‘Big Data, Media and Privacy: Do Journalism Students Feel Spied On?’ Perceptions of Data-Driven Communication, Surveillance and Professional Ethics Among Future Journalists

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The work is relevant, original, and sound from both a conceptual and methodological standpoint. It addresses a topic of clear academic and social relevance, starting with a well-defined object of study, an explicit research purpose, and research questions consistent with the problem posed.

Methodologically, the manuscript presents a recognizable design appropriate to its objectives. The author(s) define it as a cross-sectional, primarily quantitative study based on a questionnaire (N=222), complemented by a focus group that adds depth. The description of the sample, the selection procedure, the thematic blocks of the instrument, and the analysis techniques employed ensure that the methodological design is properly justified to support the results obtained.
Furthermore, the manuscript offers clear, consistent results that are well-articulated with the discussion and conclusions. The findings are significant, and the internal consistency is adequate. It is also positive that it includes information on informed consent, ethical review, and data availability, aspects that contribute to the transparency of the work. Therefore, I consider the article publishable as is, without the need for substantive modifications prior to its acceptance.

Author Response

Please also see the attachment

REVIEWER 1

 

1. Review report comment

The work is relevant, original, and sound from both a conceptual and methodological standpoint. It addresses a topic of clear academic and social relevance, starting with a well-defined object of study, an explicit research purpose, and research questions consistent with the problem posed.

Methodologically, the manuscript presents a recognizable design appropriate to its objectives. The author(s) define it as a cross-sectional, primarily quantitative study based on a questionnaire (N=222), complemented by a focus group that adds depth. The description of the sample, the selection procedure, the thematic blocks of the instrument, and the analysis techniques employed ensure that the methodological design is properly justified to support the results obtained.
Furthermore, the manuscript offers clear, consistent results that are well-articulated with the discussion and conclusions. The findings are significant, and the internal consistency is adequate. It is also positive that it includes information on informed consent, ethical review, and data availability, aspects that contribute to the transparency of the work. Therefore, I consider the article publishable as is, without the need for substantive modifications prior to its acceptance.

 

Author's reply to the review report 1

We would like to thank you for your thoughtful and positive review. We sincerely appreciate your recognition of our methodology and our efforts to promote transparency.

In any case, we have carefully reviewed the text to incorporate improvements based on feedback from the other reviewers who participated in the peer review process during this first round. We repeat our gratitude for accepting the text and for the time you have dedicated to reviewing the manuscript.

 

 

REVIEWER 2

 

1. Review report comment

The manuscript makes use of several artificial intelligence tools throughout the research process, and while the authors transparently declare their use, the way in which these tools have been integrated raises concerns that require attention.

 

Author's reply to the comment 1

The use of AI in this study has been ethical, transparent and aligned with institutional guidelines. The tools used are officially provided by the University of Seville to promote technological innovation in research.

We understand that the role of AI in research naturally generates debate and that current scholarly discussions reveal differing viewpoints among researchers. For this reason, we are carefully reviewing all editorial and reviewer comments and considering adjustments where appropriate. However, we reiterate our commitment to full transparency.

 

2. Review report comment

On the methodological side, the use of Gemini Advanced as a qualitative coding assistant introduces questions about the reliability and interpretive validity of the thematic analysis. Qualitative coding is an inherently interpretive process that requires sustained engagement with the data, theoretical sensitivity, and iterative reflexivity, qualities that AI-assisted tools cannot replicate. The authors state that a second cycle of human review was conducted, but the depth and rigour of this verification process are not described.

 

Author's reply to the comment 2

Thank you for raising these important methodological notes. We fully agree that qualitative coding is, by nature, an interpretive process that requires theoretical sensitivity, reflexivity, and deep engagement with the data. For this reason, the use of Gemini Advanced in this study didn´t replace human interpretation; rather, it served as a primary support tool to assist in organising and detecting preliminary thematic patterns in open-ended responses. Furthermore, to clarify the human contribution and guarantee reliability and interpretive validity, qualitative analysis followed a two‑stage process conducted by the researches:

- Before the interaction with AI tools, the researchers defined basic concepts, the analytical framework, and the coding orientation. The decisions about what could constitute relevant thematic categories or how to structure the coding were entirely human-driven. At this stage, Gemini Advanced was used only to generate a first-pass exploratory grouping of semantic similarities in open-ended responses. This step is comparable to using automated techniques such as word frequency clustering or topic extraction to identify possible patterns, always used as a proposal and not as definitive coding tool. Gemini did not decide about codes, categories, or interpretations. It only reduced the preliminary task of scanning large text segments for recurring ideas.

- The process was completed by human verification, recoding, and conceptual refinement. Therefore, we conducted a full second sequence of coding, which involved: reviewing all open-ended responses line by line; comparing AI-suggested gatherings with the raw data; modifying, confirming, merging, or removing AI-generated groupings; creating new codes in case it was necessary, based exclusively on human interpretation; and guaranteeing theoretical coherence with the concepts used in the quantitative sections of the study. This second cycle followed the principles of reflexive thematic analysis, including iterative engagement with the data, constant comparison, attention to context, and theoretical adequacy.

- Only after this human-driven recoding were the final themes defined, conceptualized and integrated with the quantitative findings. All interpretive decisions, the naming of the themes, and the theoretical connections, were totally human-made.

Following the recommendation, we have incorporated into the manuscript an explicit paragraph that clarifies the process through which AI-assisted tools were used during the qualitative analysis However, we believe that providing such level of detail like this one in the manuscript could unconsciously produce the opposite effect: rather than enhancing transparency regarding the use of AI and reinforcing the ethical validity of its integration, may overwhelm the reader and obscure the core methodological rationale. For this reason, and following best practices for clarity and proportionality in methodological reporting, we have opted to include two more concise paragraphs in the manuscript (lines 391-409 and 412-421) explaining the auxiliary nature of the use of Gemini Advanced and the essential role of human-led analytical task.

 

3. Review report comment

More critically, the use of NotebookLM to generate Figure 1 is problematic: a figure produced by an AI tool from survey data, without a clear description of the underlying statistical procedure or visual encoding logic, cannot be considered a scholarly data visualisation. Its inclusion without adequate methodological justification weakens the empirical credibility of the article. Authors are strongly encouraged to either replace this figure with one produced through transparent, replicable statistical procedures, or provide a detailed account of exactly how NotebookLM processed the data and what analytical decisions shaped the output.

 

Author's reply to the comment 3

We understand that, although the use of AI is openly disclosed in both the main text and in the final declaration, it is more appropriate to include graphs generated directly from the Excel file; therefore, we have proceeded with their replacement. We appreciate your concern about the visualisation of the data.

In response, we have revised this aspect of the manuscript. The original Figure 1 (line 457) has been removed and replaced with six new figures (lines 475, 495, 518, 529 and 537), each directly derived from the survey data, using transparent and replicable statistical procedures. These visualisations have been designed in accordance with standard academic practices, with clearly defined visual encoding and explicit links to the underlying data.

To further enhance clarity and interpretability, each figure has been positioned within the text along with the corresponding section in which the results are discussed. This restructuring ensures a more precise alignment between the data, the analysis, and the narrative. It also improves both readability and coherence.

We also ensured that all the data presented are traceable and reproducible, allowing readers to fully understand and replicate the analytical workflow.

We believe that these revisions directly address your concerns and significantly strengthen the robustness and transparency of the research. No AI-generated figures are included in the revised manuscript.

Thank you again for helping us improve the quality of this work.

 

 

4. Review report comment

Beyond the methodological concerns, the manuscript also shows signs that AI assistance in the writing process was not followed by sufficiently thorough human revision. The most evident symptom is the presence of a Spanish-language note.
embedded in the reference list — "(No DOI disponible; si quieres, puedo buscarlo.)" —   which is clearly an unedited prompt or response generated during AI-assisted reference management and should never appear in a submitted manuscript.

AI artifact in references (line 735): "(No DOI disponible; si quieres, puedo buscarlo.)" — a Spanish-language note left unremoved, almost certainly generated by an AI assistant. This is unacceptable in a submitted manuscript.

 

Author's reply to the comment 4

In the process of refining the writing, we used Writefull, a tool provided by the University of Seville and integrated into the Microsoft Word text processor.

I sincerely apologise for the unintended inclusion of the Spanish note. This text was generated during an internal check using an AI tool (Citewise) to help locate missing or incomplete DOIs in the reference list. It was used only for DOI lookup assistance and not for generating scientific content. In the revised text, we have carefully reviewed the entire reference list, corrected all entries with missing or truncated DOIs where available, and removed internal annotations. Also, we have mentioned the use of Citewise in the declaration of generative AI and AI-assisted technologies.

Despite conducting searches in Citewise, Google Scholar and in the publication itself, the specific reference Tahat, Khalid M., Muhammed Habes, Dana Tahat, and Mohammed Alghizzawi. 2024 did not yield any results to locate the DOI, and we concluded that no such DOI is available.

 

5. Review report comment

Additionally, the shift from third-person to second-person pronouns in the justification of the study (lines 309–313), the verbatim repetition of a theoretical quote across two
consecutive sections, and the inconsistent spelling of neologisms such as "datafied" suggest that the text was not carefully reviewed after AI-assisted drafting. The authors declare the use of Writefull for academic English improvement, yet several of these issues fall precisely within the scope of what such a tool should have flagged. Authors are advised to conduct a thorough and critical human review of the entire manuscript before resubmission, ensuring that AI-generated content has been fully absorbed, verified, and revised to meet the standards of scholarly communication.

 

Author's reply to the comment 5

We sincerely thank you for your detailed review of the text. Indeed, Writefull is a tool that suggests linguistic improvements to enhance texts in an academic style, but it has significant shortcomings in detecting deeper textual inconsistencies which, in any case, should have been spotted upon careful reading. We have therefore carried out a linguistic review to adjust the style and have made various changes, some of which are indicated as recommendations.

--In relation to the change from third-person to second-person pronouns in the justification of the study (lines 309–313 in the original manuscript), we have retained the use of the third-person verb form (lines 312-318). Furthermore, we have reviewed the entire text to prevent this error from recurring.

--Regarding the verbatim repetition of a theoretical quote across two consecutive sections, the following paragraph was, in fact, repeated verbatim and has been removed from the section ‘Digital surveillance and data capitalism: The sensation of constant monitoring”.

--Concerning the inconsistent spelling of neologisms such as “datafied”, we have standardised the spelling of this term, which appeared as “data fied” and “datafied”, opting for the standard form “datafied” (lines 162, 309, etc.) However, in the text abstract, it appears with a hyphen to indicate a word break at the end of the line (line 9).

--We have also made some changes to the citation after detecting errors in punctuation. for example (Moravec et al; 2025). These are highlighted in red. --We have also added some transitional phrases between paragraphs to avoid abrupt breaks. These can be seen in the following lines:

• Therefore, studying their perceptions… (line 312)
• Secondly, because their level… (line 314)
• Finally, because their future responsibility… (lines 316-317)
• On this basis, this need to promote research… (line 327)
• Following this procedure, the questionnaire design, structured into… (line 441)
• Given this intensive use of social networks, most respondents… (line 467)
• Moreover, concerning to professional ethics and the use of data in the journalism sector, the results… (540-541)
• Complementing the quantitative findings, the group can confirm… (574)
• b) Notably, the participation of the focus group also… (600
• In summary, the results of the research provide insight into … (623)
• In this regard, students are highly critical.. (634)
• To begin with, the tendency… (644)
• However, paradoxically, as future… (649)
• Ultimately, this hope for algorithmic… (688)

 

6. Review report comment

- Abstract formatting: The structured abstract labels (1), (2), (3) are inconsistently integrated mid-sentence, producing awkward phrasing: "Results.
  Findings (3) reveal...".

 

Author's reply to the comment 6

Thank you for your comment. The abstract has been revised to include the labels (1), (2), (3)… in accordance with MDPI style. This is a crucial change, as the abstract serves as the introduction to readers and must be absolutely clear. The changes are highlighted in red and affects to the lines 6-24).

 

7. Review report comment

Last sentence of abstract: "The study suggests that academic training on
transparency, consent, and accountability in data-driven practices is reinforced" —
grammatically incomplete or ambiguous; likely should read "should be reinforced".

 

Author's reply to the comment 7

We thank you for your comments regarding the abstracts, which are extremely valuable. We fully agree that, in its previous form, the wording (“is reinforced”) could be interpreted as incomplete or ambiguous. In any case, in response to the suggestion regarding labels (1), (2), (3) and (4), expressed in point 6, the abstract has been restructured and, consequently, the wording of some sections of the abstract, such as the results and conclusions, has been amended. These sections have been re-worked to make the text clearer. In the revised version, these sections are formulated as follows:

“(3) Results: The findings reveal widespread distrust toward social networks and political actors and a more moderate scepticism toward the news media. Students express strong ethical concerns about data use and algorithmic personalization, particularly in political communication and in relation to their future professional roles. (4) Conclusions: The study suggests that journalism students show critical awareness about algorithmic personalization. Their perceptions highlight the need for academic training in transparency, consent, and accountability.

 

8. Review report comment

-Pronoun Inconsistency (Significant)

- Lines 309–313 shift unexpectedly from third person to second person: "Studying your   perceptions is relevant... because your attitudes... because your level of media..."   — this should consistently use "their".

 

Author's reply to the comment 8

These two points have been mentioned in previous sections (point 5) and and we recognise that they offer important suggestions for improving the style. They have been duly incorporated, as described in the preceding sections.

 

9. Review report comment

Similarly line 331: "How does this perception influence your view..." in a research question that should use third-person reference.

 

Author's reply to the comment 9

Thank you for pointing out this. We have revised the research question to avoid using the second person, in relation with point 5 and 8.

In response to the third reviewer’s request to rethink the research questions in order to narrow down the two central themes, the research questions have been reformulated and now are now presented as follows:

RQ1. How do journalism students perceive algorithmic surveillance and data-driven political communication in their role as users of digital platforms?

RQ2. To what extent do journalism students express a tension between their personal concerns about surveillance and their acceptance of data-driven practices as future professionals?            

RQ3. How does this ambivalence relate to their levels of trust in media, platforms, and political actors?

 RQ4. How does this tension shape their understanding of professional ethics and the acceptable boundaries of data use in journalism and political communication?

 

 

10. Review report comment

Repetition  - The Couldry & Mejias "data colonialism" quote appears verbatim on both pages 4 and  5, suggesting poor editorial control rather than deliberate rhetorical choice.

 

Author's reply to the comment 10

Indeed, this is not an intentional repetition; therefore, in accordance with your previous instructions (point 5), it was removed from paragraph 4, as it made sense in line 185 (Digital surveillance and data capitalism: The sensation of constant monitoring). There are two further quotations from this author that have been retained in the text because they address different aspects of his theory.

 

11. Review report comment

 Minor Issues:

- Occasional awkward phrasing: "data-field environments" (line 9) — likely should be "data-filled", "data-fied environments" (line 152) and "datafied" (line 307) — inconsistent spelling of the same neologism.
  - "enslaved by algorithms" appears with different quotation marks in different
  instances (lines 472 and 588 in the original manuscript)
  - Semicolons used inconsistently in reference citations (e.g., "Moravec et al; 2025"  instead of "Moravec et al., 2025") (line 156)

 

Author's reply to the comment 11

We are extremely grateful for your proofreading work and your detailed suggestions. Most of these points, specifically, were highlighted in previous comments (point 5) and have been corrected, as previously indicated. Following a second grammar and style check, the quotation marks in the citation have also been corrected (Moravec et al., 2025).

Furthermore, in the phrase “as well as restraint, to prevent journalism from becoming enslaved by algorithms”, we have chosen to omit the quotation marks as it is not a direct quotation (line 677).

 

 

 

REVIEWER 3

 

 

1. Review report comment

 

The manuscript examines journalism students’ perceptions of Big Data, algorithmic surveillance, and data-driven political communication, with a focus on how these relate to trust and ethical orientations. The topic is timely and relevant, particularly in light of ongoing debates about datafication and the role of platforms in shaping journalistic practice. The focus on journalism students as both media users and future professionals is a useful angle, and the paper is based primarily on a reasonably sized survey dataset.

That said, in its current form the manuscript remains analytically underdeveloped and lacks a clearly articulated central contribution. While the paper raises important issues, there is still considerable work to be done to clarify the central argument and strengthen the analytical use of the empirical material before it can be considered for publication.

At present, it is not entirely clear what the paper is contributing. It is framed as addressing gaps in research on perceptions of Big Data and surveillance; however, this is an already well-established area, and the paper does not make clear what new insight it offers beyond this existing work.

To me, the more distinctive aspect of the paper lies in the tension between students’ experience of surveillance as users and their acceptance of data practices as future professionals.

This is a promising line of analysis, particularly for understanding how journalistic identities are formed within datafied environments. However, it is not developed as a central argument and appears most clearly only in the later sections of the paper. Bringing this tension to the centre of the analysis, and using it to organise both the empirical material and the theoretical framing, would significantly strengthen the paper’s contribution.

 

Author's reply to the comment 1

Thank you very much for your careful reading of the manuscript and for your insightful and constructive suggestions. We greatly appreciate your comments. In response to your observations, we have undertaken a thorough revision of the manuscript in order to clarify the central argument and strengthen its analytical development. In particular, we have made explicit from the outset the main contribution of the study, which lies in identifying and analysing the tension inherent in journalism students’ positions within datafied environments. As you pointed out, this tension—between their experience of algorithmic surveillance as users, often associated with concern and distrust, and their acceptance of data-driven practices as future professionals—constitutes a key analytical finding. This aspect is now clearly articulated in the second paragraph of the introduction (lines 340-344) and revisited in the discussion to emphasise its relevance and implications (677-683). Furthermore, the manuscript has been reorganised to ensure that this ambivalence functions as the central thread guiding the analysis. The discussion of the empirical results has been strengthened accordingly, moving from descriptive accounts towards a more developed interpretation of how this tension shapes attitudes towards trust, data use, and professional ethics. In this sense, the study now demonstrates more clearly its contribution to current debates on datafication, journalism, and democratic trust.  We have also made explicit that the analysis is primarily grounded in the concept of surveillance capitalism, which serves as the main analytical lens through which students’ perceptions are interpreted. This has enabled us to better connect the empirical findings with broader dynamics related to data extraction, platform power, and the formation of professional identities. Additionally, while Reviewer 2 highlighted the value of the research questions, we have refined their wording so that they more directly reflect the central argument and more effectively capture the tension identified in the study. These revised research questions now guide the analysis in a more focused manner. We believe that these revisions address your concerns by clarifying the manuscript’s central contribution, strengthening its analytical depth, and improving the coherence between the theoretical framework, empirical analysis, research questions, results, and conclusions. Thank you again for your valuable comments.

 

2. Review report comment

 

The paper engages with a wide range of theoretical frameworks, including surveillance capitalism, data colonialism, and trust, but these are not sufficiently integrated into the analysis. The literature review is extensive, yet much of this conceptual material is not consistently mobilised in the interpretation of the findings. For example, while concepts such as surveillance capitalism and data colonialism are used to frame the broader dynamics of data extraction and power, they are not clearly used in analysing how students understand or describe their experiences of surveillance. As a result, there is a noticeable gap between the paper’s theoretical framing and its empirical analysis. The authors would benefit from clarifying which conceptual framework is doing the primary analytical work and carrying that framework more explicitly through the results and discussion.

 

Author's reply to the comment 2

 

Thank you for your comment. We agree that the manuscript could be improved by incorporating an analysis of key concepts (surveillance capitalism, data colonialism, and trust). In response to your suggestion, we have made several changes to ensure that these concepts, which are indeed key, are reflected in the interpretation of the results.

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted (lines 340-345; 498-506; 513-515; 579-583; 630-633).

 

3. Review report comment

The methodological design is appropriate in principle, and the survey provides a solid base for exploratory analysis. However, aspects of the methodological justification require further attention. In particular, the qualitative component raises some concerns. The study includes a single focus group, but the rationale for this is not clearly explained. In its current form, it is unclear what role this component is intended to play within the overall research design. If the focus group is meant to support a mixed-methods approach or provide triangulation, one group is not sufficient to sustain such claims.

 

Author's reply to the comment 3

In the section where we describe the methodology, we have added an explanatory paragraph setting out the rationale for choosing a focus group. In this paragraph, we explain that the aim is not methodological triangulation, but rather to explore the results of the open-ended responses in greater depth and to identify trends, opinions, and attitudes, as is typical of an exploratory study (lines 391-420).

 

4. Review report comment

 

Alternatively, if it is intended as an exploratory or illustrative element, this should be explicitly stated and reflected in how the findings are presented. As it stands, the qualitative component appears somewhat underdeveloped and not fully integrated into the analytical framework, which limits its contribution to the study.

 

Author's reply to the comment 4

The core of the analysis is based on the role of quantitative analysis. Upon revisiting our manuscript, using the qualitative analysis as complementary to reinforce the quantitative results obtained but not to generate independents findings. This aspect has been corrected and specified in the text (page 10). Specifically, the semantic components identified in the focus group are used to:

• Corroborate the internal consistency of the quantitative data.
• Provide validity to the observed statistical patterns.

The lesser degree of development of the qualitative analysis is intentional and consistent with its supportive role, as now made explicit in the manuscript. Additionally, we have also added a clarification in the results obtained (page 16) that delimits the scope of the qualitative component and points out semantic evidence examples of this reinforcement for a better understanding.

 

5. Review report comment

The analysis itself is clearly presented but remains largely descriptive. Much of the results section focuses on reporting what participants think or feel, with limited interpretive development. For instance, students’ perceptions that their mobile phones are “listening” to them are reported as evidence of a broader sense of surveillance, but this is not further analysed in relation to algorithmic literacy or the opacity of data-driven systems, nor does the paper consider why such beliefs persist. Given the paper’s conceptual framing, there is scope for a more sustained analytical engagement with what these perceptions mean, how they relate to broader processes of datafication, and how they shape emerging professional identities. Bringing this level of interpretation more explicitly into the analysis would significantly strengthen the manuscript.

 

Author's reply to the comment 5

This very point was highlighted by reviewer number 2. To address this shortcoming, the changes made aim to link the findings to key concepts within the theoretical framework, interpreting the findings primarily through the lens of surveillance capitalism and, to a lesser extent, through other concepts such as data colonialism and trust. We set out below the changes made:

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted: (line 340-344; 498-506; 513-515; 578-583; 630-633).

 

6. Review report comment

There is also a tendency to move from a relatively limited and context-specific sample to broader claims about democracy, trust, and the future of journalism. For example, the discussion links students’ perceptions of data use to wider questions of democratic legitimacy, and the conclusion draws implications for the health of democratic systems and the role of universities in sustaining them. However, these connections are not clearly developed in the analysis and, at times, read more like a conceptual leap than an argument grounded in the data. The paper would benefit from more careful alignment between its empirical scope and the claims it advances, and from a clearer acknowledgement of its limitations.

 

Author's reply to the comment 6

In fact, this clarification is important because it would reinforce the integrity with regard to the objectives set, without overstepping them. To address this shortcoming, we have included some explanatory paragraphs at the points where democratic legitimacy is referred to. In the discussion section, for example, we have clarified the meaning of the interpretation provided by highlighting the limitations and restricting the interpretation to the responses of the respondents (lines 678-682; 722.727).

 

7. Review report comment

Finally, the overall structure could better support the development of the argument. There is some repetition across sections, particularly where key findings are restated without further analytical development. For example, the tension between students’ discomfort as users and their acceptance of data practices as future professionals appears in the results, discussion, and conclusion, but is largely reiterated rather than conceptually deepened. A similar pattern is evident in the treatment of ethical concerns, where calls for transparency and consent recur across sections without being further theorised. Reducing this repetition and more clearly building analytical progression across sections would strengthen the paper’s overall contribution.

 

Author's reply to the comment 7

We sincerely thank the reviewer for their constructive comments and suggestions on this work. Thank you.

As noted above, the overall structure has been drafted to highlight the central thread of the research and the development of the argument. However, we wish to make it clear that the approach does not reaffirm the findings, but rather builds on them to develop the conceptual incompatibilities of the user–professional tension identified in the results. The approach helps to explain why future journalists both resist and normalise algorithmic logics, and highlights the need for pedagogical interventions that explicitly address this duality.

In this way, the paper’s contribution lies not merely in describing students’ perceptions, but above all in revealing the tension that shapes their emerging ethical orientations and professional identities. That said, we have intended that, while the empirical sections describe how students articulate this duality, the discussion expands on its meaning for the formation of journalistic identity in datafied environments. Finally, the conclusion does not revisit the empirical results but synthesises the analytical insights derived from them, clarifying how this apparent contradiction between personal discomfort and professional acceptance is important when considering the emerging ethical orientations of future journalists."

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript makes use of several artificial intelligence tools throughout the       research process, and while the authors transparently declare their use, the way in    which these tools have been integrated raises concerns that require attention. On the   methodological side, the use of Gemini Advanced as a qualitative coding assistant     introduces questions about the reliability and interpretive validity of the thematic   analysis. Qualitative coding is an inherently interpretive process that requires       sustained engagement with the data, theoretical sensitivity, and iterative
  reflexivity — qualities that AI-assisted tools cannot replicate. The authors state
  that a second cycle of human review was conducted, but the depth and rigour of this
  verification process is not described. More critically, the use of NotebookLM to
  generate Figure 1 is problematic: a figure produced by an AI tool from survey data,
  without a clear description of the underlying statistical procedure or visual
  encoding logic, cannot be considered a scholarly data visualisation. Its inclusion
  without adequate methodological justification weakens the empirical credibility of
  the article. Authors are strongly encouraged to either replace this figure with one
  produced through transparent, replicable statistical procedures, or provide a
  detailed account of exactly how NotebookLM processed the data and what analytical
  decisions shaped the output.

  Beyond the methodological concerns, the manuscript also shows signs that AI
  assistance in the writing process was not followed by sufficiently thorough human
  revision. The most evident symptom is the presence of a Spanish-language note
  embedded in the reference list — "(No DOI disponible; si quieres, puedo buscarlo.)" —   which is clearly an unedited prompt or response generated during AI-assisted
  reference management and should never appear in a submitted manuscript. Additionally,   the shift from third-person to second-person pronouns in the justification of the
  study (lines 309–313), the verbatim repetition of a theoretical quote across two
  consecutive sections, and the inconsistent spelling of neologisms such as "datafied"
  suggest that the text was not carefully reviewed after AI-assisted drafting. The
  authors declare the use of Writefull for academic English improvement, yet several of   these issues fall precisely within the scope of what such a tool should have
  flagged. Authors are advised to conduct a thorough and critical human review of the
  entire manuscript before resubmission, ensuring that AI-generated content has been
  fully absorbed, verified, and revised to meet the standards of scholarly
  communication.

 

 - AI artifact in references (line 735): "(No DOI disponible; si quieres, puedo
  buscarlo.)" — a Spanish-language note left unremoved, almost certainly generated by
  an AI assistant. This is unacceptable in a submitted manuscript
  - Abstract formatting: The structured abstract labels (1), (2), (3) are
  inconsistently integrated mid-sentence, producing awkward phrasing: "Results.
  Findings (3) reveal..."
  - Last sentence of abstract: "The study suggests that academic training on
  transparency, consent, and accountability in data-driven practices is reinforced" —
  grammatically incomplete or ambiguous; likely should read "should be reinforced"

  Pronoun Inconsistency (Significant)

  - Lines 309–313 shift unexpectedly from third person to second person: "Studying your   perceptions is relevant... because your attitudes... because your level of media..."   — this should consistently use "their"
  - Similarly line 331: "How does this perception influence your view..." in a research   question that should use third-person reference

  Repetition

  - The Couldry & Mejias "data colonialism" quote appears verbatim on both pages 4 and
  5, suggesting poor editorial control rather than deliberate rhetorical choice

  Minor Issues

  - Occasional awkward phrasing: "data-field environments" (line 9) — likely should be
  "data-filled"
  - "data-fied environments" (line 152) and "datafied" (line 307) — inconsistent
  spelling of the same neologism
  - "enslaved by algorithms" appears with different quotation marks in different
  instances (lines 472 and 588)
  - Semicolons used inconsistently in reference citations (e.g., "Moravec et al; 2025"
  instead of "Moravec et al., 2025")

 

Comments on the Quality of English Language

The English is functional but shows clear signs of AI-assisted writing that has not
  been sufficiently reviewed by the authors. The second-person pronoun errors, the
  Spanish AI artifact, the abstract formatting issues, and the verbatim repetition all
  point to a manuscript that was drafted with AI assistance and edited inadequately.
  The declaration acknowledges use of Writefull for English improvement, but the
  remaining issues suggest this was insufficient. Revision by a proficient English
  editor is strongly recommended before publication

Author Response

Please also see the attachment

 

 

REVIEWER 1

 

1. Review report comment

The work is relevant, original, and sound from both a conceptual and methodological standpoint. It addresses a topic of clear academic and social relevance, starting with a well-defined object of study, an explicit research purpose, and research questions consistent with the problem posed.

Methodologically, the manuscript presents a recognizable design appropriate to its objectives. The author(s) define it as a cross-sectional, primarily quantitative study based on a questionnaire (N=222), complemented by a focus group that adds depth. The description of the sample, the selection procedure, the thematic blocks of the instrument, and the analysis techniques employed ensure that the methodological design is properly justified to support the results obtained.
Furthermore, the manuscript offers clear, consistent results that are well-articulated with the discussion and conclusions. The findings are significant, and the internal consistency is adequate. It is also positive that it includes information on informed consent, ethical review, and data availability, aspects that contribute to the transparency of the work. Therefore, I consider the article publishable as is, without the need for substantive modifications prior to its acceptance.

 

Author's reply to the review report 1

We would like to thank you for your thoughtful and positive review. We sincerely appreciate your recognition of our methodology and our efforts to promote transparency.

In any case, we have carefully reviewed the text to incorporate improvements based on feedback from the other reviewers who participated in the peer review process during this first round. We repeat our gratitude for accepting the text and for the time you have dedicated to reviewing the manuscript.

 

 

REVIEWER 2

 

1. Review report comment

The manuscript makes use of several artificial intelligence tools throughout the research process, and while the authors transparently declare their use, the way in which these tools have been integrated raises concerns that require attention.

 

Author's reply to the comment 1

The use of AI in this study has been ethical, transparent and aligned with institutional guidelines. The tools used are officially provided by the University of Seville to promote technological innovation in research.

We understand that the role of AI in research naturally generates debate and that current scholarly discussions reveal differing viewpoints among researchers. For this reason, we are carefully reviewing all editorial and reviewer comments and considering adjustments where appropriate. However, we reiterate our commitment to full transparency.

 

2. Review report comment

On the methodological side, the use of Gemini Advanced as a qualitative coding assistant introduces questions about the reliability and interpretive validity of the thematic analysis. Qualitative coding is an inherently interpretive process that requires sustained engagement with the data, theoretical sensitivity, and iterative reflexivity, qualities that AI-assisted tools cannot replicate. The authors state that a second cycle of human review was conducted, but the depth and rigour of this verification process are not described.

 

Author's reply to the comment 2

Thank you for raising these important methodological notes. We fully agree that qualitative coding is, by nature, an interpretive process that requires theoretical sensitivity, reflexivity, and deep engagement with the data. For this reason, the use of Gemini Advanced in this study didn´t replace human interpretation; rather, it served as a primary support tool to assist in organising and detecting preliminary thematic patterns in open-ended responses. Furthermore, to clarify the human contribution and guarantee reliability and interpretive validity, qualitative analysis followed a two‑stage process conducted by the researches:

- Before the interaction with AI tools, the researchers defined basic concepts, the analytical framework, and the coding orientation. The decisions about what could constitute relevant thematic categories or how to structure the coding were entirely human-driven. At this stage, Gemini Advanced was used only to generate a first-pass exploratory grouping of semantic similarities in open-ended responses. This step is comparable to using automated techniques such as word frequency clustering or topic extraction to identify possible patterns, always used as a proposal and not as definitive coding tool. Gemini did not decide about codes, categories, or interpretations. It only reduced the preliminary task of scanning large text segments for recurring ideas.

- The process was completed by human verification, recoding, and conceptual refinement. Therefore, we conducted a full second sequence of coding, which involved: reviewing all open-ended responses line by line; comparing AI-suggested gatherings with the raw data; modifying, confirming, merging, or removing AI-generated groupings; creating new codes in case it was necessary, based exclusively on human interpretation; and guaranteeing theoretical coherence with the concepts used in the quantitative sections of the study. This second cycle followed the principles of reflexive thematic analysis, including iterative engagement with the data, constant comparison, attention to context, and theoretical adequacy.

- Only after this human-driven recoding were the final themes defined, conceptualized and integrated with the quantitative findings. All interpretive decisions, the naming of the themes, and the theoretical connections, were totally human-made.

Following the recommendation, we have incorporated into the manuscript an explicit paragraph that clarifies the process through which AI-assisted tools were used during the qualitative analysis However, we believe that providing such level of detail like this one in the manuscript could unconsciously produce the opposite effect: rather than enhancing transparency regarding the use of AI and reinforcing the ethical validity of its integration, may overwhelm the reader and obscure the core methodological rationale. For this reason, and following best practices for clarity and proportionality in methodological reporting, we have opted to include two more concise paragraphs in the manuscript (lines 391-409 and 412-421) explaining the auxiliary nature of the use of Gemini Advanced and the essential role of human-led analytical task.

 

3. Review report comment

More critically, the use of NotebookLM to generate Figure 1 is problematic: a figure produced by an AI tool from survey data, without a clear description of the underlying statistical procedure or visual encoding logic, cannot be considered a scholarly data visualisation. Its inclusion without adequate methodological justification weakens the empirical credibility of the article. Authors are strongly encouraged to either replace this figure with one produced through transparent, replicable statistical procedures, or provide a detailed account of exactly how NotebookLM processed the data and what analytical decisions shaped the output.

 

Author's reply to the comment 3

We understand that, although the use of AI is openly disclosed in both the main text and in the final declaration, it is more appropriate to include graphs generated directly from the Excel file; therefore, we have proceeded with their replacement. We appreciate your concern about the visualisation of the data.

In response, we have revised this aspect of the manuscript. The original Figure 1 (line 457) has been removed and replaced with six new figures (lines 475, 495, 518, 529 and 537), each directly derived from the survey data, using transparent and replicable statistical procedures. These visualisations have been designed in accordance with standard academic practices, with clearly defined visual encoding and explicit links to the underlying data.

To further enhance clarity and interpretability, each figure has been positioned within the text along with the corresponding section in which the results are discussed. This restructuring ensures a more precise alignment between the data, the analysis, and the narrative. It also improves both readability and coherence.

We also ensured that all the data presented are traceable and reproducible, allowing readers to fully understand and replicate the analytical workflow.

We believe that these revisions directly address your concerns and significantly strengthen the robustness and transparency of the research. No AI-generated figures are included in the revised manuscript.

Thank you again for helping us improve the quality of this work.

 

 

4. Review report comment

Beyond the methodological concerns, the manuscript also shows signs that AI assistance in the writing process was not followed by sufficiently thorough human revision. The most evident symptom is the presence of a Spanish-language note.
embedded in the reference list — "(No DOI disponible; si quieres, puedo buscarlo.)" —   which is clearly an unedited prompt or response generated during AI-assisted reference management and should never appear in a submitted manuscript.

AI artifact in references (line 735): "(No DOI disponible; si quieres, puedo buscarlo.)" — a Spanish-language note left unremoved, almost certainly generated by an AI assistant. This is unacceptable in a submitted manuscript.

 

Author's reply to the comment 4

In the process of refining the writing, we used Writefull, a tool provided by the University of Seville and integrated into the Microsoft Word text processor.

I sincerely apologise for the unintended inclusion of the Spanish note. This text was generated during an internal check using an AI tool (Citewise) to help locate missing or incomplete DOIs in the reference list. It was used only for DOI lookup assistance and not for generating scientific content. In the revised text, we have carefully reviewed the entire reference list, corrected all entries with missing or truncated DOIs where available, and removed internal annotations. Also, we have mentioned the use of Citewise in the declaration of generative AI and AI-assisted technologies.

Despite conducting searches in Citewise, Google Scholar and in the publication itself, the specific reference Tahat, Khalid M., Muhammed Habes, Dana Tahat, and Mohammed Alghizzawi. 2024 did not yield any results to locate the DOI, and we concluded that no such DOI is available.

 

5. Review report comment

Additionally, the shift from third-person to second-person pronouns in the justification of the study (lines 309–313), the verbatim repetition of a theoretical quote across two
consecutive sections, and the inconsistent spelling of neologisms such as "datafied" suggest that the text was not carefully reviewed after AI-assisted drafting. The authors declare the use of Writefull for academic English improvement, yet several of these issues fall precisely within the scope of what such a tool should have flagged. Authors are advised to conduct a thorough and critical human review of the entire manuscript before resubmission, ensuring that AI-generated content has been fully absorbed, verified, and revised to meet the standards of scholarly communication.

 

Author's reply to the comment 5

We sincerely thank you for your detailed review of the text. Indeed, Writefull is a tool that suggests linguistic improvements to enhance texts in an academic style, but it has significant shortcomings in detecting deeper textual inconsistencies which, in any case, should have been spotted upon careful reading. We have therefore carried out a linguistic review to adjust the style and have made various changes, some of which are indicated as recommendations.

--In relation to the change from third-person to second-person pronouns in the justification of the study (lines 309–313 in the original manuscript), we have retained the use of the third-person verb form (lines 312-318). Furthermore, we have reviewed the entire text to prevent this error from recurring.

--Regarding the verbatim repetition of a theoretical quote across two consecutive sections, the following paragraph was, in fact, repeated verbatim and has been removed from the section ‘Digital surveillance and data capitalism: The sensation of constant monitoring”.

--Concerning the inconsistent spelling of neologisms such as “datafied”, we have standardised the spelling of this term, which appeared as “data fied” and “datafied”, opting for the standard form “datafied” (lines 162, 309, etc.) However, in the text abstract, it appears with a hyphen to indicate a word break at the end of the line (line 9).

--We have also made some changes to the citation after detecting errors in punctuation. for example (Moravec et al; 2025). These are highlighted in red. --We have also added some transitional phrases between paragraphs to avoid abrupt breaks. These can be seen in the following lines:

• Therefore, studying their perceptions… (line 312)
• Secondly, because their level… (line 314)
• Finally, because their future responsibility… (lines 316-317)
• On this basis, this need to promote research… (line 327)
• Following this procedure, the questionnaire design, structured into… (line 441)
• Given this intensive use of social networks, most respondents… (line 467)
• Moreover, concerning to professional ethics and the use of data in the journalism sector, the results… (540-541)
• Complementing the quantitative findings, the group can confirm… (574)
• b) Notably, the participation of the focus group also… (600
• In summary, the results of the research provide insight into … (623)
• In this regard, students are highly critical.. (634)
• To begin with, the tendency… (644)
• However, paradoxically, as future… (649)
• Ultimately, this hope for algorithmic… (688)

 

6. Review report comment

- Abstract formatting: The structured abstract labels (1), (2), (3) are inconsistently integrated mid-sentence, producing awkward phrasing: "Results.
  Findings (3) reveal...".

 

Author's reply to the comment 6

Thank you for your comment. The abstract has been revised to include the labels (1), (2), (3)… in accordance with MDPI style. This is a crucial change, as the abstract serves as the introduction to readers and must be absolutely clear. The changes are highlighted in red and affects to the lines 6-24).

 

7. Review report comment

Last sentence of abstract: "The study suggests that academic training on
transparency, consent, and accountability in data-driven practices is reinforced" —
grammatically incomplete or ambiguous; likely should read "should be reinforced".

 

Author's reply to the comment 7

We thank you for your comments regarding the abstracts, which are extremely valuable. We fully agree that, in its previous form, the wording (“is reinforced”) could be interpreted as incomplete or ambiguous. In any case, in response to the suggestion regarding labels (1), (2), (3) and (4), expressed in point 6, the abstract has been restructured and, consequently, the wording of some sections of the abstract, such as the results and conclusions, has been amended. These sections have been re-worked to make the text clearer. In the revised version, these sections are formulated as follows:

“(3) Results: The findings reveal widespread distrust toward social networks and political actors and a more moderate scepticism toward the news media. Students express strong ethical concerns about data use and algorithmic personalization, particularly in political communication and in relation to their future professional roles. (4) Conclusions: The study suggests that journalism students show critical awareness about algorithmic personalization. Their perceptions highlight the need for academic training in transparency, consent, and accountability.

 

8. Review report comment

-Pronoun Inconsistency (Significant)

- Lines 309–313 shift unexpectedly from third person to second person: "Studying your   perceptions is relevant... because your attitudes... because your level of media..."   — this should consistently use "their".

 

Author's reply to the comment 8

These two points have been mentioned in previous sections (point 5) and and we recognise that they offer important suggestions for improving the style. They have been duly incorporated, as described in the preceding sections.

 

9. Review report comment

Similarly line 331: "How does this perception influence your view..." in a research question that should use third-person reference.

 

Author's reply to the comment 9

Thank you for pointing out this. We have revised the research question to avoid using the second person, in relation with point 5 and 8.

In response to the third reviewer’s request to rethink the research questions in order to narrow down the two central themes, the research questions have been reformulated and now are now presented as follows:

RQ1. How do journalism students perceive algorithmic surveillance and data-driven political communication in their role as users of digital platforms?

RQ2. To what extent do journalism students express a tension between their personal concerns about surveillance and their acceptance of data-driven practices as future professionals?            

RQ3. How does this ambivalence relate to their levels of trust in media, platforms, and political actors?

 RQ4. How does this tension shape their understanding of professional ethics and the acceptable boundaries of data use in journalism and political communication?

 

 

10. Review report comment

Repetition  - The Couldry & Mejias "data colonialism" quote appears verbatim on both pages 4 and  5, suggesting poor editorial control rather than deliberate rhetorical choice.

 

Author's reply to the comment 10

Indeed, this is not an intentional repetition; therefore, in accordance with your previous instructions (point 5), it was removed from paragraph 4, as it made sense in line 185 (Digital surveillance and data capitalism: The sensation of constant monitoring). There are two further quotations from this author that have been retained in the text because they address different aspects of his theory.

 

11. Review report comment

 Minor Issues:

- Occasional awkward phrasing: "data-field environments" (line 9) — likely should be "data-filled", "data-fied environments" (line 152) and "datafied" (line 307) — inconsistent spelling of the same neologism.
  - "enslaved by algorithms" appears with different quotation marks in different
  instances (lines 472 and 588 in the original manuscript)
  - Semicolons used inconsistently in reference citations (e.g., "Moravec et al; 2025"  instead of "Moravec et al., 2025") (line 156)

 

Author's reply to the comment 11

We are extremely grateful for your proofreading work and your detailed suggestions. Most of these points, specifically, were highlighted in previous comments (point 5) and have been corrected, as previously indicated. Following a second grammar and style check, the quotation marks in the citation have also been corrected (Moravec et al., 2025).

Furthermore, in the phrase “as well as restraint, to prevent journalism from becoming enslaved by algorithms”, we have chosen to omit the quotation marks as it is not a direct quotation (line 677).

 

 

 

REVIEWER 3

 

 

1. Review report comment

 

The manuscript examines journalism students’ perceptions of Big Data, algorithmic surveillance, and data-driven political communication, with a focus on how these relate to trust and ethical orientations. The topic is timely and relevant, particularly in light of ongoing debates about datafication and the role of platforms in shaping journalistic practice. The focus on journalism students as both media users and future professionals is a useful angle, and the paper is based primarily on a reasonably sized survey dataset.

That said, in its current form the manuscript remains analytically underdeveloped and lacks a clearly articulated central contribution. While the paper raises important issues, there is still considerable work to be done to clarify the central argument and strengthen the analytical use of the empirical material before it can be considered for publication.

At present, it is not entirely clear what the paper is contributing. It is framed as addressing gaps in research on perceptions of Big Data and surveillance; however, this is an already well-established area, and the paper does not make clear what new insight it offers beyond this existing work.

To me, the more distinctive aspect of the paper lies in the tension between students’ experience of surveillance as users and their acceptance of data practices as future professionals.

This is a promising line of analysis, particularly for understanding how journalistic identities are formed within datafied environments. However, it is not developed as a central argument and appears most clearly only in the later sections of the paper. Bringing this tension to the centre of the analysis, and using it to organise both the empirical material and the theoretical framing, would significantly strengthen the paper’s contribution.

 

Author's reply to the comment 1

Thank you very much for your careful reading of the manuscript and for your insightful and constructive suggestions. We greatly appreciate your comments. In response to your observations, we have undertaken a thorough revision of the manuscript in order to clarify the central argument and strengthen its analytical development. In particular, we have made explicit from the outset the main contribution of the study, which lies in identifying and analysing the tension inherent in journalism students’ positions within datafied environments. As you pointed out, this tension—between their experience of algorithmic surveillance as users, often associated with concern and distrust, and their acceptance of data-driven practices as future professionals—constitutes a key analytical finding. This aspect is now clearly articulated in the second paragraph of the introduction (lines 340-344) and revisited in the discussion to emphasise its relevance and implications (677-683). Furthermore, the manuscript has been reorganised to ensure that this ambivalence functions as the central thread guiding the analysis. The discussion of the empirical results has been strengthened accordingly, moving from descriptive accounts towards a more developed interpretation of how this tension shapes attitudes towards trust, data use, and professional ethics. In this sense, the study now demonstrates more clearly its contribution to current debates on datafication, journalism, and democratic trust.  We have also made explicit that the analysis is primarily grounded in the concept of surveillance capitalism, which serves as the main analytical lens through which students’ perceptions are interpreted. This has enabled us to better connect the empirical findings with broader dynamics related to data extraction, platform power, and the formation of professional identities. Additionally, while Reviewer 2 highlighted the value of the research questions, we have refined their wording so that they more directly reflect the central argument and more effectively capture the tension identified in the study. These revised research questions now guide the analysis in a more focused manner. We believe that these revisions address your concerns by clarifying the manuscript’s central contribution, strengthening its analytical depth, and improving the coherence between the theoretical framework, empirical analysis, research questions, results, and conclusions. Thank you again for your valuable comments.

 

2. Review report comment

 

The paper engages with a wide range of theoretical frameworks, including surveillance capitalism, data colonialism, and trust, but these are not sufficiently integrated into the analysis. The literature review is extensive, yet much of this conceptual material is not consistently mobilised in the interpretation of the findings. For example, while concepts such as surveillance capitalism and data colonialism are used to frame the broader dynamics of data extraction and power, they are not clearly used in analysing how students understand or describe their experiences of surveillance. As a result, there is a noticeable gap between the paper’s theoretical framing and its empirical analysis. The authors would benefit from clarifying which conceptual framework is doing the primary analytical work and carrying that framework more explicitly through the results and discussion.

 

Author's reply to the comment 2

 

Thank you for your comment. We agree that the manuscript could be improved by incorporating an analysis of key concepts (surveillance capitalism, data colonialism, and trust). In response to your suggestion, we have made several changes to ensure that these concepts, which are indeed key, are reflected in the interpretation of the results.

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted (lines 340-345; 498-506; 513-515; 579-583; 630-633).

 

3. Review report comment

The methodological design is appropriate in principle, and the survey provides a solid base for exploratory analysis. However, aspects of the methodological justification require further attention. In particular, the qualitative component raises some concerns. The study includes a single focus group, but the rationale for this is not clearly explained. In its current form, it is unclear what role this component is intended to play within the overall research design. If the focus group is meant to support a mixed-methods approach or provide triangulation, one group is not sufficient to sustain such claims.

 

Author's reply to the comment 3

In the section where we describe the methodology, we have added an explanatory paragraph setting out the rationale for choosing a focus group. In this paragraph, we explain that the aim is not methodological triangulation, but rather to explore the results of the open-ended responses in greater depth and to identify trends, opinions, and attitudes, as is typical of an exploratory study (lines 391-420).

 

4. Review report comment

 

Alternatively, if it is intended as an exploratory or illustrative element, this should be explicitly stated and reflected in how the findings are presented. As it stands, the qualitative component appears somewhat underdeveloped and not fully integrated into the analytical framework, which limits its contribution to the study.

 

Author's reply to the comment 4

The core of the analysis is based on the role of quantitative analysis. Upon revisiting our manuscript, using the qualitative analysis as complementary to reinforce the quantitative results obtained but not to generate independents findings. This aspect has been corrected and specified in the text (page 10). Specifically, the semantic components identified in the focus group are used to:

• Corroborate the internal consistency of the quantitative data.
• Provide validity to the observed statistical patterns.

The lesser degree of development of the qualitative analysis is intentional and consistent with its supportive role, as now made explicit in the manuscript. Additionally, we have also added a clarification in the results obtained (page 16) that delimits the scope of the qualitative component and points out semantic evidence examples of this reinforcement for a better understanding.

 

5. Review report comment

The analysis itself is clearly presented but remains largely descriptive. Much of the results section focuses on reporting what participants think or feel, with limited interpretive development. For instance, students’ perceptions that their mobile phones are “listening” to them are reported as evidence of a broader sense of surveillance, but this is not further analysed in relation to algorithmic literacy or the opacity of data-driven systems, nor does the paper consider why such beliefs persist. Given the paper’s conceptual framing, there is scope for a more sustained analytical engagement with what these perceptions mean, how they relate to broader processes of datafication, and how they shape emerging professional identities. Bringing this level of interpretation more explicitly into the analysis would significantly strengthen the manuscript.

 

Author's reply to the comment 5

This very point was highlighted by reviewer number 2. To address this shortcoming, the changes made aim to link the findings to key concepts within the theoretical framework, interpreting the findings primarily through the lens of surveillance capitalism and, to a lesser extent, through other concepts such as data colonialism and trust. We set out below the changes made:

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted: (line 340-344; 498-506; 513-515; 578-583; 630-633).

 

6. Review report comment

There is also a tendency to move from a relatively limited and context-specific sample to broader claims about democracy, trust, and the future of journalism. For example, the discussion links students’ perceptions of data use to wider questions of democratic legitimacy, and the conclusion draws implications for the health of democratic systems and the role of universities in sustaining them. However, these connections are not clearly developed in the analysis and, at times, read more like a conceptual leap than an argument grounded in the data. The paper would benefit from more careful alignment between its empirical scope and the claims it advances, and from a clearer acknowledgement of its limitations.

 

Author's reply to the comment 6

In fact, this clarification is important because it would reinforce the integrity with regard to the objectives set, without overstepping them. To address this shortcoming, we have included some explanatory paragraphs at the points where democratic legitimacy is referred to. In the discussion section, for example, we have clarified the meaning of the interpretation provided by highlighting the limitations and restricting the interpretation to the responses of the respondents (lines 678-682; 722.727).

 

7. Review report comment

Finally, the overall structure could better support the development of the argument. There is some repetition across sections, particularly where key findings are restated without further analytical development. For example, the tension between students’ discomfort as users and their acceptance of data practices as future professionals appears in the results, discussion, and conclusion, but is largely reiterated rather than conceptually deepened. A similar pattern is evident in the treatment of ethical concerns, where calls for transparency and consent recur across sections without being further theorised. Reducing this repetition and more clearly building analytical progression across sections would strengthen the paper’s overall contribution.

 

Author's reply to the comment 7

We sincerely thank the reviewer for their constructive comments and suggestions on this work. Thank you.

As noted above, the overall structure has been drafted to highlight the central thread of the research and the development of the argument. However, we wish to make it clear that the approach does not reaffirm the findings, but rather builds on them to develop the conceptual incompatibilities of the user–professional tension identified in the results. The approach helps to explain why future journalists both resist and normalise algorithmic logics, and highlights the need for pedagogical interventions that explicitly address this duality.

In this way, the paper’s contribution lies not merely in describing students’ perceptions, but above all in revealing the tension that shapes their emerging ethical orientations and professional identities. That said, we have intended that, while the empirical sections describe how students articulate this duality, the discussion expands on its meaning for the formation of journalistic identity in datafied environments. Finally, the conclusion does not revisit the empirical results but synthesises the analytical insights derived from them, clarifying how this apparent contradiction between personal discomfort and professional acceptance is important when considering the emerging ethical orientations of future journalists."

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript examines journalism students’ perceptions of Big Data, algorithmic surveillance, and data-driven political communication, with a focus on how these relate to trust and ethical orientations. The topic is timely and relevant, particularly in light of ongoing debates about datafication and the role of platforms in shaping journalistic practice. The focus on journalism students as both media users and future professionals is a useful angle, and the paper is based primarily on a reasonably sized survey dataset.

That said, in its current form the manuscript remains analytically underdeveloped and lacks a clearly articulated central contribution. While the paper raises important issues, there is still considerable work to be done to clarify the central argument and strengthen the analytical use of the empirical material before it can be considered for publication.

At present, it is not entirely clear what the paper is contributing. It is framed as addressing gaps in research on perceptions of Big Data and surveillance; however, this is an already well-established area, and the paper does not make clear what new insight it offers beyond this existing work. To me, the more distinctive aspect of the paper lies in the tension between students’ experience of surveillance as users and their acceptance of data practices as future professionals. This is a promising line of analysis, particularly for understanding how journalistic identities are formed within datafied environments. However, it is not developed as a central argument and appears most clearly only in the later sections of the paper. Bringing this tension to the centre of the analysis, and using it to organise both the empirical material and the theoretical framing, would significantly strengthen the paper’s contribution.

The paper engages with a wide range of theoretical frameworks, including surveillance capitalism, data colonialism, and trust, but these are not sufficiently integrated into the analysis. The literature review is extensive, yet much of this conceptual material is not consistently mobilised in the interpretation of the findings. For example, while concepts such as surveillance capitalism and data colonialism are used to frame the broader dynamics of data extraction and power, they are not clearly used in analysing how students understand or describe their experiences of surveillance. As a result, there is a noticeable gap between the paper’s theoretical framing and its empirical analysis. The authors would benefit from clarifying which conceptual framework is doing the primary analytical work and carrying that framework more explicitly through the results and discussion.

The methodological design is appropriate in principle, and the survey provides a solid base for exploratory analysis. However, aspects of the methodological justification require further attention. In particular, the qualitative component raises some concerns. The study includes a single focus group, but the rationale for this is not clearly explained. In its current form, it is unclear what role this component is intended to play within the overall research design. If the focus group is meant to support a mixed-methods approach or provide triangulation, one group is not sufficient to sustain such claims. Alternatively, if it is intended as an exploratory or illustrative element, this should be explicitly stated and reflected in how the findings are presented. As it stands, the qualitative component appears somewhat underdeveloped and not fully integrated into the analytical framework, which limits its contribution to the study.

The analysis itself is clearly presented but remains largely descriptive. Much of the results section focuses on reporting what participants think or feel, with limited interpretive development. For instance, students’ perceptions that their mobile phones are “listening” to them are reported as evidence of a broader sense of surveillance, but this is not further analysed in relation to algorithmic literacy or the opacity of data-driven systems, nor does the paper consider why such beliefs persist. Given the paper’s conceptual framing, there is scope for a more sustained analytical engagement with what these perceptions mean, how they relate to broader processes of datafication, and how they shape emerging professional identities. Bringing this level of interpretation more explicitly into the analysis would significantly strengthen the manuscript.

There is also a tendency to move from a relatively limited and context-specific sample to broader claims about democracy, trust, and the future of journalism. For example, the discussion links students’ perceptions of data use to wider questions of democratic legitimacy , and the conclusion draws implications for the health of democratic systems and the role of universities in sustaining them . However, these connections are not clearly developed in the analysis and, at times, read more like a conceptual leap than an argument grounded in the data. The paper would benefit from more careful alignment between its empirical scope and the claims it advances, and from a clearer acknowledgement of its limitations.

Finally, the overall structure could better support the development of the argument. There is some repetition across sections, particularly where key findings are restated without further analytical development. For example, the tension between students’ discomfort as users and their acceptance of data practices as future professionals appears in the results, discussion, and conclusion , but is largely reiterated rather than conceptually deepened. A similar pattern is evident in the treatment of ethical concerns, where calls for transparency and consent recur across sections without being further theorised. Reducing this repetition and more clearly building analytical progression across sections would strengthen the paper’s overall contribution.

Author Response

Please also see the attachment

REVIEWER 1

 

1. Review report comment

The work is relevant, original, and sound from both a conceptual and methodological standpoint. It addresses a topic of clear academic and social relevance, starting with a well-defined object of study, an explicit research purpose, and research questions consistent with the problem posed.

Methodologically, the manuscript presents a recognizable design appropriate to its objectives. The author(s) define it as a cross-sectional, primarily quantitative study based on a questionnaire (N=222), complemented by a focus group that adds depth. The description of the sample, the selection procedure, the thematic blocks of the instrument, and the analysis techniques employed ensure that the methodological design is properly justified to support the results obtained.
Furthermore, the manuscript offers clear, consistent results that are well-articulated with the discussion and conclusions. The findings are significant, and the internal consistency is adequate. It is also positive that it includes information on informed consent, ethical review, and data availability, aspects that contribute to the transparency of the work. Therefore, I consider the article publishable as is, without the need for substantive modifications prior to its acceptance.

 

Author's reply to the review report 1

We would like to thank you for your thoughtful and positive review. We sincerely appreciate your recognition of our methodology and our efforts to promote transparency.

In any case, we have carefully reviewed the text to incorporate improvements based on feedback from the other reviewers who participated in the peer review process during this first round. We repeat our gratitude for accepting the text and for the time you have dedicated to reviewing the manuscript.

 

 

REVIEWER 2

 

1. Review report comment

The manuscript makes use of several artificial intelligence tools throughout the research process, and while the authors transparently declare their use, the way in which these tools have been integrated raises concerns that require attention.

 

Author's reply to the comment 1

The use of AI in this study has been ethical, transparent and aligned with institutional guidelines. The tools used are officially provided by the University of Seville to promote technological innovation in research.

We understand that the role of AI in research naturally generates debate and that current scholarly discussions reveal differing viewpoints among researchers. For this reason, we are carefully reviewing all editorial and reviewer comments and considering adjustments where appropriate. However, we reiterate our commitment to full transparency.

 

2. Review report comment

On the methodological side, the use of Gemini Advanced as a qualitative coding assistant introduces questions about the reliability and interpretive validity of the thematic analysis. Qualitative coding is an inherently interpretive process that requires sustained engagement with the data, theoretical sensitivity, and iterative reflexivity, qualities that AI-assisted tools cannot replicate. The authors state that a second cycle of human review was conducted, but the depth and rigour of this verification process are not described.

 

Author's reply to the comment 2

Thank you for raising these important methodological notes. We fully agree that qualitative coding is, by nature, an interpretive process that requires theoretical sensitivity, reflexivity, and deep engagement with the data. For this reason, the use of Gemini Advanced in this study didn´t replace human interpretation; rather, it served as a primary support tool to assist in organising and detecting preliminary thematic patterns in open-ended responses. Furthermore, to clarify the human contribution and guarantee reliability and interpretive validity, qualitative analysis followed a two‑stage process conducted by the researches:

- Before the interaction with AI tools, the researchers defined basic concepts, the analytical framework, and the coding orientation. The decisions about what could constitute relevant thematic categories or how to structure the coding were entirely human-driven. At this stage, Gemini Advanced was used only to generate a first-pass exploratory grouping of semantic similarities in open-ended responses. This step is comparable to using automated techniques such as word frequency clustering or topic extraction to identify possible patterns, always used as a proposal and not as definitive coding tool. Gemini did not decide about codes, categories, or interpretations. It only reduced the preliminary task of scanning large text segments for recurring ideas.

- The process was completed by human verification, recoding, and conceptual refinement. Therefore, we conducted a full second sequence of coding, which involved: reviewing all open-ended responses line by line; comparing AI-suggested gatherings with the raw data; modifying, confirming, merging, or removing AI-generated groupings; creating new codes in case it was necessary, based exclusively on human interpretation; and guaranteeing theoretical coherence with the concepts used in the quantitative sections of the study. This second cycle followed the principles of reflexive thematic analysis, including iterative engagement with the data, constant comparison, attention to context, and theoretical adequacy.

- Only after this human-driven recoding were the final themes defined, conceptualized and integrated with the quantitative findings. All interpretive decisions, the naming of the themes, and the theoretical connections, were totally human-made.

Following the recommendation, we have incorporated into the manuscript an explicit paragraph that clarifies the process through which AI-assisted tools were used during the qualitative analysis However, we believe that providing such level of detail like this one in the manuscript could unconsciously produce the opposite effect: rather than enhancing transparency regarding the use of AI and reinforcing the ethical validity of its integration, may overwhelm the reader and obscure the core methodological rationale. For this reason, and following best practices for clarity and proportionality in methodological reporting, we have opted to include two more concise paragraphs in the manuscript (lines 391-409 and 412-421) explaining the auxiliary nature of the use of Gemini Advanced and the essential role of human-led analytical task.

 

3. Review report comment

More critically, the use of NotebookLM to generate Figure 1 is problematic: a figure produced by an AI tool from survey data, without a clear description of the underlying statistical procedure or visual encoding logic, cannot be considered a scholarly data visualisation. Its inclusion without adequate methodological justification weakens the empirical credibility of the article. Authors are strongly encouraged to either replace this figure with one produced through transparent, replicable statistical procedures, or provide a detailed account of exactly how NotebookLM processed the data and what analytical decisions shaped the output.

 

Author's reply to the comment 3

We understand that, although the use of AI is openly disclosed in both the main text and in the final declaration, it is more appropriate to include graphs generated directly from the Excel file; therefore, we have proceeded with their replacement. We appreciate your concern about the visualisation of the data.

In response, we have revised this aspect of the manuscript. The original Figure 1 (line 457) has been removed and replaced with six new figures (lines 475, 495, 518, 529 and 537), each directly derived from the survey data, using transparent and replicable statistical procedures. These visualisations have been designed in accordance with standard academic practices, with clearly defined visual encoding and explicit links to the underlying data.

To further enhance clarity and interpretability, each figure has been positioned within the text along with the corresponding section in which the results are discussed. This restructuring ensures a more precise alignment between the data, the analysis, and the narrative. It also improves both readability and coherence.

We also ensured that all the data presented are traceable and reproducible, allowing readers to fully understand and replicate the analytical workflow.

We believe that these revisions directly address your concerns and significantly strengthen the robustness and transparency of the research. No AI-generated figures are included in the revised manuscript.

Thank you again for helping us improve the quality of this work.

 

 

4. Review report comment

Beyond the methodological concerns, the manuscript also shows signs that AI assistance in the writing process was not followed by sufficiently thorough human revision. The most evident symptom is the presence of a Spanish-language note.
embedded in the reference list — "(No DOI disponible; si quieres, puedo buscarlo.)" —   which is clearly an unedited prompt or response generated during AI-assisted reference management and should never appear in a submitted manuscript.

AI artifact in references (line 735): "(No DOI disponible; si quieres, puedo buscarlo.)" — a Spanish-language note left unremoved, almost certainly generated by an AI assistant. This is unacceptable in a submitted manuscript.

 

Author's reply to the comment 4

In the process of refining the writing, we used Writefull, a tool provided by the University of Seville and integrated into the Microsoft Word text processor.

I sincerely apologise for the unintended inclusion of the Spanish note. This text was generated during an internal check using an AI tool (Citewise) to help locate missing or incomplete DOIs in the reference list. It was used only for DOI lookup assistance and not for generating scientific content. In the revised text, we have carefully reviewed the entire reference list, corrected all entries with missing or truncated DOIs where available, and removed internal annotations. Also, we have mentioned the use of Citewise in the declaration of generative AI and AI-assisted technologies.

Despite conducting searches in Citewise, Google Scholar and in the publication itself, the specific reference Tahat, Khalid M., Muhammed Habes, Dana Tahat, and Mohammed Alghizzawi. 2024 did not yield any results to locate the DOI, and we concluded that no such DOI is available.

 

5. Review report comment

Additionally, the shift from third-person to second-person pronouns in the justification of the study (lines 309–313), the verbatim repetition of a theoretical quote across two
consecutive sections, and the inconsistent spelling of neologisms such as "datafied" suggest that the text was not carefully reviewed after AI-assisted drafting. The authors declare the use of Writefull for academic English improvement, yet several of these issues fall precisely within the scope of what such a tool should have flagged. Authors are advised to conduct a thorough and critical human review of the entire manuscript before resubmission, ensuring that AI-generated content has been fully absorbed, verified, and revised to meet the standards of scholarly communication.

 

Author's reply to the comment 5

We sincerely thank you for your detailed review of the text. Indeed, Writefull is a tool that suggests linguistic improvements to enhance texts in an academic style, but it has significant shortcomings in detecting deeper textual inconsistencies which, in any case, should have been spotted upon careful reading. We have therefore carried out a linguistic review to adjust the style and have made various changes, some of which are indicated as recommendations.

--In relation to the change from third-person to second-person pronouns in the justification of the study (lines 309–313 in the original manuscript), we have retained the use of the third-person verb form (lines 312-318). Furthermore, we have reviewed the entire text to prevent this error from recurring.

--Regarding the verbatim repetition of a theoretical quote across two consecutive sections, the following paragraph was, in fact, repeated verbatim and has been removed from the section ‘Digital surveillance and data capitalism: The sensation of constant monitoring”.

--Concerning the inconsistent spelling of neologisms such as “datafied”, we have standardised the spelling of this term, which appeared as “data fied” and “datafied”, opting for the standard form “datafied” (lines 162, 309, etc.) However, in the text abstract, it appears with a hyphen to indicate a word break at the end of the line (line 9).

--We have also made some changes to the citation after detecting errors in punctuation. for example (Moravec et al; 2025). These are highlighted in red. --We have also added some transitional phrases between paragraphs to avoid abrupt breaks. These can be seen in the following lines:

• Therefore, studying their perceptions… (line 312)
• Secondly, because their level… (line 314)
• Finally, because their future responsibility… (lines 316-317)
• On this basis, this need to promote research… (line 327)
• Following this procedure, the questionnaire design, structured into… (line 441)
• Given this intensive use of social networks, most respondents… (line 467)
• Moreover, concerning to professional ethics and the use of data in the journalism sector, the results… (540-541)
• Complementing the quantitative findings, the group can confirm… (574)
• b) Notably, the participation of the focus group also… (600
• In summary, the results of the research provide insight into … (623)
• In this regard, students are highly critical.. (634)
• To begin with, the tendency… (644)
• However, paradoxically, as future… (649)
• Ultimately, this hope for algorithmic… (688)

 

6. Review report comment

- Abstract formatting: The structured abstract labels (1), (2), (3) are inconsistently integrated mid-sentence, producing awkward phrasing: "Results.
  Findings (3) reveal...".

 

Author's reply to the comment 6

Thank you for your comment. The abstract has been revised to include the labels (1), (2), (3)… in accordance with MDPI style. This is a crucial change, as the abstract serves as the introduction to readers and must be absolutely clear. The changes are highlighted in red and affects to the lines 6-24).

 

7. Review report comment

Last sentence of abstract: "The study suggests that academic training on
transparency, consent, and accountability in data-driven practices is reinforced" —
grammatically incomplete or ambiguous; likely should read "should be reinforced".

 

Author's reply to the comment 7

We thank you for your comments regarding the abstracts, which are extremely valuable. We fully agree that, in its previous form, the wording (“is reinforced”) could be interpreted as incomplete or ambiguous. In any case, in response to the suggestion regarding labels (1), (2), (3) and (4), expressed in point 6, the abstract has been restructured and, consequently, the wording of some sections of the abstract, such as the results and conclusions, has been amended. These sections have been re-worked to make the text clearer. In the revised version, these sections are formulated as follows:

“(3) Results: The findings reveal widespread distrust toward social networks and political actors and a more moderate scepticism toward the news media. Students express strong ethical concerns about data use and algorithmic personalization, particularly in political communication and in relation to their future professional roles. (4) Conclusions: The study suggests that journalism students show critical awareness about algorithmic personalization. Their perceptions highlight the need for academic training in transparency, consent, and accountability.

 

8. Review report comment

-Pronoun Inconsistency (Significant)

- Lines 309–313 shift unexpectedly from third person to second person: "Studying your   perceptions is relevant... because your attitudes... because your level of media..."   — this should consistently use "their".

 

Author's reply to the comment 8

These two points have been mentioned in previous sections (point 5) and and we recognise that they offer important suggestions for improving the style. They have been duly incorporated, as described in the preceding sections.

 

9. Review report comment

Similarly line 331: "How does this perception influence your view..." in a research question that should use third-person reference.

 

Author's reply to the comment 9

Thank you for pointing out this. We have revised the research question to avoid using the second person, in relation with point 5 and 8.

In response to the third reviewer’s request to rethink the research questions in order to narrow down the two central themes, the research questions have been reformulated and now are now presented as follows:

RQ1. How do journalism students perceive algorithmic surveillance and data-driven political communication in their role as users of digital platforms?

RQ2. To what extent do journalism students express a tension between their personal concerns about surveillance and their acceptance of data-driven practices as future professionals?            

RQ3. How does this ambivalence relate to their levels of trust in media, platforms, and political actors?

 RQ4. How does this tension shape their understanding of professional ethics and the acceptable boundaries of data use in journalism and political communication?

 

 

10. Review report comment

Repetition  - The Couldry & Mejias "data colonialism" quote appears verbatim on both pages 4 and  5, suggesting poor editorial control rather than deliberate rhetorical choice.

 

Author's reply to the comment 10

Indeed, this is not an intentional repetition; therefore, in accordance with your previous instructions (point 5), it was removed from paragraph 4, as it made sense in line 185 (Digital surveillance and data capitalism: The sensation of constant monitoring). There are two further quotations from this author that have been retained in the text because they address different aspects of his theory.

 

11. Review report comment

 Minor Issues:

- Occasional awkward phrasing: "data-field environments" (line 9) — likely should be "data-filled", "data-fied environments" (line 152) and "datafied" (line 307) — inconsistent spelling of the same neologism.
  - "enslaved by algorithms" appears with different quotation marks in different
  instances (lines 472 and 588 in the original manuscript)
  - Semicolons used inconsistently in reference citations (e.g., "Moravec et al; 2025"  instead of "Moravec et al., 2025") (line 156)

 

Author's reply to the comment 11

We are extremely grateful for your proofreading work and your detailed suggestions. Most of these points, specifically, were highlighted in previous comments (point 5) and have been corrected, as previously indicated. Following a second grammar and style check, the quotation marks in the citation have also been corrected (Moravec et al., 2025).

Furthermore, in the phrase “as well as restraint, to prevent journalism from becoming enslaved by algorithms”, we have chosen to omit the quotation marks as it is not a direct quotation (line 677).

 

 

 

REVIEWER 3

 

 

1. Review report comment

 

The manuscript examines journalism students’ perceptions of Big Data, algorithmic surveillance, and data-driven political communication, with a focus on how these relate to trust and ethical orientations. The topic is timely and relevant, particularly in light of ongoing debates about datafication and the role of platforms in shaping journalistic practice. The focus on journalism students as both media users and future professionals is a useful angle, and the paper is based primarily on a reasonably sized survey dataset.

That said, in its current form the manuscript remains analytically underdeveloped and lacks a clearly articulated central contribution. While the paper raises important issues, there is still considerable work to be done to clarify the central argument and strengthen the analytical use of the empirical material before it can be considered for publication.

At present, it is not entirely clear what the paper is contributing. It is framed as addressing gaps in research on perceptions of Big Data and surveillance; however, this is an already well-established area, and the paper does not make clear what new insight it offers beyond this existing work.

To me, the more distinctive aspect of the paper lies in the tension between students’ experience of surveillance as users and their acceptance of data practices as future professionals.

This is a promising line of analysis, particularly for understanding how journalistic identities are formed within datafied environments. However, it is not developed as a central argument and appears most clearly only in the later sections of the paper. Bringing this tension to the centre of the analysis, and using it to organise both the empirical material and the theoretical framing, would significantly strengthen the paper’s contribution.

 

Author's reply to the comment 1

Thank you very much for your careful reading of the manuscript and for your insightful and constructive suggestions. We greatly appreciate your comments. In response to your observations, we have undertaken a thorough revision of the manuscript in order to clarify the central argument and strengthen its analytical development. In particular, we have made explicit from the outset the main contribution of the study, which lies in identifying and analysing the tension inherent in journalism students’ positions within datafied environments. As you pointed out, this tension—between their experience of algorithmic surveillance as users, often associated with concern and distrust, and their acceptance of data-driven practices as future professionals—constitutes a key analytical finding. This aspect is now clearly articulated in the second paragraph of the introduction (lines 340-344) and revisited in the discussion to emphasise its relevance and implications (677-683). Furthermore, the manuscript has been reorganised to ensure that this ambivalence functions as the central thread guiding the analysis. The discussion of the empirical results has been strengthened accordingly, moving from descriptive accounts towards a more developed interpretation of how this tension shapes attitudes towards trust, data use, and professional ethics. In this sense, the study now demonstrates more clearly its contribution to current debates on datafication, journalism, and democratic trust.  We have also made explicit that the analysis is primarily grounded in the concept of surveillance capitalism, which serves as the main analytical lens through which students’ perceptions are interpreted. This has enabled us to better connect the empirical findings with broader dynamics related to data extraction, platform power, and the formation of professional identities. Additionally, while Reviewer 2 highlighted the value of the research questions, we have refined their wording so that they more directly reflect the central argument and more effectively capture the tension identified in the study. These revised research questions now guide the analysis in a more focused manner. We believe that these revisions address your concerns by clarifying the manuscript’s central contribution, strengthening its analytical depth, and improving the coherence between the theoretical framework, empirical analysis, research questions, results, and conclusions. Thank you again for your valuable comments.

 

2. Review report comment

 

The paper engages with a wide range of theoretical frameworks, including surveillance capitalism, data colonialism, and trust, but these are not sufficiently integrated into the analysis. The literature review is extensive, yet much of this conceptual material is not consistently mobilised in the interpretation of the findings. For example, while concepts such as surveillance capitalism and data colonialism are used to frame the broader dynamics of data extraction and power, they are not clearly used in analysing how students understand or describe their experiences of surveillance. As a result, there is a noticeable gap between the paper’s theoretical framing and its empirical analysis. The authors would benefit from clarifying which conceptual framework is doing the primary analytical work and carrying that framework more explicitly through the results and discussion.

 

Author's reply to the comment 2

 

Thank you for your comment. We agree that the manuscript could be improved by incorporating an analysis of key concepts (surveillance capitalism, data colonialism, and trust). In response to your suggestion, we have made several changes to ensure that these concepts, which are indeed key, are reflected in the interpretation of the results.

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted (lines 340-345; 498-506; 513-515; 579-583; 630-633).

 

3. Review report comment

The methodological design is appropriate in principle, and the survey provides a solid base for exploratory analysis. However, aspects of the methodological justification require further attention. In particular, the qualitative component raises some concerns. The study includes a single focus group, but the rationale for this is not clearly explained. In its current form, it is unclear what role this component is intended to play within the overall research design. If the focus group is meant to support a mixed-methods approach or provide triangulation, one group is not sufficient to sustain such claims.

 

Author's reply to the comment 3

In the section where we describe the methodology, we have added an explanatory paragraph setting out the rationale for choosing a focus group. In this paragraph, we explain that the aim is not methodological triangulation, but rather to explore the results of the open-ended responses in greater depth and to identify trends, opinions, and attitudes, as is typical of an exploratory study (lines 391-420).

 

4. Review report comment

 

Alternatively, if it is intended as an exploratory or illustrative element, this should be explicitly stated and reflected in how the findings are presented. As it stands, the qualitative component appears somewhat underdeveloped and not fully integrated into the analytical framework, which limits its contribution to the study.

 

Author's reply to the comment 4

The core of the analysis is based on the role of quantitative analysis. Upon revisiting our manuscript, using the qualitative analysis as complementary to reinforce the quantitative results obtained but not to generate independents findings. This aspect has been corrected and specified in the text (page 10). Specifically, the semantic components identified in the focus group are used to:

• Corroborate the internal consistency of the quantitative data.
• Provide validity to the observed statistical patterns.

The lesser degree of development of the qualitative analysis is intentional and consistent with its supportive role, as now made explicit in the manuscript. Additionally, we have also added a clarification in the results obtained (page 16) that delimits the scope of the qualitative component and points out semantic evidence examples of this reinforcement for a better understanding.

 

5. Review report comment

The analysis itself is clearly presented but remains largely descriptive. Much of the results section focuses on reporting what participants think or feel, with limited interpretive development. For instance, students’ perceptions that their mobile phones are “listening” to them are reported as evidence of a broader sense of surveillance, but this is not further analysed in relation to algorithmic literacy or the opacity of data-driven systems, nor does the paper consider why such beliefs persist. Given the paper’s conceptual framing, there is scope for a more sustained analytical engagement with what these perceptions mean, how they relate to broader processes of datafication, and how they shape emerging professional identities. Bringing this level of interpretation more explicitly into the analysis would significantly strengthen the manuscript.

 

Author's reply to the comment 5

This very point was highlighted by reviewer number 2. To address this shortcoming, the changes made aim to link the findings to key concepts within the theoretical framework, interpreting the findings primarily through the lens of surveillance capitalism and, to a lesser extent, through other concepts such as data colonialism and trust. We set out below the changes made:

Firstly, in the theoretical framework, we have added a further paragraph to clarify which aspects of the theoretical framework govern the analytical approach, reinforcing the idea that the concept of ‘surveillance capitalism’ serves as the lens through which students’ perceptions of algorithmic surveillance or data control are interpreted: (line 340-344; 498-506; 513-515; 578-583; 630-633).

 

6. Review report comment

There is also a tendency to move from a relatively limited and context-specific sample to broader claims about democracy, trust, and the future of journalism. For example, the discussion links students’ perceptions of data use to wider questions of democratic legitimacy, and the conclusion draws implications for the health of democratic systems and the role of universities in sustaining them. However, these connections are not clearly developed in the analysis and, at times, read more like a conceptual leap than an argument grounded in the data. The paper would benefit from more careful alignment between its empirical scope and the claims it advances, and from a clearer acknowledgement of its limitations.

 

Author's reply to the comment 6

In fact, this clarification is important because it would reinforce the integrity with regard to the objectives set, without overstepping them. To address this shortcoming, we have included some explanatory paragraphs at the points where democratic legitimacy is referred to. In the discussion section, for example, we have clarified the meaning of the interpretation provided by highlighting the limitations and restricting the interpretation to the responses of the respondents (lines 678-682; 722.727).

 

7. Review report comment

Finally, the overall structure could better support the development of the argument. There is some repetition across sections, particularly where key findings are restated without further analytical development. For example, the tension between students’ discomfort as users and their acceptance of data practices as future professionals appears in the results, discussion, and conclusion, but is largely reiterated rather than conceptually deepened. A similar pattern is evident in the treatment of ethical concerns, where calls for transparency and consent recur across sections without being further theorised. Reducing this repetition and more clearly building analytical progression across sections would strengthen the paper’s overall contribution.

 

Author's reply to the comment 7

We sincerely thank the reviewer for their constructive comments and suggestions on this work. Thank you.

As noted above, the overall structure has been drafted to highlight the central thread of the research and the development of the argument. However, we wish to make it clear that the approach does not reaffirm the findings, but rather builds on them to develop the conceptual incompatibilities of the user–professional tension identified in the results. The approach helps to explain why future journalists both resist and normalise algorithmic logics, and highlights the need for pedagogical interventions that explicitly address this duality.

In this way, the paper’s contribution lies not merely in describing students’ perceptions, but above all in revealing the tension that shapes their emerging ethical orientations and professional identities. That said, we have intended that, while the empirical sections describe how students articulate this duality, the discussion expands on its meaning for the formation of journalistic identity in datafied environments. Finally, the conclusion does not revisit the empirical results but synthesises the analytical insights derived from them, clarifying how this apparent contradiction between personal discomfort and professional acceptance is important when considering the emerging ethical orientations of future journalists."

 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

While the use of Gemini Advanced as an auxiliary tool for preliminary thematic coding is acknowledged in the AI declaration and described in the Methods section, several important methodological and ethical concerns remain unaddressed.

First, submitting open-ended student responses — which contain personal opinions about surveillance, political views, and privacy concerns — to a cloud-based proprietary platform owned by Google raises significant data protection questions, particularly under GDPR, which applies to the University of Seville.

The paper claims to guarantee participants' anonymity but does not discuss whether informed consent covered the transfer of their responses to third-party AI services, nor whether a Data Processing Agreement was in place. This omission is especially striking given that the study's central theme is the sensation of being "spied on" by digital platforms: the methodological choice of routing participant data through one of the very corporations discussed in the theoretical framework warrants explicit reflection.

Furthermore, no justification is offered for selecting a proprietary cloud-based model over locally deployable open-weight alternatives (e.g. LLaMA, Mistral), which would have avoided third-party data transfer altogether and better aligned the research practice with the ethical principles the paper itself advocates.

Second, the paper does not engage with well-documented limitations of large language models when applied to qualitative analysis. These include systematic biases in pattern recognition, the tendency to impose artificial coherence on heterogeneous data, and potential cultural and linguistic skew — a particularly relevant concern given that student responses were presumably collected in Spanish and processed by a model predominantly trained on English-language data. No information is provided about the specific model version, parameter settings, or prompts used, making the AI-assisted phase of the analysis entirely non-reproducible.

The paper would also benefit from acknowledging the environmental cost of using cloud-based generative AI, an increasingly standard expectation in research involving computational tools. Addressing these gaps would strengthen the methodological transparency of the study and bring its research practices into closer alignment with the principles of accountability and informed consent that it rightly calls for in journalistic and political data use.

Comments on the Quality of English Language

Typos are still visible in the paper, which is not easy to understand provided the acknowledged use of AI tools also for writing assistance. Please, thoroughly review the text and ensure that grammar and typos are corrected before resubmitting.

Author Response

ANSWER TO REVIEWERS IN ROUND 2. POINT-TO -POINT RESPONSES REPORT.

REVIEWER 1

1. Review report comment

While the use of Gemini Advanced as an auxiliary tool for preliminary thematic coding is acknowledged in the AI declaration and described in the Methods section, several important methodological and ethical concerns remain unaddressed.

First, submitting open-ended student responses — which contain personal opinions about surveillance, political views, and privacy concerns — to a cloud-based proprietary platform owned by Google raises significant data protection questions, particularly under GDPR, which applies to the University of Seville.

The paper claims to guarantee participants' anonymity but does not discuss whether informed consent covered the transfer of their responses to third-party AI services, nor whether a Data Processing Agreement was in place. This omission is especially striking given that the study's central theme is the sensation of being "spied on" by digital platforms: the methodological choice of routing participant data through one of the very corporations discussed in the theoretical framework warrants explicit reflection.

Furthermore, no justification is offered for selecting a proprietary cloud-based model over locally deployable open-weight alternatives (e.g. LLaMA, Mistral), which would have avoided third-party data transfer altogether and better aligned the research practice with the ethical principles the paper itself advocates.

 

Author's reply to the review report 1

We would like to thank the reviewer for this important point from an ethical perspective. We fully agree that the use of AI tools on a database stored by the authors in the cloud requires a careful and thoughtful explanation, particularly when the content of the responses addresses subjective perceptions regarding the sense of surveillance on digital platforms.

 

With regard to data protection, we would like to emphasise that all responses, including open-ended ones, were collected entirely anonymously, without respondents having to provide names, surnames, email addresses, or IP addresses that would allow for identification. Respondents voluntarily accessed the Google Forms questionnaire, being aware of the ownership and corporate nature of the tool used to administer the survey. Furthermore, they were properly informed of the purpose of the research and the use of the responses, intended for the development of the research, and the anonymous nature of the responses. No special categories of personal data were collected in the questionnaire, in accordance with Article 9 of the GDPR.

Regarding the choice between a proprietary platform and the alternative of locally deployed locally, we appreciate the suggestion, which will be taken into account in future research. While we are aware that this alternative would reduce dependence on third parties, its use required technical infrastructure and institutional support that were not available in the context of this research, for which no external funding was received. We acknowledge, in any case, the methodological tension that might arise from using a tool developed by large corporations that are critically analysed within the theoretical framework.

In order to provide a more thorough and thoughtful explanation of both aspects, a new paragraph has been included in the manuscript, within the Methodology section, in which we emphasise this point, particularly from an ethical perspective (lines 412-439). Furthermore, in the section ‘Declaration of generative AI and AI-assisted technologies in the writing process’ (lines 807-811), we have reinforced the idea that the use of Gemini Advanced is auxiliary in nature and the responsibility assumed by researchers regarding all matters related to methodological design. Additionally, the section “Ethics statement” (lines 784-789) has been revised to emphasize the anonymous nature of the responses and the ethical commitment to the handling open responses.

2. Review report comment

Second, the paper does not discuss the well-documented limitations of large language models when applied to qualitative analysis. These include systematic biases in pattern recognition, the tendency to impose artificial coherence on heterogeneous data, and potential cultural and linguistic skew - a particularly relevant concern given that student responses were presumably collected in Spanish and processed by a model predominantly trained on English-language data. No information is provided about the specific model version, parameter settings, or prompts used, making the AI-assisted phase of the analysis entirely non-reproducible.

Author's reply to the review report 2

We appreciate this comment regarding the documented limitations of language models when AI is applied to qualitative analysis. We fully share the view that biases may arise in pattern detection, or even cultural and linguistic biases. The original responses were collected in Spanish, as they were obtained within the context of a Spanish university. For this reason, the human factor was emphasised at all times, whilst the use of AI was, as previously described, merely auxiliary and never a substitute for human qualitative analysis. This use, a phased process, was explained beforehand to one of the reviewers in round 1 in the following terms:

- Before the interaction with AI tools, the researchers defined basic concepts, the analytical framework, and the coding orientation. The decisions about what could constitute relevant thematic categories or how to structure the coding were entirely human-driven. At this stage, Gemini Advanced was used only to generate a first-pass exploratory grouping of semantic similarities in open-ended responses. This step is comparable to using automated techniques such as word frequency clustering or topic extraction to identify possible patterns, always used as a proposal and not as definitive coding tool. Gemini did not decide about codes, categories, or interpretations. It only reduced the preliminary task of scanning large text segments for recurring ideas.

- The process was completed by human verification, recoding, and conceptual refinement. Therefore, we conducted a full second sequence of coding, which involved: reviewing all open-ended responses line by line; comparing AI-suggested gatherings with the raw data; modifying, confirming, merging, or removing AI-generated groupings; creating new codes in case it was necessary, based exclusively on human interpretation; and guaranteeing theoretical coherence with the concepts used in the quantitative sections of the study. This second cycle followed the principles of reflexive thematic analysis, including iterative engagement with the data, constant comparison, attention to context, and theoretical adequacy.

- Only after this human-driven recoding were the final themes defined, conceptualized and integrated with the quantitative findings. All interpretive decisions, the naming of the themes, and the theoretical connections, were totally human-made.

Consequently, the use of Gemini Advanced was limited to lexical recurrences and low-level semantic similarities. As we indicated in the first report, a change was made in the first version of the manuscript, incorporating a paragraph in which we clarified the process through which AI-assisted tools were used during qualitative analysis. However, we also indicated that we believed that providing such a level of detail as that indicated in this report could unconsciously produce the opposite effect: rather than improving transparency regarding the use and reinforcing the ethical validity of its integration, we have opted to include two more concise paragraphs in the manuscript (lines 391–409 and 412–421 in the manuscript submitted to the second round reviewers), explaining the auxiliary nature of the use of Gemini Advanced and the essential role of human-led analytical tasks.

In this new version, we have expanded on the methodological explanation, specifying that the analysis was carried out using Gemini Advanced (the commercial version available at the time of the analysis) (lines 391-393 and 412-418), without additional training, using restrictive prompts designed to identify lexical recurrences. Logically, the AI-assisted phase would not be fully reproducible, due to the evolutionary nature of the model and the ongoing training of these systems, independent of the research activities. In any case, the analytical reproducibility lies in human coding, which forms the basis of qualitative analysis. The limitations that language and potential cultural biases may pose have also been highlighted, hence the importance of a human-led approach to the analysis.

3. Review report comment 3

The paper would also benefit from acknowledging the environmental cost of using cloud-based generative AI, an increasingly standard expectation in research involving computational tools. Addressing these gaps would strengthen the methodological transparency of the study and bring its research practices into closer alignment with the principles of accountability and informed consent that it rightly calls for in journalistic and political data use.

 Author's reply to the review report 3

We thank the reviewer for this comment, which will be taken into account in future research. With regard to the present study, this environmental impact has been explicitly acknowledged among the limitations of the research in the Methods section, so future researchers who read the article or use it as a documentary reference (lines 441-446).

4. Review report comment 4

Typos are still visible in the paper, which is not easy to understand provided the acknowledged use of AI tools also for writing assistance. Please, thoroughly review the text and ensure that grammar and typos are corrected before resubmitting. Comments on the Quality of English Language.

Review report comment 4

We thank you for your feedback, as it helps us to systematically improve the manuscript. We have carefully reviewed the aspects relating to language and grammar, the use of capital letters, errors in citations, and unnecessary double spaces. The changes made have been highlighted in red.

Reviewer 3 Report

Comments and Suggestions for Authors

I would like to thank the authors for their careful and considered revisions. The manuscript has improved, particularly in clarifying the central argument and strengthening the connection between the theoretical framework and the empirical analysis.

There is still some scope to further strengthen the analytical depth in places. In particular, the results section would benefit from briefly moving beyond description to indicate what these findings suggest in relation to the paper’s central argument. For example, where students describe feeling that their devices are “listening,” this could be more explicitly interpreted in relation to how students make sense of surveillance and the limited visibility of how data-driven systems operate.

Similarly, the tension between students’ experience as users and their orientation as future professionals is now clearly identified, but could be drawn through more consistently as a framing device in the analysis rather than appearing at key points.

These are relatively minor points. The paper is now suitable for publication.

Author Response

REVIEWER 2

1. Review report comment

I would like to thank the authors for their careful and considered revisions. The manuscript has improved, particularly in clarifying the central argument and strengthening the connection between the theoretical framework and the empirical analysis.

There is still some scope to further strengthen the analytical depth in places. In particular, the results section would benefit from briefly moving beyond description to indicate what these findings suggest in relation to the paper’s central argument. For example, where students describe feeling that their devices are “listening,” this could be more explicitly interpreted in relation to how students make sense of surveillance and the limited visibility of how data-driven systems operate.

Author's reply to the comment 1

We sincerely thank the reviewer for this constructive suggestion, which has helped us to further strengthen the analytical depth of the manuscript.

In response, we have revised the Results section to move more explicitly beyond description in selected passages and to clarify how the empirical findings relate to the paper’s central argument. Specifically, we have added interpretative statements that connect students’ responses to broader processes of datafication, algorithmic opacity, and contemporary forms of surveillance (pp. 11-15).

Following the reviewer’s example, we now explicitly interpret students’ perceptions that their devices are “listening” as an attempt to make sense of data‑driven systems whose operations remain largely invisible to users. We discuss how visible effects such as hyper‑contextualised advertising are transposed into a broader sense of continuous technological surveillance, analysing the responses extracted from the students' survey about algorithmic control and limited transparency (pp. 12–13).

In addition, we have strengthened the interpretation of students’ awareness of political microtargeting and their scepticism towards digital platforms, linking these findings to perceptions of depersonalised control, loss of autonomy, and the attribution of structural power to data‑driven systems (pp. 12). These revisions aim to reinforce the analytical contribution of the Results section while remaining firmly grounded in the empirical material.

At the same time, we have reinforced the idea that the respondents are still confident on democratic processes despite digital media manipulation (pp. 14-15).

 We believe that these changes improve the clarity of the argument and better articulate how students interpret and negotiate contemporary forms of surveillance within opaque data ecosystems.

 2. Review report comment

Similarly, the tension between students’ experience as users and their orientation as future professionals is now clearly identified, but could be drawn through more consistently as a framing device in the analysis rather than appearing at key points.

These are relatively minor points. The paper is now suitable for publication.

 Author's reply to the comment 2

We thank the reviewer for this careful reading and for recognising the improvements introduced in the revised version of the manuscript.

As suggested, the ambivalent position of respondents as journalism students and future journalists has now been explicitly incorporated into the introductory framework (pag 2), where it is presented as a key contextual condition shaping the analysis and reinforced in the microstrategy section (pag 3). This dual position is subsequently highlighted in the Results section, particularly when interpreting students’ perceptions of algorithmic surveillance, trust in platforms, and professional responsibility.

At the same time, we would like to clarify that this ambivalence functions as an analytical lens rather than as the central object of the analysis. The primary aims of the study are to examine:

(a) how young users embedded in highly datafied environments perceive algorithmic surveillance and data-driven political communication;

(b) how these perceptions relate to trust in digital platforms, media, and political actors; and

(c) how such perceptions are linked to broader concerns about democratic legitimacy, political microtargeting, and the opacity of data-driven systems.

Within this framework, the dual role of journalism students is mobilised to contextualise and interpret these perceptions, rather than to offer a dedicated analysis of professional identity formation per se. We believe this approach allows the article to maintain its focus on surveillance, trust, and democratic implications, while still acknowledging the specific relevance of journalism students as a strategic group.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

The reviewer acknowledges the authors' efforts to address the methodological concerns raised in the previous round, and recognises several genuine improvements in the revised manuscript, including the restructuring of the abstract, the reformulation of research questions, the expanded methodological justification, and the addition of figures and tables that strengthen the presentation of results. These revisions reflect a conscientious engagement with the review process.

However, significant concerns remain, and the reviewer is unable to recommend acceptance at this stage.

**On typographic and formatting quality.** It is difficult to understand how, after explicit concerns were raised about the quality of the manuscript, errors as visible as "Toutube" appear in the new text, or the awkward parenthetical formatting of references — where every entry systematically begins with "(" — persists throughout. These are not subtle issues; they are immediately apparent upon any visual inspection of the text. This raises questions about the level of care with which the manuscript was revised prior to resubmission, and the degree to which authors have genuinely engaged with the feedback received.

**On the use of AI tools.** The reviewer's concerns about the proficient and transparent use of AI tools have not been adequately addressed. In their response, the authors have in fact introduced additional AI tools whose use has not been properly acknowledged in the manuscript. More critically, in response to concerns about the reliability of Gemini Advanced for qualitative analysis, the authors now claim that prior conceptual work was developed before employing the tool — yet no evidence of this work is provided, nor is there any indication of how it was applied or how it constrained the model's outputs. The authors are requested to provide: (a) the conceptual coding framework developed prior to AI-assisted analysis; and (b) a systematic cross-check demonstrating the extent to which the reported results derive from researcher interpretation versus Gemini output.

More concerning still is the authors' claim, in their response, that Gemini Advanced is equivalent in nature to standard statistical tools. This statement reveals a fundamental misunderstanding of the tools being used, and one that is directly relevant to the validity of the study's findings. Unlike conventional statistical tools — including machine learning methods — the output of large language models (LLMs) is non-deterministic and non-explainable. Results can vary across runs, and the model may generate outputs that are entirely fabricated, a phenomenon well documented in the literature as "hallucination." For this reason, researchers employing LLMs are universally expected to commit to full methodological transparency: without disclosure of model version, parameters, and prompts, peers cannot replicate the analysis, cannot identify possible errors, and cannot assess the validity of the results. None of this information is currently provided, as neither it is the relevant literature. 

Furthermore, the use of LLMs entails significant computational costs, including energy consumption and water use for cooling, that require explicit justification. If conventional analytical tools could have accomplished the same task, authors must explain what Gemini Advanced adds to the analysis that those tools cannot provide. In the absence of such justification, the use of a cloud-based proprietary model of this kind is difficult to defend on either scientific or ethical grounds. The reviewer therefore considers that this methodology can only be deemed acceptable if authors provide: the raw prompts used to instruct the model; the results of the conceptual coding exercise that supposedly preceded Gemini use; and a clear account of how human and AI contributions are distinguished in the final results.

**On data privacy.** The authors respond to privacy concerns by asserting that no personal data was involved. The reviewer cannot accept this response as adequate. The data collection instrument used was Google Forms, and the analysis was subsequently conducted using Gemini Advanced — both products operated by Alphabet Inc. This means that the same corporate entity controlled both data collection and analysis. Regardless of whether students were asked to provide their names, Alphabet is technically positioned to cross-reference the two datasets and re-identify participants. Students who completed the Google Form using a personal device — the overwhelmingly most likely scenario — will have provided personally identifiable signals either directly (through a Google or Gmail account linked to their Android device or browser) or indirectly through their digital fingerprint: the specific combination of operating system, browser version, screen resolution, and device configuration constitutes a quasi-unique identifier. The claim that anonymity was guaranteed is therefore not technically supportable under the conditions described, and this has not been addressed in the revision. Authors should really engage with this issue, explaining how the referred issues have been addressed, if any, and noting the relevant literature on the topic.

**Conclusion and path forward.** The reviewer wishes to be clear that the results of this study are of genuine interest, and the underlying research questions are valuable. The reviewer's role is to help the authors deliver their findings in a form that meets scientific and ethical standards. With that goal in mind, the manuscript can only be considered acceptable after the following conditions are met:

1. A full and thorough typographic revision is carried out prior to resubmission, ensuring complete compliance with MDPI formatting standards and correct academic English throughout. Professional editing services should be used if necessary. No further typographic or formatting errors should appear in the resubmitted version.

2. Sufficient materials are made available for independent peer verification of results, including: the raw survey data, carefully anonymised before release; the conceptual coding framework developed prior to the use of Gemini Advanced; and the specific prompts used to instruct the model during analysis.

3. A full and transparent acknowledgement of all AI tools used in the research process is provided, including any tools used for writing assistance. A dedicated limitations section should explicitly recognise the known problems of employing LLMs in scientific research — non-determinism, non-explainability, hallucination risk, and reproducibility constraints — and acknowledge how these limit the generalisability of the present findings. Additionally, a paragraph of lessons learnt should be included, offering concrete recommendations for future researchers to avoid the privacy risks that have complicated the present study. As a minimum, future studies should avoid using a single platform ecosystem for both data collection and data analysis: if Google Forms is used for data collection, analysis should be conducted using tools operated by a different provider, and ideally using locally deployable open-weight models that do not require routing participant data through third-party cloud infrastructure. Data collection should in any case prioritize university labs under controlled conditions, avoiding under all circumstances using personal devices whose characteristics can be easily used to re-identify the data.

Comments on the Quality of English Language

Typos are still visible in the paper, which is not easy to understand provided the acknowledged use of AI tools also for writing assistance. Please, thoroughly review the text and ensure that grammar and typos are corrected before resubmitting.

Author Response

ANSWER TO REVIEWERS IN ROUND 3. POINT-TO -POINT RESPONSES REPORT.

Introduction

The reviewer acknowledges the authors' efforts to address the methodological concerns raised in the previous round, and recognises several genuine improvements in the revised manuscript, including the restructuring of the abstract, the reformulation of research questions, the expanded methodological justification, and the addition of figures and tables that strengthen the presentation of results. These revisions reflect a conscientious engagement with the review process.

However, significant concerns remain, and the reviewer is unable to recommend acceptance at this stage.

1. Review comment 

**On typographic and formatting quality.** It is difficult to understand how, after explicit concerns were raised about the quality of the manuscript, errors as visible as "Toutube" appear in the new text, or the awkward parenthetical formatting of references — where every entry systematically begins with "(" — persists throughout. These are not subtle issues; they are immediately apparent upon any visual inspection of the text. This raises questions about the level of care with which the manuscript was revised prior to resubmission, and the degree to which authors have genuinely engaged with the feedback received.

Author's reply to comment 1

The second round of proofreading involved a thorough review of the manuscript. Specifically, the word ‘Toutube’ had been corrected on line 364 (marked red in the second version submitted). We have checked the text again, and there are no errors in the spelling of that word. For this reason, we wanted to ensure that the reviewer had correctly received the attached file containing the changes made to the manuscript in the second version. 

Regarding the use of parentheses, according to the journal editor’s instructions, this is a valid format; we have chosen to retain this system to facilitate visual correspondence with the citations in the text.

2. Review comment

**On the use of AI tools.** The reviewer's concerns about the proficient and transparent use of AI tools have not been adequately addressed. In their response, the authors have in fact introduced additional AI tools whose use has not been properly acknowledged in the manuscript. More critically, in response to concerns about the reliability of Gemini Advanced for qualitative analysis, the authors now claim that prior conceptual work was developed before employing the tool — yet no evidence of this work is provided, nor is there any indication of how it was applied or how it constrained the model's outputs. The authors are requested to provide: (a) the conceptual coding framework developed prior to AI-assisted analysis; and (b) a systematic cross-check demonstrating the extent to which the reported results derive from researcher interpretation versus Gemini output.

More concerning still is the authors' claim, in their response, that Gemini Advanced is equivalent in nature to standard statistical tools. This statement reveals a fundamental misunderstanding of the tools being used, and one that is directly relevant to the validity of the study's findings. Unlike conventional statistical tools — including machine learning methods — the output of large language models (LLMs) is non-deterministic and non-explainable. Results can vary across runs, and the model may generate outputs that are entirely fabricated, a phenomenon well documented in the literature as "hallucination." For this reason, researchers employing LLMs are universally expected to commit to full methodological transparency: without disclosure of model version, parameters, and prompts, peers cannot replicate the analysis, cannot identify possible errors, and cannot assess the validity of the results. None of this information is currently provided, as neither it is the relevant literature. 

Furthermore, the use of LLMs entails significant computational costs, including energy consumption and water use for cooling, that require explicit justification. If conventional analytical tools could have accomplished the same task, authors must explain what Gemini Advanced adds to the analysis that those tools cannot provide. In the absence of such justification, the use of a cloud-based proprietary model of this kind is difficult to defend on either scientific or ethical grounds. The reviewer therefore considers that this methodology can only be deemed acceptable if authors provide: the raw prompts used to instruct the model; the results of the conceptual coding exercise that supposedly preceded Gemini use; and a clear account of how human and AI contributions are distinguished in the final results.

Author's reply to the comment 2

 

We appreciate the precision and level of detail in the reviewer’s comments. We agree on the importance of methodological transparency, analytical traceability and the responsible use of language models, particularly in qualitative studies. At the same time, we believe that some of the concerns raised stem from a lack of clarity in the previous version of the article, rather than from weaknesses in the methodological design or the conduct of the study. In this regard, the review has provided an opportunity to highlight procedures that were already part of the analytical work and are now documented in greater detail.

Below, we respond directly to each of the issues raised, whilst also incorporating into the revised article all the information necessary to ensure methodological transparency.

Firstly, regarding the conceptual work carried out prior to the use of AI tools, we wish to clarify that the qualitative analysis was based on a conceptual coding framework developed by the researchers prior to any interaction with Gemini Advanced. This framework was guided by the study’s central theoretical constructs—algorithmic surveillance, datafication, trust, political micro-segmentation and professional ethics—and by sensitising concepts derived from the literature on surveillance capitalism, data colonialism and algorithmic opacity. In the revised version of the manuscript, this framework is now explicitly described (edits are highlighted in red) and is also included as a structured table in the Supplementary Materials (Qualitative Coding Framework and AI-Assisted Exploratory Support), to facilitate peer review. This supplementary document includes verbatim examples of the prompts used, which were employed to request descriptive groupings of recurring terms. This framework not only preceded the use of the tool but also functioned as an explicit methodological constraint: all output generated by Gemini was evaluated solely on the basis of its consistency with these categories. We consider it important to emphasise that this procedure was part of the study’s original design, although it had not been described in the level of detail that is now included.

Secondly, regarding the cross check between human and model results (request b), the suplementary materials includes a specific section (A.3) detailing the procedure followed. The analysis was organised into two clearly distinct phases: an exploratory phase assisted by Gemini Advanced and a second phase of complete and independent manual coding. During this second phase, all model outputs were thoroughly reviewed against the original data and classified as accepted, rejected, or reformulated. This process involved verifying the semantic coherence of the groupings, identifying inconsistencies, and, where necessary, redefining or generating new categories based solely on human interpretation. Furthermore, the applied criteria are made explicit—primacy of human interpretation, theoretical coherence, empirical traceability and alignment with the study’s objectives. This level of detail clearly demonstrates that the final results do not derive from the model, but from a human analytical process.

Furthermore, we agree with the reviewer in highlighting the issues of non-determinism, opacity and the risk of hallucination associated with large-scale language models. In this regard, we wish to make it clear that we do not consider Gemini Advanced to be equivalent to conventional statistical tools, and we regret that an earlier formulation may have suggested such an equivalence. The manuscript has been revised to remove any such comparisons. We have removed any wording that might suggest equivalence with statistical tools and have included an explicit clarification: language models are non-deterministic systems, whose results may vary between runs and which may generate unverifiable content (hallucinations), as documented in the specialist literature. We emphasise that Gemini Advanced was not used as an analytical or interpretative tool, but exclusively as an auxiliary aid in an initial exploratory phase, limited to the identification of lexical recurrences and superficial semantic similarities in a large corpus of open-ended responses in Spanish.

Furthermore, regarding the possible use of undeclared AI tools, we wish to clarify that no additional systems other than Gemini Advanced were employed. References to techniques such as clustering or pattern detection were intended solely as functional descriptions of the type of task performed, and not as an indication of the actual use of specific tools. This potential ambiguity, which we acknowledge may have caused confusion, has been corrected in the revised version.

As for transparency, prompts and reproducibility, we have also responded directly to this requirement by including in the supplementary materials (section A.2) verbatim examples of the prompts used. These prompts were deliberately restrictive and non-interpretative, aimed solely at identifying lexical repetitions and superficial similarities.

Thirdly, regarding the distinction between human and AI-assisted contributions, we have clarified the entire analysis workflow in the Methods section. All interpretative work—including the validation, rejection, merging or reformulation of clusters suggested by the AI; the creation of codes; the naming of themes; and the integration of qualitative and quantitative results—was carried out entirely by the researchers, through a second phase of reflective coding in accordance with established principles of qualitative analysis. The manuscript now includes an explicit description of the systematic cross-checking process between the initial AI outputs and the raw data, indicating that only those themes fully supported by human interpretation and the theoretical coherence of the study were incorporated into the final analysis.

Finally, with regard to ethical and environmental considerations, we have revised the paragraphs that highlight limitations to explicitly address the environmental impact associated with the use of generative models in the cloud, as well as the methodological tension involved in using proprietary platforms in research critical of digital surveillance. We clarify that the use of Gemini Advanced in this study was limited in scope and exploratory in nature, constrained by the institutional infrastructure available at the time of the research, and we have made specific recommendations for future research, including the use of local or open source models where technically feasible.

We consider that these revisions substantially strengthen the methodological transparency of the study, clearly delineate the role of AI, and directly address the concerns raised by the reviewer about interpretative validity, reproducibility, and ethical accountability. 

 

3. Review comment

**On data privacy.** The authors respond to privacy concerns by asserting that no personal data was involved. The reviewer cannot accept this response as adequate. The data collection instrument used was Google Forms, and the analysis was subsequently conducted using Gemini Advanced — both products operated by Alphabet Inc. This means that the same corporate entity controlled both data collection and analysis. Regardless of whether students were asked to provide their names, Alphabet is technically positioned to cross-reference the two datasets and re-identify participants. Students who completed the Google Form using a personal device — the overwhelmingly most likely scenario — will have provided personally identifiable signals either directly (through a Google or Gmail account linked to their Android device or browser) or indirectly through their digital fingerprint: the specific combination of operating system, browser version, screen resolution, and device configuration constitutes a quasi-unique identifier. The claim that anonymity was guaranteed is therefore not technically supportable under the conditions described, and this has not been addressed in the revision. Authors should really engage with this issue, explaining how the referred issues have been addressed, if any, and noting the relevant literature on the topic.

Author's reply to comment 3

We appreciate the reviewer’s comment, which we consider relevant. The previous language on the absence of personal data did not adequately reflect the technical complexity of the digital environment in which the study was conducted and has been revised accordingly.

In the current version of the article, we explicitly acknowledge that the combined use of Google Forms and Gemini Advanced implies, in strictly technical terms, the possibility of data correlation at the infrastructure level. We have removed any claims of absolute anonymity. Students who participated in the survey were informed of the mechanism used in the investigation, as well as all its implications. We consider this to be a vital part of our research work.

However, we consider it important to clarify that the study applied measures consistent with standard practices in social research and with the principles of the General Data Protection Regulation, in particular, data minimisation and purpose limitation: no direct identifiers were collected, the data were processed exclusively in anonymised text format and no profiling of any kind was carried out.

Recent literature agrees that absolute anonymity is difficult to guarantee in contemporary digital environments and that privacy should be approached in terms of risk reduction rather than total elimination. In this regard, the risk highlighted by the reviewer is technically valid, but stems from a limitation of the current digital ecosystem, not from a specific omission in the research design.

The article has been revised to accurately reflect The context; the limits of anonymity are now made explicit, the measures taken are detailed, and this issue has been incorporated as a limitation of the study. We consider that this reformulation improves transparency and brings the work into line with contemporary methodological and ethical standards.

 

4. Review comment (summary)

**Conclusion and path forward.** The reviewer wishes to be clear that the results of this study are of genuine interest, and the underlying research questions are valuable. The reviewer's role is to help the authors deliver their findings in a form that meets scientific and ethical standards. With that goal in mind, the manuscript can only be considered acceptable after the following conditions are met:

1. A full and thorough typographic revision is carried out prior to resubmission, ensuring complete compliance with MDPI formatting standards and correct academic English throughout. Professional editing services should be used if necessary. No further typographic or formatting errors should appear in the resubmitted version.
2. Sufficient materials are made available for independent peer verification of results, including: the raw survey data, carefully anonymised before release; the conceptual coding framework developed prior to the use of Gemini Advanced; and the specific prompts used to instruct the model during analysis.
3. A full and transparent acknowledgement of all AI tools used in the research process is provided, including any tools used for writing assistance. A dedicated limitations section should explicitly recognise the known problems of employing LLMs in scientific research — non-determinism, non-explainability, hallucination risk, and reproducibility constraints — and acknowledge how these limit the generalisability of the present findings. Additionally, a paragraph of lessons learnt should be included, offering concrete recommendations for future researchers to avoid the privacy risks that have complicated the present study. As a minimum, future studies should avoid using a single platform ecosystem for both data collection and data analysis: if Google Forms is used for data collection, analysis should be conducted using tools operated by a different provider, and ideally using locally deployable open-weight models that do not require routing participant data through third-party cloud infrastructure. Data collection should in any case prioritize university labs under controlled conditions, avoiding under all circumstances using personal devices whose characteristics can be easily used to re-identify the data.

Author's reply to comment 4

We would like to thank the reviewer for the depth of his comments and for recognising the significance of the results and research questions. We have carefully addressed all the comments with the aim of bringing the article into line with the standards of transparency, methodological rigour, and ethical responsibility required by the journal. At the same time, we consider it important to distinguish between requirements necessary for the evaluation of the work and methodological recommendations which, whilst valuable, exceed what is reasonably required in the context of peer review.

4.1.

We have endeavored to address the various points raised by the reviewer, as set out in the preceding sections (number 1).

4.2.

Regarding the public availability of the survey, we have informed the journal that the Excel file is available upon request to the authors, so that there is a responsible custodian of the data and to avoid potential risks of reidentification. Given that the reviewer’s own report acknowledges the possibility of reidentification in digital environments, open publication of the data would not be compatible with current ethical standards. Therefore, we have adopted a model of controlled access upon reasoned request, a practice widely accepted in social research. We believe that this solution strikes an appropriate balance between transparency and data protection.

Supplementary materials are attached to the manuscript, in which we include the preliminary coding framework table, verbal examples of prompts used, and an outline of the process followed in the AI-human interpretation interaction.

4.3.

We have substantially improved the transparency of the manuscript by: (1) explicitly stating all the AI tools used; (2) including a detailed section on limitations that addresses nondeterminism, hallucinations, applicability and reproducibility; and (3) including a section with recommendations for future research.

With regard to the use of AI, the tools employed are those declared in the previously submitted manuscripts, specifically in the supplementary section: declaration of AI use. As for the draughting process, we would emphasise that the Writefull tool, provided by the University of Seville and integrated into the word processor, obviously has limitations that must be addressed through a careful proofreading of the text. The purpose of Writefull is to refine the writing of academic texts, suggesting changes more appropriate to scientific language. Had other tools been used, the text would be free from typographical or spelling errors, for example, which have had to be corrected during the various rounds of manuscript review.

In any case, your interest and concerns, as the stated use of AI continues to raise suspicions within the research community. Our aim has been to use these tools responsibly, with transparency and ethical rigour. Therefore, we welcome any suggestions for improvement. We are aware that differing perspectives on AI in research enrich the debate and promote ethical use, so we appreciate your interest.

Regarding the additional recommendations—such as avoiding the use of a single technology provider, employing local models, or restricting data collection to fully controlled environments, we recognise their value as best practices in certain contexts and have incorporated them as guidelines for future work. Thank you.

However, from an editorial perspective, we consider it important to note that these conditions represent optimal methodological scenarios rather than requirements established in the literature or in standard guidelines for empirical research in the social sciences. Applying them as acceptance criteria would entail substantially redefining the study design retrospectively, which does not fall within the usual scope of the peer review process. Furthermore, these conditions do not necessarily reflect widespread empirical practice in studies analysing behaviour in real digital environments.

The modifications introduced in this review significantly enhance transparency, traceability, and the critical evaluation of the use of AI in the study.

Thus, the article incorporates the reviewer’s comments constructively without compromising the integrity of the study design or the core of the research. We are confident that this version will allow for a balanced assessment of the rigour, validity, and relevance of the work within the field of communications and big data. Thank you.