Rethinking the Relation between Media and Their Audience: The Discursive Construction of the Risk of Artificial Intelligence in the Press of Belgium, France, Portugal, and Spain
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe article analyzes how news platforms in four EU countries frame AI risks, loosely defined as "potentially negative and harmful impacts of AI development and implementation" (p. 1). The paper addresses a topical and extremely popular subject. I commend the authors for choosing Romance language news outlets. Conceptually, the paper pairs audience monitoring/metrics and discursive practices. In its present form, I don't think the paper works just yet. In what follows, I indicate where I feel the paper needs to be enhanced and I offer suggestions for doing so.
1. Theoretical justification: why this study now?
Given the popularity of the subject, I feel the authors should engage more with recent work on AI media discourse (eg. Cools et al. 2024, Ananny 2024 or Marres et al. 2024) to carve out a niche for their study and position their study vis-à-vis relevant literature. At present, I feel it's too self-contained. The public service ideal of journalism is foregrounded in the section 3 but that's low-hanging fruit. What about sociotechnical imaginaries? Or journalistic metadiscourse (Carlson 2015)? Or AI controversies?
2. Analytical transparency: IR scores
A content analysis conducted by two scholars implies a degree of interrater reliability. This is omitted in the paper. How exactly were 'doubts that arose during the process' (p. 5) resolved? Quantifying the coding process (beyond percentages) would strenghten the paper considerably.
3. Interpretive leads: what about boundary work?
Research shows that whenever their authority is challenged, journalists tend to engage in boundary work (Gieryn 1983) to reaffirm the solidity of their profession's norms and procedures. It's difficult to ignore this observation when reading that 63% of the total risk occurrences have to do with dimensions that speak to journalistic labor and trustworthiness (eg. public opinion manipulation, content misappropriation, etc). The empirical examples on p. 6 illustrate boundary work rhetoric -- and they explain (in part) why journalists are so eager to write about emerging tech in dystopian terms (see also Singer 2005).
4. Intercountry comparisons: consider deleting or unpacking
I don't think section 5.2 is neccesary. Relative percentages don't speak to me and differences in risk categories are only meaningful when seen through a particular theoretical lens (see point 1 above). I think you could craft a more convincing argument about how news media cover AI by deleting the country specific comparison.
5. conclusions: remove speculative elements
The conclusion is too speculative for my taste. Rather than tell the reader that global dimensions in AI risk coverage are 'interesting', I want to learn what layer of insight these dimensions provide. Again, I feel that a clearer theoretical orientation is necessary, above and beyond the programmatic claim about media and their audiences. Finally, that "national contexts have a notable influence on" (p. 12) risk coverage does not follow logically from the preceding analysis.
I hope the authors are open to my comments and are willing to revise their work so that it can be published.
Works cited
Ananny, M. (2024). Making Generative Artificial Intelligence a Public Problem. Seeing Publics and Sociotechnical Problem-Making in Three Scenes of AI Failure. Javnost - The Public, 31(1), 89–105. https://doi.org/10.1080/13183222.2024.2319000
Carlson, M. (2015). Metajournalistic Discourse and the Meanings of Journalism: definitional control, boundary work, and legitimation. Communication Theory, 26(4), 349–368. https://doi.org/10.1111/comt.12088
Cools, H., Van Gorp, B., & Opgenhaffen, M. (2024). Where exactly between utopia and dystopia? A framing analysis of AI and automation in US newspapers. Journalism, 25(1), 3-21. https://doi.org/10.1177/14648849221122647
Gieryn, T. F. (1983). Boundary-Work and the Demarcation of Science from Non-Science: Strains and Interests in Professional Ideologies of Scientists. American Sociological Review, 48(6), 781-795.
Marres, N., Castelle, M., Gobbo, B., Poletti, C., & Tripp, J. (2024). AI as super-controversy: Eliciting AI and society controversies with an extended expert community in the UK. Big Data & Society, 11(2). https://doi.org/10.1177/20539517241255103
Singer, J. B. (2005). The political j-blogger: 'Normalizing' a new media form to fit old norms and practices. Journalism, 6(2), 173-198. https://journals.sagepub.com/doi/10.1177/1464884905051009
Comments for author File: Comments.pdf
I've made a few editorial suggestions in the attached version of the paper. These have to do with word choice, register and style. For instance, "the beginning of the third decade of the millenium" is difficult to read. How about 'the current decade'? Moreover, some turns of phrases were hard to grasp. What for instance, is the reader to make of 'content offer and approaches presented on a daily basis'? (p. 2).
Author Response
Dear reviewer,
First of all, we would like to thank you for your service to this article. Most certainly, your comments have helped us broaden the scope of our work and facilitated the improvement of its quality, depth and cohesiveness.
We have carefully read the feedback given by each reviewer. In essence, we have added part of the bibliography suggested, rewritten various parts of our article, the ones highlighted in yellow, and rewritten almost completely the discussion and results.
In the same spirit, the article has been reviewed by a native speaker.
Comment 1:
Theoretical justification: why this study now?
Given the popularity of the subject, I feel the authors should engage more with recent work on AI media discourse (eg. Cools et al. 2024, Ananny 2024 or Marres et al. 2024) to carve out a niche for their study and position their study vis-à-vis relevant literature. At present, I feel it's too self-contained. The public service ideal of journalism is foregrounded in the section 3 but that's low-hanging fruit. What about sociotechnical imaginaries? Or journalistic metadiscourse (Carlson 2015)? Or AI controversies?
Response 1:
In order to give the research niche a more solid structure we added the authors suggested by the reviewers (Cools et al. 2024; Annany, 2024 and Marres et al. 2024). In addition, we also included other authors in hopes of broadening the scope of the study of risk discourses and AI implementation in journalism (Cave et al. 2020 and Sortori & Bocca 2023). In the same spirit, we tried to reinforce the niche proposed in our article. Our research comes from the general interest to better understand the relation between audiences and the differences present in how these global topics are treated in local/national contexts.
We have incorporated the notions of socio-technical imaginary (Sartori and Bocca 2023) and the study of controversies (Mare et al. 2024) in the discussion.
Certainly, with the same textual corpus, research focused on these concepts could be carried out, since they address serious and thorough aspects of the language and representations of technology in the press and in society. Clearly, the suggested bibliography is also a contribution to our work beyond this article.
Comment 2
Analytical transparency: IR scores
A content analysis conducted by two scholars implies a degree of interrater reliability. This is omitted in the paper. How exactly were 'doubts that arose during the process' (p. 5) resolved? Quantifying the coding process (beyond percentages) would strenghten the paper considerably.
Response 2:
We have stated what we have done procedurally. We conducted a parallel analysis and solved doubts by discussing them. It is a reflective and thorough approach of scientific collaboration between two researchers with different research trajectories and from different backgrounds: one a linguist and the other a journalist.
Comment 3:
Interpretive leads: what about boundary work?
Research shows that whenever their authority is challenged, journalists tend to engage in boundary work (Gieryn 1983) to reaffirm the solidity of their profession's norms and procedures. It's difficult to ignore this observation when reading that 63% of the total risk occurrences have to do with dimensions that speak to journalistic labor and trustworthiness (eg. public opinion manipulation, content misappropriation, etc). The empirical examples on p. 6 illustrate boundary work rhetoric -- and they explain (in part) why journalists are so eager to write about emerging tech in dystopian terms (see also Singer 2005).
Response 3:
Gieryn's (1983) suggested work addresses the concept of "Boundary-Work" which refers to the activities by which scientists create and maintain a boundary between science and non-science. This process is pivotal to acquire intellectual authority, career opportunities and protect the autonomy of scientific research from political and pseudoscientific interference. Reading this work has led us to recognize other boundaries that are managed in the public space, such as what is and what is not of public interest. This reflection is complemented by the work of Ananny (2024) that investigates the role of AI in the redefinition of the public.
Comment 4
Intercountry comparisons: consider deleting or unpacking
I don't think section 5.2 is neccesary. Relative percentages don't speak to me and differences in risk categories are only meaningful when seen through a particular theoretical lens (see point 1 above). I think you could craft a more convincing argument about how news media cover AI by deleting the country specific comparison.
Response 4:
We have decided to keep this section and reinforce its purpose in order to try to achieve a better assessment of the subject. We believe that the comparison of relative values for each newspaper is a valid way of showing the different orientations or interests of the media regarding the risks of AI. This topic, which is treated descriptively in the results section, is then taken up again in the discussion.
Comment 5:
conclusions: remove speculative elements
The conclusion is too speculative for my taste. Rather than tell the reader that global dimensions in AI risk coverage are 'interesting', I want to learn what layer of insight these dimensions provide. Again, I feel that a clearer theoretical orientation is necessary, above and beyond the programmatic claim about media and their audiences. Finally, that "national contexts have a notable influence on" (p. 12) risk coverage does not follow logically from the preceding analysis.
Response 5:
The discussion and conclusions sections have received the vast majority of reformulation changes.
With the literature we have incorporated, we believe that the interpretation of the research findings has been enriched. We incorporate Marres (2024) with the notion of controversy, to which we have related our findings. We incorporated Ananny (2024) with the notion of Dual AI Identity (ontological and epistemological) which we have also related to our findings. We have also restructured the sections to make more explicit what we believe to be the contribution of our research. While these results do not directly reflect public awareness of risk in each of the countries, they open a line of research on the possible influences of the progressive monitoring and understanding of audiences on the construction of the media agenda.
Comments on the quality of the English employed
I've made a few editorial suggestions in the attached version of the paper. These have to do with word choice, register and style. For instance, "the beginning of the third decade of the millenium" is difficult to read. How about 'the current decade'? Moreover, some turns of phrases were hard to grasp. What for instance, is the reader to make of 'content offer and approaches presented on a daily basis'? (p. 2).
Response
These excerpts have been rewritten and the text has been reviewed by a native English speaker.
References Included
Ananny, M. (2024). Making Generative Artificial Intelligence a Public Problem. Seeing Publics and Sociotechnical Problem-Making in Three Scenes of AI Failure. Javnost - The Public, 31(1), 89–105. https://doi.org/10.1080/13183222.2024.2319000
Carlson, M. (2016). Metajournalistic Discourse and the Meanings of Journalism: Definitional Control, Boundary Work, and Legitimation. Communication Theory, 26(4), 349–368. https://doi.org/10.1111/comt.12088
Cave, S., Dihal, K., & Dillon, S. (Eds.). (2020). AI narratives: A history of imaginative thinking about intelligent machines. Oxford University Press.
Cools, H., Van Gorp, B., & Opgenhaffen, M. (2024). Where exactly between utopia and dystopia? A framing analysis of AI and automation in US newspapers. Journalism, 25(1), 3-21. https://doi.org/10.1177/14648849221122647
Marres, N., Castelle, M., Gobbo, B., Poletti, C., & Tripp, J. (2024). AI as super-controversy: Eliciting AI and society controversies with an extended expert community in the UK. Big Data & Society, 11(2). https://doi.org/10.1177/20539517241
Sartori, L., & Bocca, G. (2023). Minding the gap(s): Public perceptions of AI and socio-technical imaginaries. AI & Society, 38(3), 443–458. https://doi.org/10.1007/s00146-022-01422-1
Gieryn, Thomas F. (1983) Boundary-Work and the Demarcation of Science from Non-Science: Strains and Interests in Professional Ideologies of Scientists. American Sociological Review, Vol. 48, No. 6. (Dec., 1983), pp. 781-795.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe article proposes interesting research and does so in a scholarly appropriate manner. The main limitation I perceive is the imbalance between the space given to background, data collection and its criteria and the actual qualitative part, which remains too descriptive. The content analysis (I would not call it textual analysis) of the main risks perceived by audiences highlights very general "threats" that are not fully discussed. Some of these threats, such as AI 'hallucinations,' are not even explained. Additionally, a few of the entries could be aggregated. A better fine-tuning would help make the analysis more insightful. Adding literature on content analysis of news discourse could further enhance the paper.
Comments for author File: Comments.pdf
A thorough language check is advisable.
Is syllabification correctly set on English?
Author Response
Dear reviewer,
First of all, we would like to thank you for your service to this article. Most certainly, your comments have helped us broaden the scope of our work and facilitated the improvement of its quality, depth and cohesiveness.
We have carefully read the feedback given by each reviewer. In essence, we have added part of the bibliography suggested, rewritten various parts of our article, the ones highlighted in yellow, and rewritten almost completely the discussion and results.
In the same spirit, the article has been reviewed by a native speaker.
Comment 1
The article proposes interesting research and does so in a scholarly appropriate manner. The main limitation I perceive is the imbalance between the space given to background, data collection and its criteria and the actual qualitative part, which remains too descriptive.
Response 1
We have worked on rewriting the discussion and conclusion to give more value to the findings.
Comment 2
The content analysis (I would not call it textual analysis) of the main risks perceived by audiences highlights very general "threats" that are not fully discussed.
Response 2
Certainly, in this case, the work of textual analysis from discourse analysis is very close to content analysis. We call it textual analysis because it is what we do from the perspective of discourse analysis, taking into account the contexts of production.
Comment 3
Some of these threats, such as AI 'hallucinations,' are not even explained. Additionally, a few of the entries could be aggregated. A better fine-tuning would help make the analysis more insightful.
Response 3
As there are fourteen categories, we have only described the first four, which together correspond to 63% of the occurrences. Since these topics have been widely discussed in the press, we considered it was easy to access their meaning. What we provide is precisely the differences in occurrences between countries.
References Included
Ananny, M. (2024). Making Generative Artificial Intelligence a Public Problem. Seeing Publics and Sociotechnical Problem-Making in Three Scenes of AI Failure. Javnost - The Public, 31(1), 89–105. https://doi.org/10.1080/13183222.2024.2319000
Carlson, M. (2016). Metajournalistic Discourse and the Meanings of Journalism: Definitional Control, Boundary Work, and Legitimation. Communication Theory, 26(4), 349–368. https://doi.org/10.1111/comt.12088
Cave, S., Dihal, K., & Dillon, S. (Eds.). (2020). AI narratives: A history of imaginative thinking about intelligent machines. Oxford University Press.
Cools, H., Van Gorp, B., & Opgenhaffen, M. (2024). Where exactly between utopia and dystopia? A framing analysis of AI and automation in US newspapers. Journalism, 25(1), 3-21. https://doi.org/10.1177/14648849221122647
Marres, N., Castelle, M., Gobbo, B., Poletti, C., & Tripp, J. (2024). AI as super-controversy: Eliciting AI and society controversies with an extended expert community in the UK. Big Data & Society, 11(2). https://doi.org/10.1177/20539517241
Sartori, L., & Bocca, G. (2023). Minding the gap(s): Public perceptions of AI and socio-technical imaginaries. AI & Society, 38(3), 443–458. https://doi.org/10.1007/s00146-022-01422-1
Gieryn, Thomas F. (1983) Boundary-Work and the Demarcation of Science from Non-Science: Strains and Interests in Professional Ideologies of Scientists. American Sociological Review, Vol. 48, No. 6. (Dec., 1983), pp. 781-795.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe article is ok for publication. nothing dramatic, but needs a linguistic revision.
Comments on the Quality of English Languagenothing dramatic, but needs a linguistic revision.
Author Response
Dear reviewer,
First of all, we would like to thank you for your service to this article. Most certainly, your comments have helped us broaden the scope of our work and facilitated the improvement of its quality, depth and cohesiveness.
We have carefully read the feedback given by each reviewer. In essence, we have added part of the bibliography suggested, rewritten various parts of our article, the ones highlighted in yellow, and rewritten almost completely the discussion and results.
In the same spirit, the article has been reviewed by a native speaker.
References Included
Ananny, M. (2024). Making Generative Artificial Intelligence a Public Problem. Seeing Publics and Sociotechnical Problem-Making in Three Scenes of AI Failure. Javnost - The Public, 31(1), 89–105. https://doi.org/10.1080/13183222.2024.2319000
Carlson, M. (2016). Metajournalistic Discourse and the Meanings of Journalism: Definitional Control, Boundary Work, and Legitimation. Communication Theory, 26(4), 349–368. https://doi.org/10.1111/comt.12088
Cave, S., Dihal, K., & Dillon, S. (Eds.). (2020). AI narratives: A history of imaginative thinking about intelligent machines. Oxford University Press.
Cools, H., Van Gorp, B., & Opgenhaffen, M. (2024). Where exactly between utopia and dystopia? A framing analysis of AI and automation in US newspapers. Journalism, 25(1), 3-21. https://doi.org/10.1177/14648849221122647
Marres, N., Castelle, M., Gobbo, B., Poletti, C., & Tripp, J. (2024). AI as super-controversy: Eliciting AI and society controversies with an extended expert community in the UK. Big Data & Society, 11(2). https://doi.org/10.1177/20539517241
Sartori, L., & Bocca, G. (2023). Minding the gap(s): Public perceptions of AI and socio-technical imaginaries. AI & Society, 38(3), 443–458. https://doi.org/10.1007/s00146-022-01422-1
Gieryn, Thomas F. (1983) Boundary-Work and the Demarcation of Science from Non-Science: Strains and Interests in Professional Ideologies of Scientists. American Sociological Review, Vol. 48, No. 6. (Dec., 1983), pp. 781-795.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsDear authors,
Many thanks for your thoughtful responses and pertinent edits, which have enhanced the paper substantially. I feel that my concerns have been addressed and recommend publication.
Sincerely,
Tom