Artificial Intelligence Alone Will Not Democratise Education: On Educational Inequality, Techno-Solutionism and Inclusive Tools

Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe introduction is interesting and comprehensive, clearly defining the purpose of the work and its importance, including specific hypotheses. The current state of the research field and relevant publications on the topic are cited. However, the article does not present scientifically sound experiments and does not provide a substantial amount of new information as considered by the journal.
The materials should be improved and a concrete methodology should be developed. There is no empirical analysis to replicate and develop the published results. The results should provide a concise and precise description of the experimental results, their interpretation and the experimental conclusions that can be drawn. Based on the theoretical framework presented, we encourage the authors to conduct empirical research on the impact of Artificial Intelligence in Education and its potential to build personalised and democratic curricula.
The article presented would be more appropriate in the category Reviews, which according to the journal, provide concise and precise updates on the latest progress made in a given area of research. Systematic reviews should follow the PRISMA.
Author Response
Many thanks for the positive and constructive comments about our work, especially for appreciating the timeliness and relevant work discussed in this opinion paper. We have addressed the points you have brought up below:
Point 1. The manuscript does not present scientifically sound experiments
Response 1: We thank you for pointing this. There was a major mishap in the initial submission where it was not clearly communicated that this is an “opinion” paper rather than a research article or a systematic review. While we communicated this to the editor, the latex manuscript was not changed to the right template to reflect this. We apologise for this miscommunication and have now changed the template accordingly.
Point 2. Lack of a concrete methodology or empirical results that can be replicated. In the context of a review, a systematic methodology (such as PRISMA) is not described.
Response 2: We would like to refer here that this is an opinion piece rather than a research article or a systematic review. We apologise for the confusion. To improve clarity, we have added text in the abstract (line 10), section 1.3 (line 115) to clarify that this is an opinion piece. We further state in section 1.3 (lines 116-119) that the work is based on a critical scoping of the available evidence and not a systematic review.
Reviewer 2 Report
Comments and Suggestions for Authors“Techno-solutionism" is a term that is placed in the title as one of the central descriptors of the study, however, no conceptual clarification of it is presented throughout the text. It would be important to clarify this.
The authors should state the methodology of the study both in the abstract and in the text, even if it is a literature review; it is not clear which procedures were followed in the study, the databases that prevailed, the search criteria and systematization. It is important to state transparently the route followed by the study.
The objective of the study should be made more explicit and the conclusions should reflect the extent to which the proposed objective was or was not met.
It is recommended to use the full term "Artificial Intelligence" in the title of the article, in Scopus around 57,220 documents have the term "Artificial Intelligence" in the title, and a low percentage have only the abbreviations "AI", this will help to better locate and use this study in the different databases.
The authors propose 4 technological pillars:
1. Open Educational Resources:
2. A Unifying Taxonomy of Knowledge:
3. Human-centred AI:
4. Streamlined and Solidified Regulation:
In the second pillar they bet on Wikipedia, however, they do not expose the limitations of this encyclopedia, the current challenges in the use of this, the challenges with their input sources that are not always scientific, in this sense, it would be important, especially for quality education, to clarify these aspects and to emphasize scientific databases, which do have established scientific knowledge.
The solution to the technological, social, and cultural differences between nations, and the commitment to quality education for all, depends first and foremost on universal and equitable access to quality scientific information.
A section on methodology is missing; making transparent the route followed by the study may help to better understand the RESULTS and DISCUSSION sections.
Twenty-six self-citations related to authors have been identified (2, 8, 16, 23, 27, 30, 31, 32, 32, 33, 36, 37, 43, 50, 69, 70, 87, 94, 95, 96, 97, 99, 100, 102, 103, 108, 111), this is too many self-citations, it would be advisable to reduce these self-citations to less than 8.
Regarding references, it is recommended that the authors use references that are closer to their topic of study.
Comments for author File: Comments.pdf
Author Response
Many thanks for the positive and constructive comments about our work. We have addressed the points you have brought up below:
Point 1. “Techno-solutionism" should be defined as it is a critical part of the work
Response 1: We thank the reviewer for this suggestion, an explanation of this term was indeed missing in the paper. We have now added a clarification in the abstract and a longer explanation in section 3.3.2.
Point 2. A concrete methodology is not described in the abstract or in the text even in the context of a systematic review.
Response 2: We thank you for pointing this. There was a major mishap in the initial submission where it was not clearly communicated that this is an “opinion” paper rather than a research article or a systematic review. While we communicated this to the editor, the latex manuscript was not changed to the right template to reflect this. We apologise for this miscommunication and have now changed the template accordingly. Furthermore, to improve clarity, we have added text in the abstract (line 10), section 1.3 (line 115) to clarify that this is an opinion piece. We further state in section 1.3 (lines 116-119) that the work is based on a critical scoping of the available evidence and not a systematic review.
Point 3. The objective of the work should be explicitly presented and referred to where necessary.
Response 3: We thank you for pointing this out. In response, we have improved clarity by explicitly stating the goal of the work in section 1.3, motivation (lines 116-135). To further clarify, we start the section with “our primary goal is to identify several key pillars that would allow for building and maintaining a sustainable, large-scale and inclusive AI in Education ecosystem that facilitates equitable, high-quality lifelong learning opportunities for all.” (lines 116-119) that makes this goal explicit. The remainder of the section expands how we achieve this goal in this work.
Point 4. Use of full term Artificial Intelligence rather than AI in the title
Response 4: We fully agree with the proposal as our objective is also to make this work as discoverable as possible. We have made the change to the title.
Point 5. Lack of scientific backing for the case of Wikipedia as a knowledge base given its limitations.
Response 5: With discussion among the co-authors and doing a literature survey, we have concluded that there is ample evidence of the reliability of Wikipedia as a faithful knowledge source. We identify the key issue in this manuscript is that these works have not been cited to support this case. We have added additional text to section 3.2 to actively present these arguments with the citations to support the claims. We have also reduced the emphasis on Wikipedia itself and added a few more generic mentions to collaborative Encyclopedias.
Point 6. Emphasising the need for universal and equitable access to quality scientific information
Response 6: We fully agree with this claim and all the pillars proposed in this opinion paper directs to universal and equitable scientific information. OERs are openly licensed materials that can be consumed with minimal copyright restrictions. Wikipedia is an open knowledgebase with both direct and API based programmatic access to the wealth of multi-lingual information. Human-centric and open tools drive improved access to both knowhow and model/software artefacts. Regulation will enrich the ethical soundness and equity. Furthermore, we have added refinforcement of this fact as “We think that access through good connectivity... important parts of providing quality education for all (lines 272-275).
Point 7. Missing methodology and results sections
Response 7: This goes back to the previous point that this is an opinion piece rather than a research article or a systematic review. We apologise for the confusion. To improve clarity, we have added text in the abstract (line 10), section 1.2 (line 115) to clarify that this is an opinion piece. We further state in section 1.3 (lines 116-119) that the work is based on a critical scoping of the available evidence and not a systematic review.
Point 8. The number of self-citations seems excessive and can be limited to self references closer to the topic
Response 8: We agree that there are many self-citations in the manuscript and apologise for this. This is due partly to the fact that this work builds heavily on various directions of research work that has been conducted for over 10+ years. The opinion piece, due to its objective to identify a Techno-social solution, covers diverse research domains (Computer Science, Artificial Intelligence, Education, Policy and so on...) where the authors also come from different domains. While we have significantly reduced the number of self-citations, due to this, we strongly believe that several of the cited papers are essential for us to communicate a clear picture of this work and its context within the fie With the recommendation of the assistant editor, we have reduced the number of citations significantly where 13 citations have been removed/replaced.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe contribution is certainly of interest, timeliness and relevance. However, there are some limitations that need to be resolved:
- disambiguate and contextualize from the beginning what is meant by the key concepts of the contribution, such as AI, Inclusion, TEL, which, as we know, take on various contextual meanings, as it happens in the text presented. It is necessary to clarify to the reader how they are used
- a restructuring of the paragraphs is suggested to show the logic of the discussion more clearly
- the authors focus on the role of OER but without sufficiently developing the relationship with AI or, at least, with the type of AI they intend
- the text seems to indicate different objectives each time, declared at the end or within the individual paragraphs, giving the sensation of never understanding what point the authors want to reach/demonstrate (e.g. lines 62-70, 144-145, 156-160, 330-343, 628-637). Once it has been clarified what the real point of the work is, it must be said in the abstrcat, introduction and resumed eacht time till it is neeeded till the conclusions
- topics are introduced that are not then adequately taken up (e.g. blended learning line 69; 157-158)
- The only figure inserted does not seem relevant or that it adds value, moreover it does not fully correspond to the theses subsequently developed (i.e., the four pillars of par. 3.1.). The authors, instead, could make greater use of diagrams and figures to support their theses more clearly
- some passages are very dense and full of implications to disambiguate and clarify (lines 218-222)
- some passages are unnecessarily redundant (e.g. 241-243, 257-263 and similar)
- what happens to the 4 pillars of 3.1 in the conclusions? Moreover, in the conclusions, I would expect a final summary table that summarizes the discussion and is accompanied by practical indications, limits and solutions, rather than a repetition of things already said and a new list of points to address, which, however, should be presented in a more graphically appealing way
Author Response
Many thanks for the positive and constructive comments about our work, especially for appreciating the timeliness and relevant work discussed in this opinion paper. We have addressed the points you have brought up below:
Point 1. Disambiguate and contextualize from the beginning what is meant by the key concepts of the contribution, such as AI, Inclusion and TEL
Response 1: We thank you for pointing this. We have added additional text to define the scope of these key topics to improve clarity. AI and Educational Technology are scoped in section 1.4 (lines 140-145 and lines 154-156 respectively). The scope of inclusion is set in section 1.2 lines 111-112.
Point 2. A restructuring of the paragraphs is suggested to show the logic of the discussion more clearly
Response 2: We thank you for recommending this. We have discussed this among the co-authors and have done some restructuring. In the new revision, we start with an introduction where the reinforcement of exclusion through AI is pointed out (sec. 1.2), followed by the motivation inspired by the former effect (sec. 1.3) and summed up by stating the role Educational Technology and AI can play in this democratisation process (sec. 1.4). The promise and the peril of AI in Education is described next (Sec. 2). Following this background, the 4 pillars are proposed (sec.3) with a subsection further ellaborating each pillar (sec 3.1-3.4) where the human-centric AI tools that can take multiple facets are further expanded (sec. 3.3.1-3.3.3). Based on the background presented in sections 2 and 3, the conclusions are presented in sec. 4. A new subsection has been added to the introduction subtitled “paper overview” (sec. 1.1, lines 64-71) where this flow of information is clearly described in the paper itself.
Point 3. The relationship between AI and OER needs to be clearly developed.
Response 3: We have modified the text in several places. Specifically, in section 2.1, Promise, lines 177-183, describing how 100,000s of materials are discoverable with personalisation using AI technologies such as search and recommender systems. Also, we modified section 3.1, OERs, lines 416-420 and lines 423-426 to describe how automatic translation, transcription, quality prediction, information retrieval and personalisation can improve the utility of OER justifying the power of the pillar.
Point 4. The manuscript points out different objectives in different parts of the paper. This needs to be addressed.
Response 4: We thank you for pointing this out. In response, we have improved clarity by explicitly stating the goal of the work in section 1.3, motivation (lines 116-135). To further clarify, we start the section with “our primary goal is to identify several key pillars that would allow for building and maintaining a sustainable, large-scale and inclusive AI in Education ecosystem that facilitates equitable, high-quality lifelong learning opportunities for all.” (lines 116-119) that makes this goal explicit. Then we remind the reader of this goal in the abstract (lines 10-12), section 1.4, section 3 (lines 358-360) and the conclusion (section 4, lines 674-675) as it improves readability.
Point 5. Topics are introduced that are not then adequately taken up (e.g. blended learning line 69; 157-158)
Response 5: The topic of blended learning has now been introduced in the paragraph where it was mentioned. We have also gone through the paper and clarified a couple of other concepts that were not introduced at the right time. We thank the reviewer for noticing this.
Point 6. The only figure inserted does not seem relevant or that it adds value, moreover it does not fully correspond to the theses subsequently developed (i.e., the four pillars of par. 3.1.). The authors, instead, could make greater use of diagrams and figures to support their theses more clearly
Response 6: We thank you for the recommendation. With this recommendation, we have modified and improved the figure to show how the pillars address several key challenges in democratising education and repositioned it in the conclusion where it provides a holistic, graphical summary of the information in the opinion piece. We have discussed this among the co-authors and feel that the figure allows the reader to get a higher level of understanding and information about the four pillars presented in this opinion piece. A strong reason to retain this figure is that the paper has a lot of information and the figure can represent the summary very effectively in the conclusion section.
Point 7. Some passages need are redundant (e.g. 241-243, 257-263 and similar) and needs clarity (lines 218-222)
Response 7: We have carefully analysed the parts pointed and come to a conclusion that while similar, the pointed extracts in the manuscript discuss different issues (new lines 270-272 discuss infrastructure such as connectivity, devices, servers etc. while new lines 288-292 discuss biases due to the AI models trained for popular languages that worsen the challenge for low resource languages). We have rewritten lines 246-251 to improve clarity.
Point 8. Summarise the conclusion in terms of the 4 pillars in the final section with a graphically apealing manner
Response 8: The four pillars have now been introduced in the conclusions (sec. 4 lines 658-664 and lines 678-688). We have also restructured some of the material and paragraphs to make it more easily readable. Additionally, we have introduced a figure that summarises how the four pillars are connected to addressing some of the key challenges in building a sustainable AIEd ecosystem (figure 1)
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsCongratulations to the authors for the new version of the article and for clarifying and resolving previous comments.
Please take into account the number of self-citations in the next revision.
Author Response
Many thanks for the positive and constructive comments about our work. We have addressed the points you have brought up below:
Point 1. The number of self-citations need to be reduced
Response 1: We thank the reviewer for this suggestion. While we reiterate the challenge of reducing the citations due to the multi-faceted nature of the paper, we have further removed 4 self-citations bringing the number of self-citations to 12 (out of 120) in line with the editor’s recommendation of keeping it at 10%.
Reviewer 2 Report
Comments and Suggestions for AuthorsIn this new version of the study, the authors have been addressing the comments made, which has resulted in a more complete text, however, depending on the final version of the text, there are still three recommendations that have not been resolved:
- Recommendation 1.- Number of self-citations
- Recommendation 2.- Theoretical framework of "Techno-solutionism".
- Recommendation 3.- Complete the methodology.
The details are shown below:
- Recommendation 1.- Number of self-citations
The number of citations related to authors has decreased considerably compared to the initial version, however, there are still 15 self-citations: 2, 5, 6, 7, 24, 25, 25, 32, 35, 36, 41, 92, 93, 95, 97, 98. It would be appropriate for these self-citations to be below 8.
- Recommendation 2.- Theoretical framework of "Techno-solutionism".
The authors have added a concept on "Techno-solutionism" between lines 603-607, however, it seems more of a personal perception as it lacks any scientific citation.
Since "Techno-solutionism" is one of the main categories of the study, it should have theoretical support. There are several studies that address this category as a central aspect and that can help researchers to build a theoretical basis in this regard.
- Recommendation 3.- Complete the methodology.
Important note: consider providing more details in the methodology to allow better reproducibility of the study, e.g., descriptors used, search equation, databases examined, study selection criteria, time period, etc. Anything that can clarify the route followed by the authors. This is very important to have an idea of how the research proceeded.
Good luck in this final stage.
Comments for author File: Comments.pdf
Author Response
Many thanks for the positive and constructive comments about our work. We have addressed the points you have brought up below:
Point 1. Number of self-citations
Response 1: We thank the reviewer for this suggestion. While we reiterate the challenge of reducing the citations due to the multi-faceted nature of the paper, we have further removed 4 self-citations bringing the number of self-citations to 12 (out of 120) in line with the editor’s recommendation of keeping it at 10%.
Point 2. Theoretical framework of "Techno-solutionism".
Response 2: We thank you for pointing out this potential improvement. We have further elaborated the definition of Techno-solutionism in section 3.3.2 lines 607-614 and supported our description with 3 additional references (110-112) to elaborate the definition.
Point 3. Complete the methodology
Response 3: We appreciate this recommendation. This work is an “opinion” paper rather than a research article or a systematic review. We communicated this clearly via our last revision by 1) changing the template accordingly and explicitly and clearly stating this in several relevant places in the paper (abstract and section 1.3). We have further improved clarity by describing the method used to compile our opinion in section 1.3 lines 124-128. We hope this elaboration provides further clarity on this matter.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript is much improved
Comments on the Quality of English LanguageEnglish requires revisions
Author Response
Many thanks for the positive and constructive comments about our work. We have addressed the points you have brought up below:
Point 1. English requires revision
Response 1: We thank the reviewer for this suggestion. We have discussed and have had the native English speakers among the authors revise the language quality of the paper in this revision.