The Role of Artificial Intelligence in Gaming
Round 1
Reviewer 1 Report (Previous Reviewer 2)
Comments and Suggestions for AuthorsAfter a thorough review, it is evident that the author has invested significant effort in revising the manuscript. The revisions effectively address all the previous comments and concerns, substantially improving the clarity, logic, and academic rigor of the paper.
Minor comment: line 96 says 2005 and 2025; line 17 says 2016 and 2025. Please align the date.
Author Response
We sincerely thank the reviewer for the positive assessment of the revised manuscript.
Regarding the minor comment, we appreciate the reviewer for noting the inconsistency in the date range. We have corrected and aligned the stated period throughout the manuscript. The appropriate range is now consistently reported as 2016–2025 in all relevant sections, including the sentence originally located at line 96.
Reviewer 2 Report (Previous Reviewer 1)
Comments and Suggestions for AuthorsI'm happy to finally see that the authors took into account my previous suggestions and resubmitted a significantly improved version of the paper: I believe the expanded keyword list should now cover well over 99% of relevant publications and so is the fact that they now also consider conference papers as well. The tables now have results (e.g. one example of many: institutions who publish in this area) which make much more sense for anyone who is actively working in the field and knows the authors and institutions which produce high quality and quantity of publications.
I also think the analysis, especially in Section 4 of the paper (Discussion section) actually now has more relevance to guide someone who is interested in the trends of this important topic.
My main comment for possible revision: as Section 4 is by far the most important and interesting section (for me at least) consider giving a bit more intuition into the trends. E.g. in figure 9 the analysis is a bit confusing in some aspects, please be more specific about the trends, and I'm not entirely convinced that you explain clearly the four trends you mention as evidenced by the papers you found in your analysis.
Author Response
We sincerely thank the reviewer for the positive and constructive assessment of the revised manuscript and for acknowledging the substantial improvements introduced. We appreciate the reviewer’s observation regarding the importance of Section 4 and the need for greater clarity and intuition in the interpretation of the emerging trends.
In response to this valuable comment, we have expanded and refined Section 4.3, with particular focus on enhancing the explanation of the thematic trends derived from the strategic diagram (Figure 9).
Reviewer 3 Report (New Reviewer)
Comments and Suggestions for AuthorsThis study provided a comprehensive bibliometric analysis of 5114 documents on 536 Artificial Intelligence (AI) in gaming. The references cited in this paper essentially cover key and representative research achievements in the field. The study holds certain academic significance for enhancing people's understanding of the current applications and prospects of artificial intelligence in gaming. Overall, the research content is relatively complete, but there are still several issues that require further revision and refinement.
1. The content of this review paper primarily lists and presents the distribution of research findings, but lacks detailed introduction to the specific applications of artificial intelligence in gaming.
2. It is recommended to conduct a comparative analysis of the advantages and disadvantages of different methods.
3. Is the content in line 193 a figure or a table?
4. In Section 4.1, the references cited in the discussion of the four thematic areas are generally outdated and do not represent the latest research progress. It is advised to update the literature.
5. The conclusion section should more clearly outline future directions for the application of artificial intelligence in gaming.
The English could be improved to more clearly express the research.
Author Response
Dear reviewer:
Thank you very much for taking the time to review our manuscript again and for your assessment of it. We appreciate your suggestions and comments. In response, we will provide a targeted answer, and the corresponding revision will be reflected in the manuscript highlighted in yellow. Thank you for reading and reviewing our work.
1. Reviewer Comment: The content of this review paper primarily lists and presents the distribution of research findings, but lacks detailed introduction to the specific applications of artificial intelligence in gaming.
Thank you for this observation. We have strengthened the manuscript by adding a paragraph describing key application domains of AI in gaming, including gameplay automation, player modeling, adaptive learning systems, content generation, and health-related serious games.
Changes in Manuscript (added in Section 1, end of the Introduction): “In addition to mapping research production, recent work highlights concrete applications of AI in gaming, including automated gameplay agents, adaptive difficulty systems, procedural content generation, personalized learning in educational games, affective computing for emotion-aware interaction, and diagnostic or therapeutic use of AI-assisted serious games."
2. Reviewer Comment: It is recommended to conduct a comparative analysis of the advantages and disadvantages of different methods.
We agree with the reviewer’s suggestion. A new paragraph has been added in Section 4 to compare strengths and limitations of the most relevant AI techniques (reinforcement learning, deep learning, machine learning classifiers, and procedural generation approaches).
Changes in Manuscript (added at the end of Section 4.2): “In this regard, a comparative perspective reveals that reinforcement learning excels in autonomous decision-making in complex environments but requires large computational resources; deep learning provides high representational capacity yet suffers from limited interpretability; classical machine learning methods remain efficient for player modeling but struggle with scalability; and procedural content generation offers adaptability but requires careful human oversight to ensure coherence. Understanding these strengths and limitations helps contextualize methodological choices across different game-related applications.”
3. Reviewer Comment: Is the content in line 193 a figure or a table?
We thank the reviewer for pointing out this ambiguity. The content referred to is a figure, and we have clarified this in the text.
4. Reviewer Comment: In Section 4.1, the references cited in the discussion of the four thematic areas are generally outdated and do not represent the latest research progress. It is advised to update the literature.
We thank the reviewer for this observation. We would like to clarify that Section 4.1 focuses specifically on high-impact thematic clusters, and therefore our discussion relies primarily on the most cited and foundational publications in the field. These works—although not always the most recent—represent the studies that have exerted the strongest and most enduring scientific influence, as confirmed by their citation metrics and centrality within the bibliometric network.
Highly cited papers often require several years to accumulate scientific impact. For this reason, bibliometric analyses commonly highlight publications that, while not very recent, constitute the intellectual pillars of a research domain. This justifies their inclusion in Section 4.1.
5. Reviewer Comment: The conclusion section should more clearly outline future directions for the application of artificial intelligence in gaming.
We have expanded the conclusion to explicitly highlight future directions, including explainable AI in gaming, hybrid learning architectures, ethical safeguards, and human-centered adaptive systems.
Changes in Manuscript (added to the end of the Conclusion): “Future research is expected to advance toward explainable and transparent AI for improving trust in game-based systems, hybrid learning architectures that combine symbolic and neural approaches, and human-centered adaptive engines capable of modeling affect, motivation, and long-term learning behaviors. Ethical considerations (privacy, fairness, and responsible data use) will become increasingly important as AI-driven gaming applications expand into education, healthcare, and professional training contexts.”
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper aims to track the “rapid growth of research” in using AI and related techniques in the domain of games and gaming in general, including recreational and serious applications.
While this paper follows an established methodology, PRISMA workflow, which is appropriate in the literature for such analysis, there a several serious issues with the actual analysis conducted by the authors.
The most important issue, is that the keyword query performed (see Figure 2) is incorrect. By demanding general terms such as Artificial Intelligence (AI) or, to a lesser extent, Machine Learning, most of the relevant papers are not discoverable. I tried to look at the keywords of well-known papers, e.g. from the University of Alberta, which has one of the premier groups in gaming AI research, and almost none of their papers include such general terms. The same holds true for several papers that I have personally read during this last year, e.g. research topics such as applying search techniques to games, or regarding AI and game design and in general the keywords are much more specific than just AI or ML; AI appears nowhere that I checked and rarely does ML appear as a keyword. Instead you should look for terms such as Monte Carlo Tree Search (MCTS), reinforcement learning, procedural content generation, or even more specific, such as minimax, etc. which are much more specific keywords traditionally used. This might explain why you have found only 60 relevant paper in IEEE ToG over 20 years, when in fact this journal publishes papers which deal with these specific topics that you tried to capture in your analysis.
Furthermore, you removed most papers you did discover because they were conference paper. I don’t understand this decision. Several computing and AI conferences have archival proceedings and thus considered on par with most equivalent journals. For example, IJCAI is an A* conference considered on par with the top AI journals. I know for a fact that many AI papers with algorithms and/or applications to games have been published there over the last 20 years. What is the reasoning for excluding such high-quality papers? Perhaps it might be sufficient to just look at the top conferences only, there are relevant lists published for AI conferences, or try to only exclude conference papers which also have an extended journal publication as well.
Because of these serious shortcomings, you seem to have missed most of the papers, especially those not relating to Machine Learning, that you should have included in your analysis. In understand that the topic you tried to tackle is immense, but perhaps you should have done a more thorough analysis or at least tried to analyze a subpart of it. As you seem to have found a lot of general papers that are addressed to a more general audience, and less technical papers overall, meaning papers presenting specific algorithms - at least this is what I can conclude from reading your paper and following the general steps of the analysis you document - perhaps you might have wanted to restrict yourselves to just a specific area of AI as applied to gaming, or to primarily non-technical papers only.
For example, I searched for a paper that is not included at your top citations “A Survey of Monte Carlo Tree Search Methods”, as I had read it a while back it’s a very well-known paper. It has at least 2000 citations currently so should be #4 in your list. But it’s not there because of your analysis failing to find most of the papers. This is in addition to the low number of ToG papers you mention. Or how few papers you found around 2005 when in fact dozens if not hundreds of such papers were published in top conferences and journals even back then.
I did not check extensively parts of your analysis, once I realized the key omissions in paper retrievals, but I also believe the way you use Lotka or Bradford's law is applicable to such a broad domain with tens of thousands of papers in the last 20 years. While I do agree that some authors are more prolific in this area, it’s so big, that I find it dubious that such a law hold true. In fact, if your analysis were thorough this kind of information might be indeed interesting to publish.
Note also some other inconsistencies I found. Check the numbers of published papers across different tables. For example, you mention in Table 2 that Greece has published 28 papers, but in Table 5, you have two Greek universities with 25+22=47>28 papers! Check all your facts please.
Finally, it might be useful to have a clear message of what you are trying to achieve with your paper. Could it be, for example, that Lotka law holds? Could it be you want to points out specific areas of research that are drawing the most attention in the AI and games literature? E.g. I personally have noticed a huge increase in papers using LLMs in gaming over the last couple of years.
Author Response
Please, find enclosed a detailed response.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper presents a comprehensive bibliometric analysis of AI in gaming, offering valuable insights into the field’s intellectual structure and thematic evolution. However, several critical areas require refinement to enhance its rigor, clarity, and impact.
(1) the manuscript refers key figures but fails to leverage them for interpretive depth. Figure 1’s "Foundational to Applied" continuum is described in text but lacks granularity. labeling sub-themes would strengthen its role as a visual anchor for the field’s scope. Additionally, Figure 5’s citation peak in 2015 is linked to Mnih et al.’s work, but the figure does not overlay citation data with landmark publication years, missing an opportunity to visually validate this causal claim. Figures should be designed to stand alone as explanatory tools rather than passive illustrations.
(2) in line 148, the authors removed conference papers. It is better to explain why conference papers are removed. It is well-known that several leading conferences are welcomed and have high impact in AI field.
(3) table 6 shows the impact of most cited documents in journals. the authors do not pay attention to the impact of journals. 1219 papers from different journals; while different journals have different impact. It is better to assign different weight to different journals. I am not sure if this could affect the conclusion of the manuscript.
Author Response
Please, find enclosed a detailed response.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI accept your comments regarding the survey paper and the inconsistency in the tables that I pointed out. I cannot fully check these, but they sound at least plausible.
However, I find it rather puzzling that you chose to just disagree with most of my recommendations rather than addressing them, and without giving not even a hint of an argument that your point of view is valid! In particular, I pointed out some very very serious issues, the two key ones I summarize here:
1. The most serious is that the keyword search you did does NOT cover all relevant articles. You need to search for more specific keywords than just AI and ML. The fact that you do it also in the abstract as you state in your response, does not suffice, as a LOT of publications do not put such incredibly generic terms anywhere, because they are not useful. In fact, you mostly get articles which are more generic and far fewer technical contributions, which are the most important ones that you need to consider.
2. Unlike in several other fields, in computing and in particular in AI and games, a lot of publications are done in conferences. So while in other disciplines, e.g. economics, that I know personally about, and lots of others, it is acceptable to not consider conference publications, here you are missing some (significant) part of the research output that your analysis is meant to cover. You can check this with the university faculty review procedures in several countries, e.g. UK.
Let me give you some evidence that support my point of view.
I checked the publications of just ONE researcher at the University of Alberta which has the most prolific games group in the world. Just from that one researcher I found 26 articles that fit your criteria (and that's just one researcher not the whole university). In addition, close to 100 conference papers, which some match the journal articles, but a lot of others are NOT covered by journal papers. These are published in top archival conference. Check the scholar page below to verify:
https://scholar.google.com/citations?hl=el&user=PYtPCHoAAAAJ&view_op=list_works&sortby=pubdate
The University of Malta also has a very strong AI and games group, (with dozens if not hundred of publications, both journal and conference papers) as well as other universities which I do not see in Table 5. This support the statements that I make in this review.
Therefore, you need to redo the entire search and analysis that you have conducted, in order for your results to be valid.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper has been improved. The paper foscuses on AI in gaming, while the meaning of gaming is general. It is better to delve into a detailed analysis of the gaming types such as education, healthcare, and human–computer interaction. Or it is better to delve into a detailed summary of video games, serious games, and so on. Otherwise, the title of the paper starts AI but lacks AI factors; the title of the paper ends with gaming but lacks specific games or gaming type analysis.
