Generative AI in Game Development: A Qualitative Research Synthesis
Open accessThe DOI matches its Crossref record; 57 of the 57 references it cites were found in a registry or an index; 26 of the 104 sources its claims cite were checked for support.
Abstract
Generative Artificial Intelligence (GenAI) is currently reshaping game development practices, production pipelines, and value networks in an unprecedentedly pervasive manner with cascading consequences remaining unclear. In the last five years since GenAI’s inception, a growing body of qualitative research has explored these early transformations from different settings and demographic angles. However, these studies often contextualise and consolidate their findings weakly with related work; for research to keep up with and support stakeholders in this development, the current moment calls for a synthesis of the findings emerged thus far. Here, we address this need through a qualitative research synthesis via meta-ethnography. We followed PRISMA-S to systematically search the relevant literature from 2020-2025, including major HCI and games research databases. We then synthesised the ten eligible studies, conducting reciprocal translation and line-of-argument synthesis guided by eMERGe, informed by CASP quality appraisal. We identified nine overarching themes, provide recommendations, and contextualise our insights in wider game production trajectories. With this work, we seek to provide practitioners, researchers and policy-makers with grounded insights to guide practice, research and governance.
Overview
CoChat Summary
TL;DR
This paper reports the first qualitative research synthesis on generative AI in game development, synthesizing 10 eligible qualitative studies (2020–June 2025) via meta-ethnography. It identifies nine overarching themes showing GenAI as a provisional, assistant-like technology whose value is concentrated in human-steered ideation and prototyping, with integration conditioned by pipeline fit, governance, and contested authorship/labour concerns.
Key findings
- Across 10 included studies, the synthesis produced nine overarching themes (C1–C9) spanning human-in-the-loop refinement, early-stage ideation, contested efficiency, pipeline/artefact constraints, socio-technical positioning, access/adoption, risks/ethics/labour, governance, and authorship/materiality.
- Reciprocal translation yielded a dense Production network (79 unique interpretations, 105 links), a more compact Learning network (37 unique interpretations, 35 links), and a cross-group layer (24 links). Relations mostly Enrich or Support rather than directly Translate; contradictions were very rare (Production 1.9%; cross-environment links dominated by Enriches at 83%).
- Findings converge on four points: GenAI’s present value is upstream (ideation) rather than end-to-end authorship; human intervention remains necessary for output integration; efficiencies are conditional rather than general; and realised benefit depends on pipeline fit and existing governance.
- CASP appraisal found all 10 studies clearly stated aims and proposed appropriate research design, but variation was highest for criteria 7–9 (researcher–participant relationship, ethics, rigorous data analysis), with only 4 rated Yes for rigorous data analysis, 3 Somewhat, and 3 No. The paper states more than half of studies show medium to strong methodological weaknesses.
Methods
- Interpretive qualitative research synthesis via meta-ethnography, following eMERGe reporting, PRISMA-S search reporting, and CASP quality appraisal; analysis used reciprocal translation and line-of-argument synthesis.
- Systematic search covered January 2020–June 2025 across ACM Digital Library, Scopus, DiGRA and FDG proceedings, and Google Scholar; after duplicate removal 320 records remained; title/abstract screening identified 7 clearly eligible and 17 borderline studies; full-text review confirmed 3 more, yielding 10 included studies.
- Included studies were split into Learning (4 studies) and Production (6 studies) groups; extraction was done in Atlas.ti 25, with two-stage reciprocal translation first within and then across groups, using the oldest study as an index and custom connectors (Translates, Supports, Enriches, Contradicts, Originally part of).
Limitations
- The paper states the evidence base remains modest and uneven, with primary studies varying in conceptual depth and methodological quality; this asymmetry shapes translation network density and limits cross-corpus precision. Poor second-order interpretations in some studies required augmentation, adding an interpretive layer.
- The paper states studies are few, scattered over more than three years with typically short data collection intervals, so the synthesis cannot track changes over time and must not be understood as a snapshot of the present. It also states game development is not unified, the synthesis cannot account for regional/cultural/community differences in sufficient detail, comprehensive game-development theory is largely absent, and wider industry forces are not integrated at the same level as GenAI.
- Evident from the text: only 10 studies are included, many with small samples and mostly capturing early adoption; empirical coverage is uneven (2D imagery and code relatively well evidenced, while 3D, animation, audio, and mixed-pipeline artefacts remain under-studied); no included studies are longitudinal; US industry studies focus only on small studios or solo developers; and some studies lacked consistent reporting of model choice/configuration and qualitative coding standards.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
24 Valid · 18 Not valid · 91 Not checked · 228 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“However, practitioners also often subvert these conventions by misusing or stretching tools [23].”
What the sources say
- Could not verifycites [23] [23] Andrew Hugill and Hongji Yang. 2013. The creative turn: new challenges for computing. International Journal of Creative Computing 1, 1 (2013), 4-19. - sourceExperimental — accuracy not yet measured
Section 1, “Creative Computing,” discusses software constraints on creative practice, and section 2, “The creative turn,” says people often adapt their practices to suit software. Neither says practitioners often subvert platform conventions by misusing or stretching tools. Searches for repurposing tools, unintended uses, and stretching tools found no passage addressing that behavior. (LLM, conf=0.72)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
Every source the paper cites for this claim was checked and supports it.
In the paper
“In a UK classroom, French et al. [20] examined early integration of LLM and TTIG tools into individual game projects.”
What the sources say
- Supportedcites [20] [20] Fiona French, David Levi, Csaba Maczo, Aiste Simonaityte, Stefanos Triantafyllidis, and Gergo Varda. 2023. Creative Use of OpenAI in Education: Case Studies from Game Development. Multimodal Technologies and Interaction 7, 8 (Aug. 2023), 81. doi:10.3390/mti7080081 Number: 8 Publisher: Multidisciplinary Digital Publishing Institute. - source
The Method and Rationale (p. 3) places the study among games-programming students at London Metropolitan University and identifies ChatGPT and Dall-E as the tools of interest. The Discussion (p. 17) describes game-development projects using Dall-E and ChatGPT, while Limitations and Implications (p. 18) says students individually planned how to test OpenAI tools for game development. (LLM, conf=0.92)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Shifting toward a production-oriented model, Shields et al. [67] documented a US serious games project where students and academics develop an educational visual novel.”
What the sources say
- Supportedcites [67] [67] Samuel Shields, Alexander Calderwood, Shi Johnson-Bey, Noah Wardrip-Fruin, Michael Mateas, and Edward Melcer. 2024. Generating Together: Lessons Learned from Developing an Educational Visual Novel with AI Collaboration. In 2024 IEEE Conference on Games (CoG) . 1-8. doi:10.1109/ CoG60054.2024.10645553 ISSN: 2325-4289. - source
The Introduction (p. 2) describes the authors’ work on Academical as an interactive-narrative serious game. The Writer’s Room section (p. 5) says undergraduate and master’s students participated in its development, while the Production Environment section (p. 4) describes the academic development team. The author affiliations (p. 2) place the project in the United States. (LLM, conf=0.90)
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Boucher et al. [5] studied a US university summer programme simulating commercial game development with small, diverse intern teams.”
What the sources say
- Supportedcites [5] [5] Josiah D Boucher, Gillian Smith, and Yunus Doğan Telliel. 2024. Is Resistance Futile?: Early Career Game Developers, Generative AI, and Ethical Skepticism. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI '24) . Association for Computing Machinery, New York, NY, USA, 1-13. doi:10.1145/3613904.3641889 - source
The Introduction describes a US summer professional-development programme in which interns form teams and make mobile games through to publication. Section 3.1 reports 26 interns across four teams with varied backgrounds, while Section 3.3 locates their workplace on a university campus. Together, these passages support the description of small, diverse teams working in an industry-oriented game-development programme; page numbers are unavailable in our copy. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, even among later studies, substantive crossreference to other empirical studies in game production remains rare.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
116 more claims
Create a free account to see all 361 claims and every source verdict.
- Probably not a substantial claim. 161
- It may state two separate claims as one. 139
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 8 Description of Included Studies · p. 12
“| [67] | Industry Professionals | 9 | Postmortem case study | - | Reflective analysis as case studies | L |”
§ 8.1 Group A: Learning · p. 13
“Shields et al. [67] documented a US serious games project where students and academics develop an educational visual novel.”
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 3×
“GPT-3 and 3.5 [7] popularised conversational interaction through ChatGPT , while GitHub Copilot demonstrated how code generation could accelerate engine scripting and gameplay logic.”
InfluentialHigh-Resolution Image Synthesis with Latent Diffusion Models
Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01042 · Cited by 20,302
Learning Transferable Visual Models From Natural Language Supervision
Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. (2021) · arXiv · doi:10.48550/arxiv.2103.00020 · Cited by 15,293
Rethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., Koffel, J. B., PRISMA-S Group, Blunt, H., Brigham, T., et al. (2021) · Systematic Reviews · doi:10.1186/s13643-020-01542-z · Cited by 3,714 · cited 3×
“To foster transparency and reproducibility, we implement the Pre-ferred Reporting Items for Systematic Reviews and Meta-Analyses literature search extension (PRISMA-S) guidelines for reporting literature searches in systematic reviews [62], thus adding detail to eMERGe steps 5–7.”
InfluentialNoblit, G., & Hare, R. (1988) · doi:10.4135/9781412985000 · Cited by 2,723 · cited 4×
“Given the interpretative nature of meta-ethnography, the partial integrations in the primary literature above can be used as contrast and complement our synthesis.”
InfluentialHierarchical Text-Conditional Image Generation with CLIP Latents
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022) · arXiv · doi:10.48550/arxiv.2204.06125 · Cited by 2,511
Zero-shot text-to-image generation
Ramesh, A., Pavlov, M., Goh, G., Gray, S., Voss, C., Radford, A., Chen, M., & Sutskever, I. (2021) · arXiv · doi:10.48550/arxiv.2102.12092 · Cited by 2,417
Long, H. A., French, D. P., & Brooks, J. M. (2020) · Research Methods in Medicine & Health Sciences · doi:10.1177/2632084320947559 · Cited by 1,112 · cited 9×
“…meta-ethnography [10, 11], we use quality appraisal not to determine study inclusion, but to (1) support interpretive confidence of each study’s findings during extraction and translation [35] and to (2) critically reflect on methodology as a lens to pinpoint gaps in research (our fourth RQ, Sec.”
InfluentialUsing meta ethnography to synthesise qualitative research: a worked example
Britten, N., Campbell, R., Pope, C., Donovan, J., Morgan, M., & Pill, R. (2002) · Journal of Health Services Research & Policy · doi:10.1258/135581902320432732 · Cited by 1,110
Research works printed but not linked: 10
- Harold Cohen and Pamela McCorduck. 1991. Aaron’s code: meta-art, artificial intelligence, and the work of Harold Cohen. W. H. Freeman & Co.
- Jonathan Ho Ajay Jain and Pieter Abbeel. 2020. Denoising Diffusion Probabilistic Models. Advances in Neural Information Processing Systems 33 (2020) 6840–6851.
- Anna Kantosalo and Tapio Takala. 2020. Five C’s for Human–Computer Co-Creativity: An Update on Classical Creativity Perspectives. In International Conference on Computational Creativity. Association for Computational Creativity, 17–24.
- Anna Kantosalo and Hannu Toivonen. 2016. Modes for creative human-computer collaboration: Alternating and task-divided co-creativity. In Proceedings of the seventh international conference on computational creativity. 77–84.
- Annakaisa Kultima. 2018. Game design praxiology. Ph. D. Dissertation. University of Tampere.
- Midjourney. 2022. Midjourney. https://www.midjourney.com/home/
- OpenAI. 2022. DALL·E 2. https://openai.com/dall-e-2/
- National Information Standards Organization. 2022. Contributor Roles Taxonomy (CRediT). http://credit.niso.org/
- Jay Ratican and James Hutson. 2024. Adaptive worlds: Generative AI in game design and future of gaming and interactive media. ISRG Journal of Arts Humanities and Social Sciences 2 5 (2024).
- Donald A Schön. 1983. The reflective practitioner: How professionals think in action. Basic Books.
Cited by
Design of Generative AI Game NPCs Based on Human-Computer Interaction
Huzhongchenwei (2026) · Frontiers in Computing and Intelligent Systems · doi:10.54097/hy3mx976
Surendran, A., & Cooper, S. (2026) · International Conference on the Foundations of Digital Games (FDG) · doi:10.1145/3815598.3815713
AI as a Democratizing Force in Indie Game Development
Madanamootoo, B., & Alla, J. (2026) · arXiv · doi:10.48550/arxiv.2608.07825
“Generative AI's first application to games (2022–2024) extended content democratization: image, audio, code, and text generation, which is where 60% of Steam's AI disclosures sit [14] and where the prior qualitative literature focused [15–18].”
Hu, J., Li, J., Wang, Z., Li, Z., Chan, K.-C., Ma, X., Jiang, Z., & Zhang, Y. (2026) · AHFE International · doi:10.54941/ahfe1007640
Metadata
- Authors
- Ternar, D.-A., Denisova, A., Cunha, J. M., Kultima, A., & Guckelsberger, C.
- Affiliations
- Department of Computer Science, Aalto University, Espoo, Finland
- University of York (GB)
- University of Coimbra (PT)
- Aalto University (FI)
- ORCID
- Venue
- Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems
- Publisher
- ACM
- Published
- 2026-04-13
- Pages
- 1-24
- OpenAlex
- W4415089230
- Semantic Scholar
- 9d03c44c4614a3e38440254f7178b4c33ca3a70d
- Cited by
- 6
- References
- 46
- Citations by year
- 2026: 6
- FWCI
- 50.61
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by)
- Aaltodoc (Aalto University) (submittedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- White Rose Research Online (University of Leeds, The University of Sheffield, University of York) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- Fundação para a Ciência e a Tecnologia (UID/00326, C645008882-00000055, 00326)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Social Sciences › Sociology and Political Science › Digital Games and Media
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- generative AI, game development, value networks, video game industry, AI governance, meta-ethnography, evidence synthesis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Generative AI in Game Development: A Qualitative Research Synthesis. https://app.cochat.ai/papers/10.1145/3772318.3791206
- The paper
- Ternar, D.-A., Denisova, A., Cunha, J. M., Kultima, A., & Guckelsberger, C. (2026). Generative AI in Game Development: A Qualitative Research Synthesis. Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3772318.3791206
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.