adminGenerative AI in game development is often presented as a way to save time, automate routine tasks and help teams move faster. But discussions at the GamesIndustry.biz HR Summit on October 1 highlighted a more complicated picture: potential efficiency gains can come with staff anxiety, data-security concerns, inconsistent workflows and additional work for HR teams.
Attendees described a technology that may be useful in specific situations, but is not a universal fix. For studios, the practical challenge is to decide where AI is appropriate, set clear rules and check whether its use genuinely improves both the work and the experience of the people doing it.
Generative AI in game development: promise and pressure
Some summit participants said AI could help with tasks such as generating code or handling repetitive work that does not require a creative team’s full attention. One senior leader said their company would not allow AI-generated content in a finished game, while still considering whether the technology could support other activities.
That distinction matters: a studio can assess AI separately for different tasks rather than treating adoption as an all-or-nothing decision. Before introducing a tool, teams can identify the intended use, the information it may handle and who will review its output.
At the same time, employees and managers are uncertain about how AI may affect roles and future planning. Some workers have reportedly asked whether the company’s AI use means their jobs are at risk. Summit attendees noted that even leadership teams do not know exactly how the technology or its impact on work will develop.
Company policies are becoming more common, but keeping them current is difficult. One leader described having to reconsider guidance soon after writing it because of newly emerging security concerns, including prompt injection. A practical policy should therefore have a clear owner and a regular review schedule, rather than being treated as a document that never needs updating.
Data protection and clear rules
Protecting confidential information was a major concern. Even if a company uses an enterprise AI service with terms that prevent submitted data from being used to train the provider’s models, employees may still use personal accounts or other tools. That can expose confidential material or information covered by a non-disclosure agreement.
Studios can reduce uncertainty by explaining which tools are approved, what information must not be entered, and what kinds of work are permitted. These rules are more useful when employees can find them easily and understand the reasons behind them. Teams should also establish a review process for AI-assisted work, particularly when code or creative assets may enter a production pipeline.
Some companies are still experimenting, while others are testing AI in selected workflows and weighing possible time savings against financial, operational and legal risks. A small, clearly defined trial can help a studio learn whether a tool is useful before expanding its use.
Why AI workflows can create extra work
Uncontrolled use can lead to rework. One example shared at the summit involved an artist using Gemini to produce shader code while a programmer was away. The code was inefficient, failed to follow the studio’s coding standards and conflicted with its AI rules. The issue was discovered weeks later, and the build had to be reverted.
Another example concerned an outsourced artwork that had been generated with AI. Reviewers failed to catch visible problems before approving it, and the community reacted negatively. In that case, the expected savings in time and cost did not materialise.
These examples show why human review and documented procedures matter. AI tools may expose gaps in a workflow when teams rely on individual knowledge that has not been written down. Before piloting AI, studios can document the existing process, define quality checks and decide who is responsible for approving the result.
Speed alone is not a sufficient measure of success. A workflow may produce more code or content in less time but still make the job more frustrating if employees spend long periods trying to get an AI tool to follow instructions. Teams should consider output quality and employee feedback alongside efficiency.
How AI affects HR teams
Attendees identified some potential HR uses. AI may help employees organise their thoughts when drafting messages, and automated systems can answer routine questions about subjects such as sickness or maternity leave.
However, HR teams also reported receiving lengthy grievance letters prepared with generative AI. A legal representative said they received around ten such letters from clients each week, and described documents that could contain extensive details that were not accurate, alongside unusual or irrelevant legal references. HR staff still need to read and respond to formal complaints, so AI-assisted writing may increase rather than reduce their workload.
One attendee recalled a grievance of about nine pages, most of which was difficult to understand. The issue itself was relatively minor and might have been resolved through a direct conversation. Attendees also cautioned that a person who cannot explain the contents of an AI-generated submission may face difficulties if the matter reaches a tribunal.
A blanket ban on AI-assisted complaints may not be suitable for everyone, including people who need writing support. Possible approaches discussed included setting a reasonable word limit and asking for specific dates, events and details. Clear, consistent procedures can help HR teams understand the substance of a complaint without dismissing an employee who needs assistance expressing it.
The human side of AI at work
Automated answers can provide information, but they may not reveal why someone is asking. A question about maternity leave, for example, could be connected to a need for additional support. A question about sickness absence might reflect concern about time away for an operation and recovery.
HR professionals can recognise context, ask follow-up questions and respond with empathy in ways an automated system may not. For that reason, AI can support routine administration, but organisations should consider when a person needs to be involved. A useful policy should make it easy for employees to reach a human contact when an issue is personal or complex.
Conclusion: make AI useful, not just faster
The summit’s discussions suggest that generative AI in game development can offer practical help, but the benefits depend on careful implementation. Clear policies, data safeguards, documented workflows, human review and regular employee feedback can help companies avoid preventable mistakes and assess whether a tool is genuinely useful. The key question is not simply whether AI can make a task faster, but whether it improves the quality of the work without creating more risk or burden elsewhere.
