Half of Game Developers Now Say Generative AI Is Making Things Worse

That combination is the interesting part. A tool people are increasingly using is a tool they increasingly think is bad for them. This piece is about what the survey actually measured, why those two numbers can move together, and what it does not tell us — including the causal claim almost every headline attached to it.

Key takeaways

• 52% now say generative AI has a negative industry impact, up from 30% a year earlier; only 7% say positive.
• Usage is unchanged at around 36%, but it is concentrated outside development — 58% in publishing and marketing roles against 30% at game studios.
• 28% were laid off in the past two years (33% in the US); 36% of those laid off over a year ago are still unemployed.
• 82% of US respondents support unionisation; roughly 10% are actually in a union.
• The survey measures belief and experience, not causation. It does not show AI caused the layoffs.

The sentiment collapse, by discipline

Aggregate numbers hide the shape of this. Broken out by what people actually do, negativity tracks exposure almost perfectly:

• Visual and technical art — 64% negative
• Design and narrative — 63% negative
• Programming — 59% negative

These are the three disciplines where generative tools produce output that resembles the deliverable. An image model makes something that looks like concept art. A language model writes something that reads like barks or item descriptions. A coding assistant emits something shaped like an implementation.

The pattern to notice is that negativity is highest where the technology is most capable, not where it is most mysterious. This is not a story about people fearing a thing they have not seen. The people closest to the output are the least impressed by what it means for them.

There is a second split that gets less attention. Adoption is 58% among professionals in publishing, marketing and support functions, against 30% at game studios themselves. Generative AI has landed hardest in the parts of the industry that produce store descriptions, localisation drafts, marketing copy and support responses — text where "good enough and instant" is a real business case. Actual game development has taken it up at roughly half that rate.

Among those who do use the tools, ChatGPT dominates at 74%, followed by Google Gemini at 37% and Microsoft Copilot at 22%. Those are unremarkable market-share numbers; the interesting result is that heavy use and negative sentiment coexist in the same population. If you want the practical view of getting useful output from these tools rather than the industrial one, our guide to writing AI prompts covers the mechanics.

The layoffs, and why the recovery figure is the alarming one

The AI result did not arrive in a vacuum. It arrived in the fourth consecutive year of contraction.

28% of respondents had been laid off in the previous two years, rising to 33% in the United States. 17% had been laid off in the previous twelve months alone — up from 11% in the twelve months before that, meaning the rate of cuts was still accelerating at the time of the survey rather than levelling off. Half said their current or most recent employer had conducted layoffs in the past year: 67% at AAA studios, 33% at indie studios.

But the number that deserves the most attention is the one about what happened next. 48% of laid-off respondents were still looking for work. And 36% of people laid off more than a year earlier were still unemployed.

That distinction matters more than the headline layoff count. In an ordinary cyclical downturn, cut roles reappear — at a competitor, at a new studio, in a different genre — and the churn shows up as short unemployment spells. When more than a third of people are still out of work a year later, the roles are not reappearing. That is the signature of a structural contraction: the industry is not redistributing this labour, it is shedding it.

When respondents were asked why their own layoff happened, they cited restructuring (43%), budget cuts and market conditions (38%) and project cancellation (32%). Not automation.

The pressure is financial, and it is visible in the deals as much as the surveys — the debt loaded onto EA in the largest buyout the industry has seen creates exactly the kind of fixed annual cost that gets met through headcount. Meanwhile 74% of surveyed students said they were worried about their prospects, which is the same finding measured at the entry point.

What the survey does not show

Here is where most coverage of this report went wrong, including headlines that ran the AI statistic directly above the layoff statistic.

The survey does not establish that generative AI caused these job losses. It cannot. It is an opinion and experience survey. It records that developers feel negatively about AI and separately records that developers have been laid off, and when asked directly about the cause of their own layoff, respondents named restructuring, budgets and cancellations.

The honest reading is a correlation of mood and circumstance. People working in a contracting industry, watching tools emerge that produce approximations of their deliverables, report a sharply negative view of those tools. That is entirely rational and it is also not evidence of a mechanism. Both of the following are consistent with this data:

• Studios are using generative tools to avoid rehiring, so AI is suppressing recovery even where it did not cause the initial cuts.
• Studios cut for straightforward financial reasons after a post-pandemic over-expansion, and generative AI is a parallel development that developers resent for reasons of craft and career security rather than direct displacement.

Nothing in the survey distinguishes these. Anyone asserting one confidently is doing so from somewhere other than this report.

There is also a sampling caveat worth stating. GDC's survey draws heavily on people connected to the conference — a self-selecting population that skews toward North America, toward developers engaged with the industry's professional discourse, and away from mobile studios in Asia, which is where a large share of the world's game revenue and headcount actually sits. It is a good instrument for measuring Western developer sentiment. It is not a global census.

And it is now seven months old. Published in January 2026, it predates a year that has included further consolidation and the completion of the largest leveraged buyout in corporate history. If the trend held, the 2027 figure will be worse.

The organising number

The result most likely to have consequences is the labour one.

82% of US respondents said they favour unionisation. 5% opposed. 13% were unsure. Actual union membership sat at about 10%, plus 2% in a company union — and 62% of non-members said they were interested in joining.

An eight-to-one support ratio with roughly one-in-ten membership is an enormous gap between sentiment and structure. Gaps like that tend to close in one of two directions: either the conditions driving support ease, or the organising infrastructure catches up with the sentiment. Four consecutive years of layoffs, a third of the US workforce personally affected, and a technology that developers overwhelmingly read as a threat to their craft are not conditions that ease on their own.

The AI result and the union result are probably the same result, measured twice.

What is genuinely unknown

Reasonably established: the survey's own findings — sentiment percentages, layoff experience, unionisation attitudes, tool adoption — as self-reported by more than 2,300 professionals in late 2025 and published in January 2026.

Not established: whether generative AI is displacing game industry jobs, at what rate, and in which disciplines; whether adoption has risen since January; whether the negative sentiment reflects direct experience of displacement or a broader assessment of craft and quality; and whether the union support converts into membership. The survey is a snapshot of what a particular slice of the industry believes, which is a genuinely valuable thing to know and a different thing from knowing what is happening to the industry.

The one claim the data supports without qualification is the one in the headline: within a single year, the professionals who build games moved from mostly ambivalent about generative AI to majority-negative, and they did it while continuing to use it.

Sources: GDC — 2026 State of the Game Industry; Game Developer — one in four developers laid off over the past two years; Business Wire — 2026 State of the Game Industry report announcement. Survey figures are as published in January 2026; this article is current as of 13 August 2026.