Tarn Adams has spent more than two decades building Dwarf Fortress, one of the most obsessively detailed simulation games ever made. So when he sat down with PC Gamer at Gamescom 2026 and said the game industry is in the grip of AI-driven layoffs and executive delusion, it carried a different kind of weight than the usual developer-versus-suits complaint. "Everyone I know, their bosses are slowly getting psychosis, right?" he said, describing managers who now ask staff "did you use AI on your commits?" as a substitute for actually understanding the work. The clip resurfaced on Hacker News in September 2026, right as a fresh round of AI-attributed layoffs hit both games and the wider tech industry, and the comment thread that followed ended up saying almost as much about the moment as Adams did.
What Adams Actually Said at Gamescom 2026
The interview, conducted by PC Gamer's Joshua Wolens, wasn't primarily about AI — it was a wide-ranging conversation about Dwarf Fortress's ongoing development. But Adams kept circling back to what he's hearing from friends elsewhere in the industry: a pattern of executives who have convinced themselves that generative tools are close to making entire teams optional.
"A CEO Press a Button That Makes a Game"
Adams's sharpest line named the contradiction directly. "They're trying to have a CEO press a button that makes a game, and then everyone else somehow buys it without a job," he said. The line works because it exposes the logical hole at the center of AI-replaces-labor thinking in entertainment: if AI eliminates enough paying jobs across the economy, it also eliminates the customers who were supposed to buy whatever the AI makes.
The Psychosis Line
His most-quoted remark went further than skepticism about the technology and into a diagnosis of the people deploying it: "I mean, everyone I know, their bosses are slowly getting psychosis, right? They're just like, 'yeah, but did you use AI on your commits?'" Kotaku's write-up of the same interview reported that Adams pointed to tools like Claude and GitHub Copilot specifically — not as useless, but as being pushed into workflows by management that treats AI adoption itself as a performance metric, regardless of whether it produces better code. He added a practical counterpoint that rarely makes it into the boardroom pitch: debugging AI-generated code often takes longer than just writing it correctly the first time.
A Pattern Adams Has Seen Before
The Sewage Plant Story
Adams's skepticism isn't purely about AI — it's about how corporate management treats work it doesn't understand. He pointed to his own father being laid off from a sewage treatment plant, cut by managers who didn't grasp what the role actually did and assumed it could simply be absorbed away. In his framing, the AI moment is the same mistake at a much larger scale: executives who can't evaluate the real output of programmers or artists defaulting to "we'll probably have machines do the job" instead.
"Horrible, Greedy People"
Adams has made a version of this argument before. In a separate PC Gamer interview about the wave of studio layoffs, he was blunter about the executives responsible, saying "I think they're horrible… greedy, greedy people." Read together, his Gamescom comments aren't a one-off reaction to a news cycle — they're a continuation of an argument he's been making for months: that the layoffs are a leadership failure being laundered through an AI narrative.
"A Pop and a Reckoning"
Asked where this trajectory ends, Adams didn't predict a soft landing: "I don't see that going anywhere sustainable, and I feel like there will simply be a pop and a reckoning and then [it's on repeat] unless people do something else." That line is where the Hacker News thread got its title, and it frames the current wave of cuts as a bubble rather than a permanent restructuring of how games get made.
The Boss's Pitch vs. the Reality on the Ground
Strip away the rhetoric and Adams is describing a gap between what AI is promised to do inside a studio and what actually happens once it gets forced into daily workflows.
| What management asks for | The stated goal | What Adams says actually happens |
|---|---|---|
| "Did you use AI on your commits?" | Prove productivity gains from tools like Claude and Copilot | Debugging AI-generated code often takes longer than writing it correctly the first time |
| "We'll probably have machines do the job" | Replace headcount with AI output | Cuts land on roles leadership doesn't understand, echoing his father's layoff from a job management couldn't evaluate |
| "A CEO presses a button that makes a game" | Collapse the whole production pipeline into automation | The same automation would eliminate the paying customers needed to buy the product |
Why This Landed So Hard on Hacker News
Attention Is Fixed, Content Isn't
The top-voted comment on the thread, from user munificent, reframed Adams's complaint as an economics problem rather than a labor problem: "Human attention is a finite resource," while AI and other tools have "drastically lowered the cost to produce a work of digital media." The point is that you can generate unlimited additional content, but you cannot generate additional hours in a day for people to consume it. The comment cited figures often used to illustrate the scale of the existing glut — roughly 20 million videos uploaded to YouTube every day, and an estimated 600 million active blogs — numbers that were already large before generative AI started adding to the pile.
The Pushback: Is Supply Actually Exploding?
Not everyone bought the framing wholesale. Commenter asdff argued that rising spam volume doesn't mean the supply of things people actually want has grown: theatrical film releases, traditionally published books, and major game launches haven't multiplied tenfold just because AI can generate rough drafts of all three. The disagreement matters because it runs through the same fault line as the layoffs themselves — whether AI is creating a real oversupply of quality work, or just an oversupply of noise that management is choosing to count as output anyway.
Executives in an Echo Chamber
A third strand, from commenter martythemaniak, diagnosed the "psychosis" Adams described in more structural terms: Silicon Valley executives operate inside closed social and professional circles where only a narrow set of ideas circulate, producing synchronized behavior across competing companies — simultaneous mass layoffs at profitable firms, simultaneous AI pivots, and the same round-trip on pandemic-era hiring, first up and then sharply down. In that reading, Adams isn't describing individual bad bosses so much as an industry-wide groupthink arriving at the same conclusion at the same time.
Why Games Are Just the Visible Edge of This
The reason Adams's comments generalized so easily beyond games is that the underlying math — falling production costs colliding with a fixed amount of human attention — applies to every creative field AI touches, not just game studios. Commenters on the same thread extended the point to poetry, noting that almost no one can make a living writing poetry anymore, and to blogging and video, where the volume figures cited above already dwarfed what any audience could meaningfully consume even before generative tools entered the pipeline. Several commenters used the term "enshittification" to describe what they see as AI's actual near-term effect: not better products, but a faster rate of bloated, lower-quality output crowding out work that took real time to make.
The Counter-Argument: AI as an Excuse, Not a Cause
A more skeptical reading raised in the same debate is that AI efficiency gains are doing less work here than the layoff announcements suggest — that AI functions as a convenient justification for cuts studios wanted to make anyway, in an industry that has long preferred smaller, more disposable teams. Under that view, the "psychosis" isn't really about believing AI works; it's about having a socially acceptable reason to say so out loud.
What's Still Unresolved
Adams's remarks are an anecdote from one well-respected but idiosyncratic developer, not a labor study — he's describing what he hears from friends, not citing industry-wide data. Whether AI adoption is actually driving the current wave of game-industry layoffs, or merely providing cover for cuts that would have happened anyway, remains contested even among the people arguing about it on Hacker News. What both sides of that argument seem to agree on is the "pop and a reckoning" framing itself: a business model built on convincing customers to buy games made by ever-fewer people isn't obviously sustainable, whether or not AI ever closes the quality gap Adams says it hasn't closed yet.
If Adams is right, the reckoning he's predicting won't stay confined to games — it will show up everywhere the same equation applies: cheaper production, fixed attention, and executives betting the gap closes before the audience notices.
-EditorZ
Photo by kate.sade on Unsplash
Post a Comment