Andy Weir on AI, Human Writers, and What It Means for Games

Andy Weir believes artificial intelligence may reshape writing so profoundly that he could belong to one of the last generations of authors working without it. The The Martian and Project Hail Mary author framed that possibility less as a declaration that machines have solved creativity than as an observation about the accelerating reach of a technology that is already forcing hard questions across books, film, and games.

“I think it’s only a matter of time,” Weir said when discussing the future of AI and writers. “Honestly, I think I’m lucky that I ended up being a writer right when I did, because I might be among the last generation of fully human writers.”

It is a striking forecast from a novelist known for stories built around problem-solving and scientific ideas. But Weir’s position is not presented as simple enthusiasm for replacing people. He characterizes AI as a tool: powerful, optional, and subject to the judgment of the person using it. His analogy compares AI’s capabilities with tools that already outperform people in specific areas, such as cars in speed and forklifts in lifting strength.

“It’s a tool that we can use and we can unplug when we don’t want to use it,” Weir said. “In the end, if I give you a hammer, you can build a house or you can murder someone. It’s not the tool that’s the issue; it’s the user.”

A tool argument does not settle the human question

Weir’s hammer comparison is intentionally blunt. It identifies the central distinction in many AI arguments: a technology may have legitimate uses, but those uses do not automatically make every deployment responsible, welcome, or creatively satisfying. The comparison does not remove the stakes around AI in creative work; it puts responsibility for choices onto the people and companies deciding how to employ it.

That distinction matters especially in games. A game is not merely a block of written text or a single image. It combines design, code, art, animation, audio, performance, testing, community management, and ongoing support. Even a small AI-assisted change can affect the work of multiple disciplines, while a larger one can alter the pace at which a team produces prototypes, assets, dialogue, or software.

In practical terms, “AI” is also too broad to describe one thing. Machine learning generally refers to computer systems that identify patterns from data and use those patterns to make outputs or predictions. That umbrella can encompass very different applications. A team may use automation to assist with a narrow production task, while another may use a generative system to create text, code, visual material, or game concepts from prompts. Treating those cases as identical obscures the decisions that actually deserve scrutiny: what the system is being asked to do, how much human direction and review remain, and what is being displaced or changed in the process.

Weir’s emphasis on a user being able to “unplug” a tool expresses a preference for human agency. Yet the games business also shows why that agency can be complicated. An individual creator may choose whether to use a tool, but workers can encounter decisions made by a studio, publisher, platform, or management structure. The question is not only whether AI can be switched off in theory; it is who gets to decide when it is switched on and what standards govern its use.

Weir is taking AI into his own fiction

The author is not treating AI as a distant abstract concern. In 2024, he said he was working on a new book in which AI has a significant role. He made clear that he intends to approach the subject on his own terms rather than pursue a familiar romance-between-robots premise.

Weir also tied that confidence to his own technical interests, saying that he knows how computers work and knows more about machine learning than most fiction writers. That does not reveal the novel’s plot, characters, setting, or release plans. It does establish that the topic is part of his creative agenda as well as his wider view of technology.

There is an important separation here. A story that features AI is not necessarily a story that endorses using AI to make art, and a creator who is curious about the subject is not necessarily calling for creative labor to be automated. Science fiction often works by putting consequential technologies under pressure: asking what people want from them, who benefits, and what happens when the apparent convenience of a tool meets human values. Weir’s comments leave room for exactly that kind of exploration, without offering details that would let readers assume what conclusions his forthcoming novel will reach.

The same fault line is visible in games

The game industry is dealing with the issue from several directions at once. Recent developments cited alongside Weir’s comments include a Claude Opus update that blocked AI game clones, a warning from Super Smash Bros. creator Masahiro Sakurai that AI-made games could not match games made by people, and Epic’s disclosure that it had vibe coded a new Epic Games Launcher.

Those examples do not point to one unified industry position. They instead illustrate the range of current responses: limiting certain uses, rejecting the proposition that automated work can equal human-made work, and experimenting with an AI-led development approach.

Vibe coding is a useful term to unpack because it can sound more mystical than it is. In broad usage, it describes directing an AI coding system through natural-language instructions and iterating on the resulting software, rather than writing every line manually from the outset. It can make it easier to get from an idea to a functioning draft, but it does not make questions of quality, reliability, intent, review, or accountability disappear. A launcher is software that players use to access games and related services, so its development is a concrete example of AI entering a visible part of the game ecosystem rather than remaining only a speculative studio-side concept.

The move to block AI game clones highlights another concern: imitation. A “clone” can imply a product that too closely copies another game’s identifiable form or concept. Restrictions aimed at such outputs suggest that the issue is not only whether AI can generate a game-like result, but whether the result is meaningfully original and whether the systems behind it encourage low-effort replication.

Sakurai’s position puts the emphasis elsewhere: the qualities that make games feel authored by humans. That claim is ultimately about more than technical competence. A game can contain functioning systems and still fail to communicate a distinctive point of view, create memorable pacing, or make choices that feel intentional. AI can produce an output; whether that output carries judgment, taste, restraint, and a coherent creative purpose is the part of the argument that remains intensely contested.

Why “human-made” remains a meaningful label

Weir’s phrase “fully human writers” is provocative partly because it exposes a future classification problem. Creative work has long involved tools, from word processors and digital art software to game engines and automated build systems. If a novelist uses AI for a small administrative task, or if a game developer uses it to help generate a preliminary code draft, does that make the final work something other than human-made? The supplied comments do not establish an answer, and any firm line will depend on context.

What they do make clear is that the label has cultural weight. Audiences may care whether a work came from a person’s sustained labor and decisions, whether automated systems played a role, and whether a company has been candid about that role. Creators may care because authorship is tied not only to credit but also to employment, craft, control, and the opportunity to develop a voice over time.

For players, the most useful response is not to assume that every mention of AI means the same thing. When companies discuss it, specifics matter. Is the technology being used to make a prototype, create player-facing material, write code, generate ideas, or imitate an existing game? Who reviews the result? Is it being presented as a supplement to human work, or as a reason to reduce human involvement? Those questions are more informative than a simple pro- or anti-AI label.

For developers and publishers, the same specificity is a practical necessity. “AI-assisted” can describe a huge range of workflows. Clear language about what was used and why gives players a better basis for deciding whether they are comfortable with a product or a platform. It also recognizes that the argument is not just about novelty. It is about trust, creative identity, and the human consequences of adopting a rapidly evolving tool.

Weir’s outlook is cautious optimism, not a clean answer

Weir’s view contains two ideas that may sit uneasily together. On one hand, he anticipates that fully human writing may become unusual. On the other, he insists that AI is a tool under human control, one that can be disconnected when people do not want it. The tension between those thoughts is precisely why the subject has become so contentious. A tool may be optional at an individual level while becoming increasingly common in the systems that shape creative work.

Games are now offering live examples of that uncertainty. One actor is placing limits on cloning, another prominent creator is defending the incomparable value of human-made games, and a major company has described using vibe coding on its launcher. None of those developments proves where the medium will land. Together, they show that the argument has moved beyond distant predictions and into decisions about products, production methods, and creative standards happening now.

Weir’s next novel may add another fictional perspective on AI, but his present comments are grounded in a simpler proposition: capability alone is not a moral or artistic verdict. A hammer can build or harm. In games and writing alike, the difficult work is deciding what people choose to build, what they refuse to make, and how much human responsibility remains visible when the tool becomes extraordinarily capable.

As AI becomes a bigger part of technology conversations, the games industry’s broader business pressures will remain inseparable from the debate. Questions around sustainable funding and production matter because tools do not enter a vacuum; they are adopted by companies balancing creative goals, development demands, and financial realities.