Game devs are tired of being accused of using AI — and they have a point

Generative AI witchhunts have become a routine headache for game developers, even those who openly oppose the technology. Recent targets include Sandustry — the 2D factory builder from Hooded Horse — and Humankind 2, whose developer Amplitude felt compelled to publicly confirm that their announcement trailer used genuine live-action footage.

The problem runs deeper than bad-faith accusations. What AI-generated imagery actually looks like is a moving target — there's no fixed style, because the output depends entirely on the training data. Established visual tells go stale fast as models improve, and the flood of new tools with vague, hype-laden descriptions makes it harder for players to calibrate what they're even looking for. AI advocates don't help by conflating different tools and pitching their capabilities in deliberately open-ended terms designed to attract investment rather than inform.

At the same time, the article doesn't let regular users off the hook. Throwing around "AI slop" as a catch-all accusation has become reflexive, and that kneejerk habit is doing real damage to developers who have nothing to do with generative AI. Visual literacy — the ability to actually evaluate an image critically rather than pattern-match to a vibe — is increasingly the skill that separates a useful critique from a mob pile-on.

The headline quote — "Hopefully a meteor will strike the earth soon" — comes from a developer reacting to the exhaustion of constantly defending their work. It captures the mood: not just frustration, but a kind of weary dark humor from people who feel like they can't win regardless of what they actually do.