Fallout 4 Tester: AI Can’t Match My ‘Professional Stupidity’ — Superbombs, Crashes and Glorious Chaos

Fallout 4 Tester: AI Can’t Match My ‘Professional Stupidity’ — Superbombs, Crashes and Glorious Chaos

The day a tester made Fallout 4 explode (on purpose)

Here’s a delightfully dumb thing a human tester did that, he insists, no AI could’ve dreamed up. Colin McInerney — a one-time Fallout 4 tester who now runs Pedalboard Games — told a GDC interviewer that he intentionally broke the game in ways that feel very human: creative, silly, and utterly relentless.

His experiment was gloriously absurd. He gave his save an absurd amount of experience points (we’re talking a billion), which shot him up to level 247. Then he used a launcher that normally spat out two nukes and slapped on a mod that turned each shot into ten bombs. He ran around the Wasteland tossing these “superbombs” everywhere and, unsurprisingly, the console’s RAM hit the wall. Result: four crashes in a single morning. Chaos achieved.

Why this matters: humans, AI, and the weirdness of testing

The game industry is very keen on AI — concept art, code helpers, even automated testing get the AI treatment. But Colin’s point is simple and important for gamers: people can try stuff that’s bizarre, out-of-spec, or just plain dumb, and those stunts often reveal problems machines won’t think to check for.

That kind of “professional stupidity” isn’t an insult — it’s a testing superpower. Human testers invent improbable scenarios, chain unlikely mods and cheats, and stress systems in ways that reveal crashes, memory issues, or gameplay problems that automated scripts miss. Those discoveries make the final game better, and they matter when studios are patching or porting titles.

Context: McInerney shared this anecdote at GDC in an interview with GamesRadar+. Fallout 4 originally launched in 2015 and, fun fact for people still playing, it recently arrived on the Nintendo Switch 2 last month. Meanwhile, generative AI tools keep popping up across studios — sometimes controversially — but anecdotes like this remind us humans still bring a weird, valuable edge.

Gamers should care because testing like this shapes stability and polish. When humans break things in ridiculous ways, developers learn how the game behaves at the limits, which leads to real fixes. Plus, it’s just comforting to know someone out there is intentionally being gloriously silly so our playthroughs are less likely to end with an unexpected crash.