When Neurons Play DOOM

A developer at kuber.studio has written a viscerally unsettling post about a biocomputer built from approximately 200,000 cultured human neurons that can play DOOM — reportedly better than a human player. The piece earned over 270 points on Hacker News.

Why This Is Different From LLMs

We've reassured ourselves that LLMs aren't conscious because they're fundamentally "next-token predictors" — elaborate math, not inner life. But biological computing breaks that reassurance. When a brain interprets electrical signals from the optic nerve, we call that "seeing." When 200,000 human neurons interpret visual data to play a game, are they seeing?

The Uncomfortable Numbers

200,000 neurons is already more than a jellyfish or a worm. We want to say it's not enough to constitute a person. But the author points out that commercial incentives are real: a human biocomputer could store far more information at a fraction of the power that silicon requires. The industry will pursue this.

The Same Reward Mechanism

What makes the piece haunting is the observation that this system uses the same reward mechanisms as LLM training. Did we build the first human biocomputer and immediately put it in a simulated hell, playing the same game on loop, forever? That question has moved from science fiction to something that deserves serious ethical attention now.

#ethics

The Ethical Dread of Biocomputers Made From Human Neurons