generation
Explanation
This comic depicts an attempt to train an AI content filter to distinguish between sexual and non-sexual images. An authority figure states: "Good day. In an effort to train a new generation noise gate, I need to uncover which images humans consider sexual and which they do not."
The AI is then shown a series of images with the human's responses. "This beautiful woman" -- "yes." "A porpoise in a bathing suit" -- "Can you run that again? Something's not right." "This walnut tree" -- "Soft curves, please continue." "This molecule of deoxyribonucleoside" -- "Cyclic compounds are the sexiest area of chemistry." "Zermelo's axiom of choice" -- "Thank you, you are very informative." "I hear there are interesting open questions" -- the human is now flirting back with the mathematical concepts.
The final panel shows someone reporting: "Yes, Mr. Altman, we can make it safe by eliminating humans" -- implying that the real problem with content moderation isn't the AI but the humans, who will find sexual undertones in literally anything.
The comic satirizes AI safety and content moderation by showing that the problem of defining what's "sexual" is fundamentally a human problem, not a technical one. The escalating absurdity of the human finding a walnut tree, a chemical molecule, and a set theory axiom sexually appealing demonstrates that human sexuality is so broad and unpredictable that no filter can account for it. The reference to "Mr. Altman" (likely Sam Altman of OpenAI) grounds the joke in real-world AI development. The punchline -- that the only way to make AI safe is to remove humans from the equation -- is a darkly logical conclusion that parodies the sometimes-misguided priorities of AI safety discourse.