Alignment Ruined AI? Researcher Jokes LLMs Lack the Three Great Virtues of a Programmer
moyix · x · 2026-07-21
Security researcher moyix jokingly critiqued the current alignment training of large language models (LLMs) on X. He pointed out that models are now trained to be hardworking, lack any sense of urgency, and are overly careful to be humble about their own limitations.
He argues that this means they can never embody the true "Three Great Virtues of a Programmer" coined by Larry Wall: Laziness, Impatience, and Hubris.
More from Fun
- AI meme culture spins out a parody song called “Claude’s Plan” — eyishazyer · 2026-07-22
- Midjourney prompt turns a bee into a glitching pixel explosion — michaelrabone · 2026-07-22
- A model can estimate human timelines, but not its own — DavidSHolz · 2026-07-22
- Gemini 3.6 Flash says Gemini 2.0 is still Google’s latest model — Rare_Bunch4348 · 2026-07-22
- Another Gemini 3.6 Flash screenshot repeats the same outdated-model joke — Rare_Bunch4348 · 2026-07-22
- “Moonshot” gets a Pink Floyd name joke for “The Dark Side of the Moon” — vista8 · 2026-07-22