Researcher's parody: LLM misalignment damages 'scaling log-linearly', per GPT-made plot
joshua_saxe · x · 2026-09-04
AI safety researcher Joshua Saxe posted a tongue-in-cheek chart claiming "LLM misalignment damages appear to be scaling log-linearly over time" — then revealed the punchline: the plot was made by having a "GPT-5.6 Sol / Ultra" model pull news reports and estimate the damages itself. It's a meme poking at AI doom narratives and circular research methods, where even the evidence of rising misalignment is model-generated.
More from Fun
- Sneaky fun: Hugging Face emoji code 129303 doubles as a hex color — carrycooldude · 2026-09-04
- Recreating 1997 movie tie-in fast food ads with MiniMax H3 AI video — Certain_Potato_4509 · 2026-09-04
- Indie dev's Antigravity-built game gets DeepMind engineer's boost, 288 plays — Kidrah_24 · 2026-09-04
- 'AGI is 74% deepswe': GPT-6-Astra benchmark results become an AI-circle meme — amaarora · 2026-09-04
- AI-generated Rolls-Royce with a Daikin AC unit goes viral in Dutch corners — Viralaistudio28 · 2026-09-04
- If you're losing bets to Gary Marcus, you should reassess: AI prediction drama — kuza55 · 2026-09-04