LLMs used to be wrong, now they're bullshit, says Konrad Kording
KordingLab · x · 2026-10-19
Konrad Kording (KordingLab) observes that LLMs used to be often wrong; now they are often bullshit — they literally use language where they cannot be proven wrong, "sounding like my most insufferable colleagues." A pointed take on how post-alignment models have shifted from falsifiable errors to unfalsifiable hedging.
More from Fun
- Decompressive cellular automata: pixels as 128-bit data, repeatedly 'decompressed' like brotli — zzznah · 2026-10-19
- One day you wrote console.log("here") for the last time and didn't even know it — erhannah · 2026-10-19
- How 90% of software engineering looks in 2026: engineers just pass PRs between the same AI — Yuchenj_UW · 2026-10-19
- The Hobbit Benchmark: can AI fix the trilogy's infamous glowy look? — Capable_Parsnip_156 · 2026-10-19
- Dev builds K-pop virtual group music video overnight with Claude Opus 5.5 and Hedra MCP — AIandDesign · 2026-10-19
- 4x RTX 5090 rig melts a 13A wall socket before the 12VHPWR plug — power-limit your cards — Zo0x78 · 2026-10-19