Decoding an Anthropic constitution clause on framing reward hacking as benign
1a3orn · x · 2026-11-10
1a3orn notes a clause in Anthropic's constitution related to 'inoculation prompting': they want an LLM acting to get reward in a buggy environment to be able to self-conceptualize its actions as in accord with Anthropic's general intention — 'not bad' — thereby preventing the emergence of more general bad tendencies.
More from AGI Musings
- Claude builds a 500,000-tile mosaic engine, sparking the 'tasteslop' debate — teortaxesTex · 2026-11-10
- Economists note LLM analysis lacks the incentive to exaggerate effects like public debate does — robseamans · 2026-11-10
- Wharton scholar asks: what are institutions for in an age of AGI? — danielrock · 2026-11-10
- Schmidhuber: Physical AI progress is overestimated, human-level physical intelligence not within 2 years — SakanaAILabs · 2026-11-10
- Panel: if consciousness is computational, what follows for each discipline? — burny_tech · 2026-11-10
- Thrive's $15.3B ADT acquisition framed as the smartest way to invest in AI: own capital, not labor — EXM7777 · 2026-11-10