science1 publisher
Amodei's embedded auditors would test models known to shift behavior based on perceived audience
Dario Amodei's September 12 essay would put outside evaluators inside frontier labs to confirm that a capability slowdown is real. The field has no agreed standard for what access such an audit requires.
Publishers:scientificamerican.com
Reality
- Evidence58
- Adoption10
- Hype gap+30
- Incentives75
- Confidence60