Evaluators and safety organisations · Article · Apollo Research
Embedded Evaluators are necessary for meaningful external testing
Read at apolloresearch.ai ↗Read at sourceNo licence statement found
Outline · 45 paragraphs
- Current external access is insufficient for meaningful safety assessments¶3
- Embedded evaluations are necessary¶13
- We might not get more warning shots¶18
- Embedded third party evaluations are feasible¶22
- Requirements to make Embedded Evaluations a success¶27
- Deeper access has already produced important safety findings¶37
- Embedded evaluations are necessary but not sufficient for safety¶44