.png)
BlueDot Narrated
Audio versions of the core readings, blog posts, and papers from BlueDot courses.
BlueDot Narrated
Do the Biorisk Evaluations of AI Labs Actually Measure the Risk of Developing Bioweapons?
•
BlueDot Impact
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Audio versions of blogs and papers from BlueDot courses.
Original text: https://epoch.ai/gradient-updates/do-the-biorisk-evaluations-of-ai-labs-actually-measure-the-risk-of-developing-bioweapons
Author(s): Anson Ho and Arden Berg.
A podcast by BlueDot Impact.
1. Publicly described benchmarks are common but saturate rapidly, with uncertain implications for biorisk
2. We know little about most other AI biorisk evaluations
Evaluations are generally light on detail, and often by design
In practice, many biorisk evaluations only tell us about one model
3. Existing evaluations do not fully address the positions most skeptical of LLM-driven biorisk
The need for somatic tacit knowledge
The importance of infrastructure access
These objections are valid but do not rule out AI biorisk concerns
Discussion