Reproduce “The Reality of AI and Biorisk” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Aidan Peppin, Anka Reuel, Stephen Casper, Elliot Jones, Andrew Strait, Usman Anwar, Anurag Agrawal, Sayash Kapoor and 5 more
This paper critically assesses the empirical evidence supporting the claim that Artificial Intelligence systems significantly increase biological risks, finding that popular concerns are not supported by available scientific data. It concludes that current AI models do not pose an immediate or substantial increase in biorisk and advocates for more scientifically rigorous, 'whole-chain' risk analyses to inform future policy and research.
Key insights
- Existing research on AI biorisks often lacks crucial baselines (e.g., comparing LLM access to internet-only access) and rigorous methodological transparency.
- A comprehensive 'whole-chain risk analysis' is necessary, as focusing solely on AI's ability to provide information or assist in design neglects the formidable barriers to physical execution of biological threats.
- The efficacy of AI biological tools is significantly constrained by the scarcity of relevant data and the inherent complexities of biological systems, limiting their ability to overcome real-world experimental challenges.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.