tech
Behind the Curtain: The AI titans' biggest private fear
There's rare agreement among the top AI experts about the seriousness of the risk.
TL;DR
- AI leaders privately express their greatest concern is the creation of a killer pathogen.
- Advanced AI models could enable a bad actor to create a deadly pathogen before detection and prevention systems are ready.
- A study of 272 researchers found a 12% chance of catastrophic outcomes from AI's dangerous capabilities by 2030, with chances exceeding 20% without mitigation.
- 'Catastrophic' is defined as over 1 million deaths or $100 billion in damage.
- Top AI executives have publicly warned about the risk of AI-derived bioweapons.
- AI could expedite the engineering of novel pathogens by mapping biological vulnerabilities and generating genetic tweaks.
- The spread of such a pathogen could overwhelm biosurveillance and containment systems.
- Open-source AI models pose a unique challenge as they can be adapted outside regulatory regimes.