This position involves working with Max Tegmark and colleagues at the Beneficial AI Foundation on turbocharging formal verification with AI tools as described here and in Towards Guaranteed Safe AI (a high-level introduction is given in the 2nd half of this TED talk).  The core idea is to deploy not untrusted neural networks, but AI-written verified code implementing machine-learned algorithms and knowledge. The position can be either remote or by MIT in Cambridge, Massachusetts.

The ideal candidate

• Significant experience in formal verification, ideally with PhD or extensive work in the field
• Want to ensure that powerful AI systems stay under human control
• Concerned that AGI is not many decades away,  and that current AI safety approaches may be inadequate for such a transformative technology
• Strong programmer
• Likes mathematics and machine learning
• Self-motivated and enjoys working fairly independently