Davidad Dalrymple: Towards Provably Safe AI

5 Sep 2024 · The Gradient – Substack UK

UK government-funded research on formal verification and guaranteed AI safety could inform how Australian agencies approach frontier AI risk frameworks.

Key points

Summary

This item is a podcast episode from The Gradient featuring Davidad Dalrymple, Programme Director at the UK's Advanced Research and Invention Agency (ARIA). The conversation covers technical AI safety concepts including formal verification, the Open Agency Architecture, the Semantic and Deontic Sufficiency Hypotheses, and ARIA's Safeguarded AI Programme. Dalrymple also discusses AGI timelines, race dynamics, and collective deliberation for value specification. The extracted text is primarily a chapter outline rather than a transcript, limiting the depth of analysis possible.

Implications for Australian agencies

Implications are AI-generated. Starting points, not advice.