LessWrong post critiquing formal verification for AI safety

[Steve Omohundro]

Andrew Dickson recently posted this critique of the "Provably Safe AI" and "Guaranteed Safe AI" proposals on LessWrong:

Limitations on Formal Verification for AI Safety

https://www.lesswrong.com/posts/B2bg677TaS4cmDPzL/limitations-on-formal-verification-for-ai-safety

He's skeptical that formal methods can be used for AI safety based on his analysis of their current uses. 

I wrote a bunch of responses to his points (19 posts at last count!) and other people on LessWrong have been posting insightful comments and questions. 

Hopefully I've been able to communicate my perspective, but I'm sure he'd appreciate input from other people working in this area.

Best,

Steve