Scenarios for the future of AI: unblocking the logjam

AI is likely to become more capable more quickly than most people expect. However, the public discussion about scenarios for the future of AI is often stuck in an unproductive impasse. This talk suggests how to unblock that logjam. It goes on to outline a roadmap toward what has been called BGI, namely beneficial general intelligence.

Topics covered include:
*) Real-world explanations of the catastrophic risks posed by increasingly powerful AI
*) The technical feasibility of safe AI superintelligence
*) The political and geo-political feasibility of safe AI superintelligence

The presentation draws upon the speaker's many discussions over the last 20 years with leaders from the worlds of politics, technology, law, economics, and philosophy.