Episode 44

The last of four episodes in which I quiz Eric Bruylant (my guest from episodes 16–19) on some follow-up issues related to our earlier discussion of the dangers of runaway artificial intelligence.

I quiz Eric on the plausibility of low-level AIs/robots taking over the economy, he explains the subtle (and mindstretching) idea of "anthropic capture" as a way of keeping an intelligence explosion in check, and we conclude by looking at some of the problems with the "coherent extrapolated volition" (CEV) approach to AI safety.

---

Some relevant links:

• artificial general intelligence
• Nick Bostrom
• N. Bostrom, Superintelligence: Paths, strategies and dangers (2014)
• simulation hypothesis
• AI takeover
• Coherent Extrapolated Volition