Oliver Sourbut comments on Talking publicly about AI risk

[Oliver Sourbut]

I wrote a little here about unpluggability (and crossposted on LessWrong/​AF)