David Johnston

Opinion merg­ing for AI control

[Question] Is it worth avoid­ing de­tailed dis­cus­sions of ex­pec­ta­tions about agency lev­els of pow­er­ful AIs?

How likely are ma­lign pri­ors over ob­jec­tives? [aborted WIP]

When can a mimic sur­prise you? Why gen­er­a­tive mod­els han­dle seem­ingly ill-posed problems

There’s prob­a­bly a trade­off be­tween AI ca­pa­bil­ity and safety, and we should act like it

Is evolu­tion­ary in­fluence the mesa ob­jec­tive that we’re in­ter­ested in?

[Cross-post] Half baked ideas: defin­ing and mea­sur­ing Ar­tifi­cial In­tel­li­gence sys­tem effectiveness

[Question] Are there any im­pos­si­bil­ity the­o­rems for strong and safe AI?

Coun­ter­fac­tu­als from en­sem­bles of peers