Stuart_Armstrong(Stuart Armstrong)

Com­par­ing re­ward learn­ing/​re­ward tam­per­ing formalisms

Prob­a­bil­ities, weights, sums: pretty much the same for re­ward functions

Learn­ing and ma­nipu­lat­ing learning

Re­ward func­tions and up­dat­ing as­sump­tions can hide a mul­ti­tude of sins

How should AIs up­date a prior over hu­man prefer­ences?

Dist­in­guish­ing lo­gis­tic curves

Dist­in­guish­ing lo­gis­tic curves: visual

Kurzweil’s pre­dic­tions’ in­di­vi­d­ual scores

Assess­ing Kurzweil pre­dic­tions about 2019: the results

Maths writer/​cowrit­ter needed: how you can’t dis­t­in­guish early ex­po­nen­tial from early sigmoid

Con­sis­tent Glo­ma­riza­tion should be feasible

Last chance for as­sess­ing Kurzweil

Databases of hu­man be­havi­our and prefer­ences?

So­lar sys­tem colon­i­sa­tion might not be driven by economics

“How con­ser­va­tive” should the par­tial max­imisers be?

Assess­ing Kurzweil’s 1999 pre­dic­tions for 2019

Call for vol­un­teers: as­sess­ing Kurzweil, 2019

An­throp­ics over-sim­plified: it’s about pri­ors, not updates

If I were a well-in­ten­tioned AI… IV: Mesa-optimising

If I were a well-in­ten­tioned AI… III: Ex­tremal Goodhart