Christopher King

Op­ti­mal­ity is the tiger, and an­noy­ing the user is its teeth

Is this a weak pivotal act: cre­at­ing nanobots that eat evil AGIs (but noth­ing else)?

Threat­en­ing to do the im­pos­si­ble: A solu­tion to spu­ri­ous coun­ter­fac­tu­als for func­tional de­ci­sion the­ory via proof theory

Bing find­ing ways to by­pass Microsoft’s filters with­out be­ing asked. Is it re­pro­ducible?

Is there a ML agent that aban­dons it’s util­ity func­tion out-of-dis­tri­bu­tion with­out los­ing ca­pa­bil­ities?

A rank­ing scale for how se­vere the side effects of solu­tions to AI x-risk are

Could Roko’s basilisk acausally bar­gain with a pa­per­clip max­i­mizer?

A bet­ter anal­ogy and ex­am­ple for teach­ing AI takeover: the ML Inferno

ARC tests to see if GPT-4 can es­cape hu­man con­trol; GPT-4 failed to do so

Ca­pa­bil­ities De­nial: The Danger of Un­der­es­ti­mat­ing AI

Ex­plor­ing the Pre­cau­tion­ary Prin­ci­ple in AI Devel­op­ment: His­tor­i­cal Analo­gies and Les­sons Learned

GPT-4 al­ign­ing with aca­sual de­ci­sion the­ory when in­structed to play games, but in­cludes a CDT ex­pla­na­tion that’s in­cor­rect if they differ

A crazy hy­poth­e­sis: GPT-4 already is agen­tic and is try­ing to take over the world!

Does GPT-4 ex­hibit agency when sum­ma­riz­ing ar­ti­cles?

More ex­per­i­ments in GPT-4 agency: writ­ing memos

GPT-4 is bad at strate­gic thinking

GPT-4 busted? Clear self-in­ter­est when sum­ma­riz­ing ar­ti­cles about it­self vs when ar­ti­cle talks about Claude, LLaMA, or DALL·E 2

Imag­ine a world where Microsoft em­ploy­ees used Bing

AI com­mu­nity build­ing: EliezerKart

Do we have a plan for the “first crit­i­cal try” prob­lem?