Machine Intelligence Research Institute

The Ma­chine In­tel­li­gence Re­search In­sti­tute, a pri­mary host of more for­mal and tech­ni­cal work on value al­ign­ment the­ory. (See Re­searchers in value al­ign­ment the­ory for a list of all re­searchers in the field and their af­fili­a­tions.) Cur­rently sup­ported pri­mar­ily by in­di­vi­d­ual donors, MIRI’s spe­cial­iza­tion is in sup­port­ing tech­ni­cally able re­searchers who are com­ing in from out­side academia, sup­port­ing ba­sic re­search that may not im­me­di­ately lead to pub­li­ca­tions, and car­ry­ing out very-high-lev­er­age ad­di­tional op­er­a­tions that can op­er­ate with­out aca­demic over­head or go­ing through a larger bu­reau­cracy. MIRI is also the de facto cen­ter for work on de­ci­sion the­ory and Vingean re­flec­tion. MIRI op­er­ates out of Berkeley, CA near the UC Berkeley cam­pus that hosts Stu­ar­tRus­sell, but is not di­rectly af­fili­ated with UC Berkeley.