..read more

SoGive works with ma­jor donors. As part of our work, we meet with sev­eral (10-30 per year) char­i­ties, gen­er­ally ones recom­mended by eval­u­a­tors we trust, or (oc­ca­sion­ally) recom­mended by our own re­search. We learn a lot through these con­ver­sa­tions. This sug­gests that we might want to pub­lish our call notes so that oth­ers can also learn about the char­i­ties we speak with. Given that we take notes dur­ing the calls any­way, it might seem that it would be low cost for us to sim­ply pub­lish those. This would be de­cep­tive. There is a non-triv­ial time cost for us ..read more

We just pub­lished an in­ter­view: Dean Spears on why ba­bies are born small in Ut­tar Pradesh, and how to save their lives. Listen on Spo­tify or click through for other au­dio op­tions, the tran­script, and re­lated links. Below are the epi­sode sum­mary and some key ex­cerpts. Epi­sode summary I work in a place called Ut­tar Pradesh, which is a state in In­dia with 240 mil­lion peo­ple. One in ev­ery 33 peo­ple in the whole world lives in Ut­tar Pradesh. It would be the fifth largest coun­try if it were its own coun­try. And if it were its own coun­try, you’d prob­a­bly know about its hu­m ..read more

Man­i­fund is a philan­thropic startup that runs a web­site and pro­grams to fund awe­some pro­jects. From Jan­uary to now, we wrapped up 3 differ­ent pro­grams for im­pact cer­tifi­cates (aka ven­ture-style fund­ing for char­ity pro­jects): ACX Grants, Man­i­fold Com­mu­nity Fund, and the Chi­natalk es­say com­pe­ti­tion. Over­all, we’ve learned a lot and are happy with the pro­jects we’ve funded, but are less ex­cited by im­pact certs than be­fore — it’s been hard to get in­vestor in­ter­est, and we still haven’t found a use case where certs led to bet­ter fund­ing de­ci­sions. For the next ..read more

This is a cross-post and you can see the origi­nal here, writ­ten in 2022. I am not the origi­nal au­thor, but I thought it was good for more EAs to know about this. I am post­ing anony­mously for ob­vi­ous rea­sons, but I am a long­stand­ing EA who is con­cerned about Tor­res’s effects on our com­mu­nity. An in­com­plete summary Introduction This post com­piles ev­i­dence that Émile P. Tor­res, a philos­o­phy stu­dent at Leib­niz Univer­sität Han­nover in Ger­many, has a long pat­tern of con­cern­ing be­hav­ior, which in­cludes gross dis­tor­tion and falsifi­ca­tion, per­sis­tent ha­rass­ment ..read more

Ab­stract: Mo­ti­vated by the ques­tion of re­spon­si­ble AI and value al­ign­ment, I seek to offer a uniquely Fou­cauldian re­con­struc­tion of the prob­lem as the emer­gence of an eth­i­cal sub­ject in a dis­ci­plinary set­ting. This re­con­struc­tion con­trasts with the strictly hu­man-ori­ented pro­gramme typ­i­cal to cur­rent schol­ar­ship that of­ten views tech­nol­ogy in in­stru­men­tal terms. With this in mind, I prob­le­ma­tise the con­cept of a tech­nolog­i­cal sub­jec­tivity through an ex­plo­ra­tion of var­i­ous as­pects of ChatGPT in light of Fou­cault’s work, ar­gu­ing that cur­r ..read more

This an­nounce­ment was writ­ten by Toby Trem­lett, but don’t worry, I won’t an­swer the ques­tions for Lewis. Lewis Bol­lard, Pro­gram Direc­tor of Farm An­i­mal Welfare at Open Philan­thropy, will be hold­ing an AMA on Wed­nes­day 8th of May. Put all your ques­tions for him on this thread be­fore Wed­nes­day (you can add ques­tions later, but he may not see them). Lewis leads Open Philan­thropy’s Farm An­i­mal Welfare Strat­egy, which you can read more about here. Open Philan­thropy has given over 400 grants in its Farm An­i­mal Welfare fo­cus area, rang­ing from $15,000 to sup­port an­i­ma ..read more

An­nounc­ing open ap­pli­ca­tions for the AI Safety Ca­reers Course In­dia 2024! Ax­iom Fu­tures has launched its flag­ship AI Safety Ca­reers Course 2024 to equip emerg­ing tal­ent work­ing on In­dia with foun­da­tional knowl­edge in AI safety. Spread out across 8-10 weeks, the pro­gram will provide can­di­dates with key skills and net­work­ing op­por­tu­ni­ties to take their first step to­ward an im­pact­ful ca­reer in the do­main. Each week will cor­re­spond with a cur­ricu­lum mod­ule that can­di­dates will be ex­pected to com­plete, and dis­cuss with their co­hort dur­ing the fa­cil­i­ta ..read more

About a week ago, Spencer Green­berg and I were de­bat­ing what pro­por­tion of Effec­tive Altru­ists be­lieve en­light­en­ment is real. Since he has a large au­di­ence on X, we thought a poll would be a good way to in­crease our con­fi­dence in our predictions Be­fore I share my com­men­tary, I think in hind­sight it would have been bet­ter to ask the ques­tion like this: ‘Do you be­lieve that awak­en­ing/​en­light­en­ment (which frees a per­son from most or all suffer­ing for ex­tended pe­ri­ods, like weeks at a time) is a real phe­nomenon that some peo­ple achieve (e.g., through med­i­ta­ti ..read more

Cross­posted from LessWrong: https://​​www.less­wrong.com/​​posts/​​zjGh93nzTTMkHL2uY/​​the-in­ten­tional-stance-llms-edi­tion In memo­riam of Daniel C. Den­nett. tl;dr: I sketch out what it means to ap­ply Den­nett’s In­ten­tional Stance to LLMs. I ar­gue that the in­ten­tional vo­cab­u­lary is already ubiquitous in ex­per­i­men­ta­tion with these sys­tems there­fore what is miss­ing is the the­o­ret­i­cal frame­work to jus­tify this us­age. I aim to make up for that and ex­plain why the in­ten­tional stance is the best available ex­plana­tory tool for LLM be­hav­ior. Choos­ing Between Stance ..read more