Announcement: AI alignment prize round 4 winners

We (Zvi Mow­show­itz and Vladimir Slep­nev) are happy to an­nounce the re­sults of the fourth round of the AI Align­ment Prize, funded by Paul Chris­ti­ano. From July 15 to De­cem­ber 31, 2018 we re­ceived 10 en­tries, and are award­ing four prizes for a to­tal of $20,000.

The winners

We are award­ing two first prizes of $7,500 each. One of them goes to Alexan­der Turner for Pe­nal­iz­ing Im­pact via At­tain­able Utility Preser­va­tion; the other goes to Abram Dem­ski and Scott Garrabrant for the Embed­ded Agency se­quence.

We are also award­ing two sec­ond prizes of $2,500 each: to Ryan Carey for Ad­dress­ing three prob­lems with coun­ter­fac­tual cor­rigi­bil­ity, and to Wei Dai for Three AI Safety Re­lated Ideas and Two Ne­glected Prob­lems in Hu­man-AI Safety.

We will con­tact each win­ner by email to ar­range trans­fer of money. Many thanks to ev­ery­one else who par­ti­ci­pated!

Mov­ing on

This con­cludes the AI Align­ment Prize for now. It has stim­u­lated a lot of good work dur­ing its year-long run, but par­ti­ci­pa­tion has been slow­ing down from round to round, and we don’t think it’s worth con­tin­u­ing in its cur­rent form.

Once again, we’d like to thank ev­ery­one who sent us ar­ti­cles! And spe­cial thanks to Ben and Oliver from the LW2.0 team for their en­thu­si­asm and help.