BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Amplification and dialogue as mechanisms for safe advanced AI - Be
 th Barnes\, Computer Lab\, University of Cambridge
DTSTART:20180124T170000Z
DTEND:20180124T183000Z
UID:TALK99949@talks.cam.ac.uk
CONTACT:Adrià Garriga Alonso
DESCRIPTION:Slides: https://valuealignment.ml/talks/2018-01-24-amplificati
 on.pdf\n\nThese techniques come at the problem of safety from a fairly dif
 ferent angle than the things we've discussed so far.\n\nAmplification is t
 he idea of bootstrapping a trusted core system\, increasing its capabiliti
 es while maintaining safety properties. Paul Christiano and the OpenAI saf
 ety team have worked on these ideas. One current suggestion for how to do 
 this has a lot in common with functional programming. For some more discus
 sion see e.g.  https://ai-alignment.com/alba-an-explicit-proposal-for-alig
 ned-ai-17a55f60bbcf
LOCATION: Cambridge University Engineering Department\, CBL Seminar room B
 E4-38.  For directions see http://learning.eng.cam.ac.uk/Public/Directions
END:VEVENT
END:VCALENDAR