robertskmiles

Early Access Video: Reward Hacking: Concrete Problems in AI Safety Part 3 (Patreon)

Published:

2017-08-11 20:53:36

Imported:

2023-12

Tags:

concrete problems early access early access video

Content

Sometimes AI can find ways to 'cheat' and get more reward than we intended by doing something unexpected.

The first proper video shot in the new studio! There's still work to do there, and I had to record this one twice because the first time I had the audio recorder settings wrong... Anyway, reward hacking is a big enough subject that it deserves more than one video, I hope you like this first one!

Files

Reward Hacking: Concrete Problems in AI Safety Part 3

Sometimes AI can find ways to 'cheat' and get more reward than we intended by doing something unexpected. The Concrete Problems in AI Safety Playlist: https://www.youtube.com/playlist?list=PLqL14ZxTTA4fEp5ltiNinNHdkPuLK4778 The Computerphile video: https://www.youtube.com/watch?v=9nktr1MgS-A The paper 'Concrete Problems in AI Safety': https://arxiv.org/pdf/1606.06565.pdf SethBling's channel: https://www.youtube.com/user/sethbling With thanks to my excellent Patreon supporters: https://www.patreon.com/robertskmiles Jordan Medina FHI's own Kyle Scott Jason Hise David Rasmussen James McCuen Richárd Nagyfi Ammar Mousali Joshua Richardson Fabian Consiglio Jonatan R Øystein Flygt Björn Mosten Michael Greve robertvanduursen The Guru Of Vision Fabrizio Pisani Alexander Hartvig Nielsen Volodymyr David Tjäder Paul Mason Ben Scanlon Julius Brash Mike Bird Peggy Youell Konstantin Shabashov Almighty Dodd DGJono Matthias Meger Scott Stevens Emilio Alvarez Benjamin Aaron Degenhart Michael Ore Robert Bridges Dmitri Afanasjev Brian Sandberg Einar Ueland Lo Rez C3POehne Stephen Paul Marcel Ward Andrew Weir Pontus Carlsson Taylor Smith Ben Archer Ivan Pochesnev Scott McCarthy Kabilan Kabilan Kabilan Kabilan Phil Philip Alexander Christopher Tendayi Mawushe Gabriel Behm Anne Kohlbrenner Jake Fish Jennifer Autumn Latham