PDA

View Full Version : The Royal Institution: How Can Machines Learn Human Values? (Brian Christian)



Aragorn
12th March 2021, 03:05
DESCRIPTION


Brian explores everything that goes wrong when we build AI systems and introduces the people that fix them. Brian Christian is the author of 'The Most Human Human', which was named a Wall Street Journal bestseller, a New York Times Editors’ Choice, and a New Yorker favourite book of the year. Brian's book 'The Alignment Problem' is available here (https://geni.us/24uce).

Recent years have seen an eruption of concern about machine learning. When the systems we attempt to teach don’t do what we want or what we expect, ethical and potentially existential risks emerge.




DURATION



Presentation: 1 hour 2 minutes
Q&A: 29 minutes








https://www.youtube.com/watch?v=bqWAVNjk-cg



https://www.youtube.com/watch?v=Gfd0HraqyHI

Elen
12th March 2021, 06:45
The Royal Institution: How Can Machines Learn Human Values? (Brian Christian)

DESCRIPTION

Brian explores everything that goes wrong when we build AI systems and introduces the people that fix them. Brian Christian is the author of 'The Most Human Human', which was named a Wall Street Journal bestseller, a New York Times Editors’ Choice, and a New Yorker favourite book of the year. Brian's book 'The Alignment Problem' is available here.

Recent years have seen an eruption of concern about machine learning. When the systems we attempt to teach don’t do what we want or what we expect, ethical and potentially existential risks emerge.







This is sooo interesting, it explains why algorithms can go wrong and in favour of some, but not in favour of others. Prison systems...medicine...military and education? OMG. Thanks for sharing Aragorn. :)

Dreamtimer
13th March 2021, 02:57
Brian seems like a very interesting fellow. And quite smart. (I've gone off on a tangent looking for his poetry)

This will be a good listen.

Thank you, Aragorn.