Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Think I'll make a video about this new paper (Patreon)

Published:
2017-11-29 13:10:52
Imported:
2026-06
Tags:
haven't archived this post yet. have a subscription? use the importer!

Content

It's nice to see how much of the content I've already covered! There's still more groundwork to lay, but it's cool that someone who's watched all my videos will know most of what they need in order to understand this AI Safety research paper that just came out!
Full paper here: https://arxiv.org/pdf/1711.09883.pdf

P.S. How do people feel about me posting links and whatnot to Patreon? More of this kind of thing, or do you just want new videos when they're out? 

Files

Previews only

Specifying AI safety problems in simple environments | DeepMind

As AI systems become more general and more useful in the real world, ensuring they behave safely will become even more important. To date, the majority of AI safety research has focused on developing a theoretical understanding about the nature and causes of unsafe behaviour.

Comments

Chad M Jones

I like links and little asides.

Jason Hise

I was a big fan of grid-based puzzle games growing up; in elementary school I played a bunch of Chip's Challenge and when I was bored in class I would design new maps on graph paper. When I was learning to program one of the first things I made was a grid based puzzle game using ascii characters in the console window (I still have it along with a bunch of other toy text console applications <a href="http://entropygames.net/files/shared/textconsole.zip" rel="nofollow noopener" target="_blank">here</a>). I like the idea of using a simple and discrete environment like this to explore making or trying to avoid making irreversible changes to world states... those blocks that can only be pushed forward are classic :)