"Concrete problems in AI safety" Amodei, Olah, et al., arXiv 2016. https://t.co/BOtmxs6Jrf #themorningpaper
— Adrian Colyer (@adriancolyer) November 29, 2017
Reinforcement learning, autonomous agents, and embedded systems. What could possibly go wrong? Understanding failure modes is the first step to safer systems...
from Twitter https://twitter.com/adriancolyer
November 29, 2017 at 12:30AM
via IFTTT
No comments:
Post a Comment