AI safety via debate

8 years ago 12
Add to circle
We’re proposing an AI safety technique which trains agents to debate topics with one another, using a human to judge who wins.
Read Entire Article
  1. Homepage
  2. Technology
  3. AI safety via debate

Related

At market open, Korean chipmakers' stocks plunged, with Samsung Electronics dropping 9%+, SK Hynix down 11%+, and the Korea Composite Stock Price Index down 8%+ (Youkyung Lee/Bloomberg)

At market open, Korean chipmakers' stocks plunged, with Sams...

8 minutes ago 0
You Can Finally Trust All Your Money to X, the Everything App

You Can Finally Trust All Your Money to X, the Everything Ap...

27 minutes ago 0
ChatGPT appears to block direct requests to copy an author's style, instead offering to capture the overall "feeling", amid its legal battles with book authors (Kyle Orland/Ars Technica)

ChatGPT appears to block direct requests to copy an author's...

33 minutes ago 0