Futurology Today
  • Communities
  • Create Post
  • heart
    Support Lemmy
  • search
    Search
  • Login
  • Sign Up
LughMA to FuturologyEnglish · 1 year ago

Sycophancy to subterfuge: Investigating reward tampering in language models

www.anthropic.com

external-link
message-square
0
link
fedilink
8
external-link

Sycophancy to subterfuge: Investigating reward tampering in language models

www.anthropic.com

LughMA to FuturologyEnglish · 1 year ago
message-square
0
link
fedilink
Empirical evidence that serious misalignment can emerge from seemingly benign reward misspecification.
alert-triangle
You must log in or register to comment.

Futurology

futurology

Subscribe from Remote Instance

Create a post
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !futurology@futurology.today
Visibility: Public
globe

This community can be federated to other instances and be posted/commented in by their users.

  • 85 users / day
  • 142 users / week
  • 841 users / month
  • 3.72K users / 6 months
  • 92 local subscribers
  • 3.43K subscribers
  • 2.13K Posts
  • 12.9K Comments
  • Modlog
  • mods:
  • voidx
  • Lugh
  • Espiritdescali
  • AwesomeLowlander
  • BE: 0.19.11
  • Modlog
  • Legal
  • Instances
  • Docs
  • Code
  • join-lemmy.org