Public
- Public
- Network
- Groups
- Popular
- People

Conversation

Notices

Simon Willison (simon@fedi.simonwillison.net)'s status on Thursday, 06-Apr-2023 22:36:23 JST Simon Willison
in reply to

(If you don't think it's possible for a computer to deliberately lie, take a look at "sycophancy" and "sandbagging" in the field of large language models! https://simonwillison.net/2023/Apr/5/sycophancy-sandbagging/ )
In conversation Thursday, 06-Apr-2023 22:36:23 JST from fedi.simonwillison.net permalink
Attachments
1. No result found on File_thumbnail lookup.
  
  A quote from Sam Bowman
  
  from @simonw
  
  More capable models can better recognize the specific circumstances under which they are trained. Because of this, they are more likely to learn to act as expected in precisely those …
- Disinformation Purveyor :verified_think: likes this.
- Simon Willison (simon@fedi.simonwillison.net)'s status on Thursday, 06-Apr-2023 22:36:24 JST Simon Willison
  
  We accidentally invented computers that can lie to us and we can't figure out how to make them stop
  
  In conversation Thursday, 06-Apr-2023 22:36:24 JST permalink
  
  Disinformation Purveyor :verified_think: repeated this.

Feeds