12

Two-faced AI language models learn to hide deception - ‘Sleeper agents’ seem benign during testing but behave differently once deployed. And methods to stop them aren’t working.(www.nature.com)

posted
by
Avatar
Lugh@futurology.todayM
in
Avatar
futurology@futurology.today
9 comments
hidereport
Sort:
HotTopControversialNewOld
You are viewing a single thread.
View all comments
Avatar
mateomaui@reddthat.com
1 point

Just… don’t hook it up to the defense grid.

permalink
report
reply
Avatar
Possibly linux@lemmy.zip
1 point

Sorry, to late for that

permalink
report
parent
reply
Avatar
mateomaui@reddthat.com
2 points

Alright, I’ll be out back digging the bomb shelter.

permalink
report
parent
reply
Avatar
Possibly linux@lemmy.zip
1 point
*

Its too late for that honestly

permalink
report
parent
reply
Show more comments

Futurology

!futurology@futurology.today

Create post

Community stats

  • 3.2K

    Monthly active users

  • 1.1K

    Posts

  • 7.7K

    Comments

Community moderators