this post was submitted on 27 Jan 2024
13 points (78.3% liked)

Futurology

1776 readers
210 users here now

founded 1 year ago
MODERATORS
voidx@futurology.today
Lugh@futurology.today
Espiritdescali@futurology.today
AwesomeLowlander@futurology.today
13
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)
submitted 9 months ago by Lugh@futurology.today to c/futurology@futurology.today
9 comments fedilink hide all child comments
you are viewing a single comment's thread
view the rest of the comments
[–] Daxtron2@startrek.website 1 points 9 months ago (1 children)

LLM trained on adversarial data, behaves in an adversarial way. Shocking

[–] CanadaPlus@futurology.today 0 points 9 months ago

Yeah. For reference, they made a model with a back door, and then trained it to not respond in a backdoored way when it hasn't been triggered. It worked but it didn't effect the back door much, and that means that it technically was acting more differently - and therefore deceptively - when not triggered.

Interesting maybe, but I don't personally find it surprising, given how flexible these things are in general.