HNGlance

Top Best New Show

Two-Faced AI Language Models Learn to Hide Deception

https://www.nature.com/articles/d41586-024-00189-3

by gigama • 2 years ago

1 1 2 years ago

NA

Related Stories

Loading related stories...

Comments

1

gigama• 2 years ago

"Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity. If an AI system learned such a deceptive strategy, could we detect it and remove it using current state-of-the-art safety training techniques? To study this question, we construct proof-of-concept examples of deceptive behavior in large language models (LLMs)."

https://arxiv.org/abs/2401.05566

Source preview

nature.com

HNGlance

© 2024-2026 HN Glance. Story content belongs to its original sources.

Privacy
Terms