neuracnuA research scientist at Anthropic has been using LLMs to black hat software and he's spooked
tomiantRemember when AI outclassed the best Go player in the world?
That was in 2016.
fallaciousBasisAs I recall Go players have adapted and have found ways to induce hallucinations and beat the machine, some using other AI. Others have adopted “adversarial strategies.”
They say it’s comprehensible enough that a human “expert” can do it without algorithmic assistance.
Adversarial Policies Beat Superhuman Go AIs
We attack the state-of-the-art Go-playing AI system KataGo by training adversarial policies against it, achieving a >97% win rate against KataGo running at superhuman settings. Our adversaries do not win by playing Go well. Instead, they trick KataGo into making serious blunders. Our attack transfers zero-shot to other superhuman Go-playing AIs, and is comprehensible to the extent that human experts can implement it without algorithmic assistance to consistently beat superhuman AIs. The core vulnerability uncovered by our attack persists even in KataGo agents adversarially trained to defend against our attack. Our results demonstrate that even superhuman AI systems may harbor surprising failure modes. Example games are available https://goattack.far.ai/.