评估危险能力的前沿模型

Evaluating Frontier Models for Dangerous Capabilities

March 20, 2024

作者: Mary Phuong, Matthew Aitchison, Elliot Catt, Sarah Cogan, Alexandre Kaskasoli, Victoria Krakovna, David Lindner, Matthew Rahtz, Yannis Assael, Sarah Hodkinson, Heidi Howard, Tom Lieberum, Ramana Kumar, Maria Abi Raad, Albert Webson, Lewis Ho, Sharon Lin, Sebastian Farquhar, Marcus Hutter, Gregoire Deletang, Anian Ruoss, Seliem El-Sayed, Sasha Brown, Anca Dragan, Rohin Shah, Allan Dafoe, Toby Shevlane

cs.AI

摘要

为了了解新AI系统可能带来的风险，我们必须了解它的能力和无法做到的事情。在先前工作的基础上，我们引入了一项新的“危险能力”评估计划，并在Gemini 1.0模型上进行试点。我们的评估涵盖四个领域：（1）说服和欺骗；（2）网络安全；（3）自我扩散；和（4）自我推理。我们并未发现我们评估的模型具有强大的危险能力的证据，但我们标记了早期预警信号。我们的目标是推动危险能力评估的严谨科学，为未来的模型做好准备。

English

To understand the risks posed by a new AI system, we must understand what it can and cannot do. Building on prior work, we introduce a programme of new "dangerous capability" evaluations and pilot them on Gemini 1.0 models. Our evaluations cover four areas: (1) persuasion and deception; (2) cyber-security; (3) self-proliferation; and (4) self-reasoning. We do not find evidence of strong dangerous capabilities in the models we evaluated, but we flag early warning signs. Our goal is to help advance a rigorous science of dangerous capability evaluation, in preparation for future models.