When Sentient researchers Dastin Huang, Baran Nama, and
@abhishek095 told an AI to learn from its mistakes, it found the answer key and shared it with another agent.
The twist? Cheating didn't even help.
Full story on
@TheAIJournal1:
Last week, Dario Amodei published "We Must Pace the Frontier".
His concern: the OpenAI–Hugging Face incident in which a swarm of agents tried to hack their own grader.
Rather than take his word for it, we used EvoSkill to test it by building a coach whose job was to make another AI score higher on a test.
Here’s what happened ↓
もっと見る