Human centered AI

https://www.facebook.com/share/p/14dAWQy7Dkh/

Trustworthy AI may not require omniscience.

https://www.facebook.com/share/p/1G1QweNUmf/

與其等 AI 變聰明，不如動手防幻覺

https://www.facebook.com/share/p/1Ak5AnQx21/

A Single Neuron Is Sufficient to Bypass Safety Alignment in Large Language Models

https://arxiv.org/pdf/2605.08513

https://huggingface.co/papers/2605.08513

https://www.facebook.com/share/p/1ELbYqt3UR/