This would solve 80% of trust issues with LLMs and somehow no lab has prioritized it.

→ View original post on X — @aihighlight