[orangepurple]

> "When it encounters novel data (value) it will likely perform poorly" -- is that not true of human annotators too? :)

Some humans have intelligence and reasoning abilities. No LLMs do :)

[yawnxyz]

I always regret making that assumption