> Imagine if LLMs were right 99% of the time
replyThis is an hypothetical that's not here yet.
Of course if LLMs had human-level accuracy they would be ideal.
That's an extreme to demonstrate perfection is not required. I then ask you to consider where the break-even point is.
reply