Comment by cheema33
10 days ago
> As soon as it starts returning to me factual, confirm-able answers consistently. Then I'll use it.
Humans don't behave this way. Yet, we still employ humans. AI is unreliable. We all agree. But that does not make it useless.
Ok, how about: Once AI is less overconfident than a median teammate, it may be worth something. It's not there yet.
I agree. Overconfidence and sycophancy is the real problem. This should be the focus of development energy. The models are already capable; now they need to be reliable.
That's why we introduced computers. If they aren't reliable anymore we can stop using them.