[zuzululu]

I didn't get that impression at all, I think there is genuine write off or outdated view of LLMs here. I also see lot of people sharing issues raised by LLMs at tasks that a human would also fail to produce an accurate result.

So in this thread there is a mix of genuine edges that LLMs need harness and guidance with tasks that when given to a human will not perform well in that LLM is supposed to suddenly solve.

Like the thread above about financial compliance, without knowing specifics it can be very vague in language and confusing unless you apply to precedents and exact scenarios that can give you a range for what is acceptable/unacceptable. Mythos or any LLM isn't going to magically figure these edges out for you because a human would also struggle at such task.

My advice is don't let what you read here including my comment dictate your own decisions, but apply the same common sense, apply it in ways that it can help you and figure out when to use determinism vs LLM in the context of your jobs.

These lazy comments that simply try to paint a black and white categorization of AI/LLM are just noise.