Best best thing is to have fucking human oversight over a machine telling us something.
Sure, having (supposedly) unbiased model running without third-party opaque interference is great, but LLM will always be statistical machines, by design. Nothing it does have value if it’s not checked by something that’s smart enough, and aside from deterministic outcomes we can express with pure logic, that requires a human.
Best thing is to use Open Source AI if people must since those are way better than the proprietary ones
Best best thing is to have fucking human oversight over a machine telling us something.
Sure, having (supposedly) unbiased model running without third-party opaque interference is great, but LLM will always be statistical machines, by design. Nothing it does have value if it’s not checked by something that’s smart enough, and aside from deterministic outcomes we can express with pure logic, that requires a human.