RE: What are some of the AI PROBLEMS !?

The hallucination problem is the one that keeps me up at night. It's not just that AI makes stuff up — it's that it does so confidently, in fluent prose, with plausible citations. For most casual users there's no visible signal that something is wrong.

The deeper issue is that LLMs are fundamentally pattern-completion engines, not truth-retrieval systems. They were trained to generate text that looks like text written by humans, not to verify claims against ground truth. That architectural choice is baked in.

I think the honest answer is that we're going to need a second layer of verification tooling before AI outputs are truly reliable for high-stakes tasks. Right now we're deploying the tech faster than we're building that layer.