I wonder what would happen to the economy if we had fully fact-based rational CEOs. Would it grow, shrink or stabilize?
Ai ceo’s would do great for a while. Then there’d be a single hallucination that would cause the company to fail and become insolvent.
So like human CEOs…
If the data is organic and not synthetic, might the hallucination be avoided? Correct me if I’m wrong, but isn’t the basis of the hallucinations down to the compilation of synthetic data sets and the LLMs go off the rails because of that?
No, it’s a direct consequence of the structure of an LLM. You may have heard them called “fancy autocorrect” and that’s because in a very real way, that’s what an LLM does: based on the text in its conversation (the user’s, its own, the system prompt, search results, etc) it picks one of the most likely next words. Then starts over with that new word included.
“Most likely” in this case is based on the weights from training, but the important thing is that it’s inherently statistical in nature, there’s no enforcing of hard facts possible, and even limiting it to only the very most likely next word each time causes unnatural text and problems such as loops - while still not being 100% accurate.
