Comment by vbezhenar
9 days ago
Why is that a problem, anyway? If machines can play chess better than any human, it is reasonable to assume that they can write articles better than many humans. What's wrong with Internet filled with good content generated by AI?
AI can't create good (meaning truthful) content. It's literally impossible for LLMs to be hallucination free. It's just not how they work.
The problem is going to get worse as hallucinations are used as training data because even the AI companies can't tell the difference between AI content and human content.