[Barbing]

Which lab is going to stop AI from hallucinating first?

What’s taking them so long? If OpenAI was going to do it, why didn’t Anthropic do it first, or vice versa? Or if we assume collusion: shouldn’t DeepSeek have released a hallucination-free LLM to steal the US’s thunder?

(Good points on AMP and walled garden)

[tough]

the whole point of an llm is to generate "plausible sounding" tokens one after each other.

It can sadly not hold any concept of truth or reality as we humans understand it

[Barbing]

We’ll be awaiting a source a while! https://news.ycombinator.com/item?id=48873333