[The28thDuck]

I feel intuitively this makes sense. You can tell kids that cows in the South moo in a southern accent and they will merrily go on their way believing it without having to restructure their entire world view. It goes with the problem of “understanding” vs parroting.

Human-centric example but you get the point.

[dTal]

Kids, but not adults. What's the difference? A more interconnected world model with underlying structure. LLMs have such structure as well, proportional to how well they're trained. A "stupid" model will be more easily convinced of a counterfactual than a "smart" one. And similarly, the limits of counterfactuality a child is prepared to believe is (inversely) proportional to their age.

[chrisnight]

There is a certain balance in this act though. Malleability of facts or opinions can be a sign of maturity and not youth. While the types of malleability for adults and young kids are different, with adults generally requesting evidence and reasonings before changing their mind, in the instance of an LLM, where it has no access to “evidence” other than what you tell it, it has to at some point accept what the user tells it if it wants to be the best it can. Otherwise you’ll get Bing Chat again with the “I don’t believe you.” responses to pure facts.

[davidguetta]

lol how many adults believe the earth is flat >< ?