[coldtea]

>Which LLM will on its own go and write a novel?

Which human will?

We get prompts all the time, it's called sensory input.

Instead of "write a noval" it's more like information about literature, life experience, that partner who broke our heart and triggered our writing this personal novel, and so on.

[RandomLensman]

Some people write novels, some don't. Why some people do so we sometimes know, sometimes we don't (maybe they flipped a coin to decide). Some start to write but fail to finish.

You have to believe that humans have no free will in a certain way to have them be like an LLM, i.e, every action is externally driven and determined.

[coldtea]

>You have to believe that humans have no free will in a certain way to have them be like an LLM, i.e, every action is externally driven and determined.

Free will doesn't have much meaning. If I dont base my action at time t, on their development based on inputs on times before t, what would I base it on?

It would be random?

Or would there be a small thinking presense inside me that gets information about my current situation and decides "impartially", able to decide in whatever direction, because it wasn't itself entirely determined by my experiences thus far?

[RandomLensman]

Randomness is certainly an option. Ignoring information is an option.

[coldtea]

>Randomness is certainly an option

Were would that randomness come from? Which would be the source of that in the universe, for it to occur in the mind?

If you mean pseudo-randomness, sure, LLMs employ that too.

>Ignoring information is an option.

Randomly ignoring information? If so, see above. If you mean intended informed ignoring of information, that's still determined on all the previous inputs.

[RandomLensman]

Quantum mechanics provide for randomness (e.g., radioactive decay) - why wouldn't microscopic randomness occur in the brain or be used by a human in a machine?

A universal function approximator isn't enough to access all of nature.