On: Unsupervised sentiment neuron

[gallerdude]

A weird thought: at some point AI short stories may be far more profound than our own.

[beaconstudios → gallerdude]

at the moment, AI short stories are derivative, so it's unlikely. They may well be better than the average, if trained on highly regarded works, but they're not completely novel.

[happycube → beaconstudios]

At the moment RNN's can't remember context, so they can make stuff that looks correct, but only on the surface.

I think that'll change, eventually...

[visarga → happycube]

We need some kind of hierarchical approach, and/or memory.

[RobertoG → beaconstudios]

As they say: “When you take stuff from one writer it’s plagiarism, but when you take from many writers it’s called research.”

[haddr → beaconstudios]

But given the whole corpus of all human-written texts, couldn't they be as creative as us?

[jsmthrowaway → haddr]

It depends on your philosophy on what defines creativity, and whether all creativity is derivative of observing others.

[haddr → jsmthrowaway]

I don't have one, and I guess that's still somehow unsolved problem. But there are some works on creativity and AI. This is definitely an interesting space to observe.