Hacker News new | ask | show | jobs
by edward_rolf 3314 days ago
You both make great points.

Can we find this conundrum in other domains, say, neural networks? Can a machine become creative when we teach it to adhere to common knowledge?

3 comments

Neural networks also need to balance exploration and exploitation. You will find it in the form of:

+ generalization / overfitting

+ cross-domain knowledge transfer

In a more embodied approach it is free energy minimization as advocated by Friston. Consider us divided by the world through a Markov blanket, or even better consider our actions and perceptions divided by one. How do we continuously surprise ourselves without getting mad?

A guy who would say it's the same thing is probably Polani with empowerment: https://arxiv.org/abs/1310.1863. The rational choice is to navigate to that location in state space where you have most decisions.

I think curiosity-based research is quite interesting from the perspective of rationality and creativity.

Personally, I see no reason why creative machines couldn't be built.

Another thought on the local/less local/global maxima problem: it seems possible that, taking a very abstract view of things, the whole reason we developed the sort of thought which distinguishes humans from other animals is to address the escaping local-maxima problems (summed up by the situation where you're in a maze, but have to go 'backward' in some sense in order to get out). While discursive thought may be worse than analogical thought at this sort of thing, our lower 'animal brain' functionality is even more geared to dealing with local maxima/minima.

Well, what you're describing falls exactly into the second category of exploration that 'westoncb was outlining. Only one way to find out!