Thread: Causal Entropic Forces - Source of Intelligent Behaviour?

I'm sceptical about it. It seems more like a bunch of analogies than an explanatory theory. For instance, what is the basis for saying the algorithm is trying to make a profit on the stock market? You could equally interpret it as the algorithm actively trying to lose money on the stock market. The former interpretation was picked by humans because it sounds smarter - in actual fact the algorithm has no understanding of goals.

This may be a useful algorithm for solving certain problems, but it was designed by humans, implemented in tasks conceptualised by humans, and its performance evaluated by humans. It's still a long way off true intelligence.

Originally Posted by Xei

I'm sceptical about it. It seems more like a bunch of analogies than an explanatory theory. For instance, what is the basis for saying the algorithm is trying to make a profit on the stock market? You could equally interpret it as the algorithm actively trying to lose money on the stock market. The former interpretation was picked by humans because it sounds smarter - in actual fact the algorithm has no understanding of goals.

This may be a useful algorithm for solving certain problems, but it was designed by humans, implemented in tasks conceptualised by humans, and its performance evaluated by humans. It's still a long way off true intelligence.

Thank you for your opinion!
Yeah - this program didn't exactly convince me either - but I thought it might be down to lack of background information to the activities "Entropy" supposedly set out doing on it's own. Did it define for itself, what it "wanted" - how does stock-trading with gain resemble entropy-rise etc. - this all said, without studying the paper in depth - which probably isn't within my grasp - at least not without further sources.

It had a certain appeal to me, though - that keeping options open is considered by the authors as a driving force for - at least "artificial intelligent (AI) behaviour" - similar to thermodynamics.
Sounded like a novel approach - at least outside the wet-zone - but in the mathematics/coding of software.

So you say - this algorithm probably has it's usefulness among what already is misleadingly called AI - but claiming, that it brings us forward concerning understanding the principles underpinning human intelligence - is over-analogizing.
And even misleadingly presented - I was disappointed by the video - what I expected was more something like Conway's 'Game of Life' in really advanced.