This is an infamous example from just a year ago, that no matter how powerful AI chess engines have become of late i.e AlphaZero, Stockfish, Komodo, Houdini etc would all crush to oblivion any human player, that they are still not human.
Any human would instantly recognize that white just moves king to any white square and draw guaranteed by the 50-move rule, but comps evaluate by piece value and do not (yet) have the ability to compute beyond 50-moves ahead. So, they are still stupid in this very one area and cannot behave like humans, even now. One tiny crumb of comfort for us mere mortals, is that the almighty chess-engines of today still have a tiny kink of artificiality about them in some very specific moments, like the above study.
If you put this position on Stockfish, you immediately appear as O, O. Stockfish immediately understands that the match is a draw.
What are you saying?
The computers you're talking about are many years old ...
Stockfish understands absolutely nothing, just a basic machine haha. No human logic ever present here, just 100% math computation. Nothing more, nothing less. Computers have no vision whatsoever, never have had, never will have, it's all number crunching alone, haha
Machines only do whatever they are pre-programmed to do, haha. Nothing else!! Zero human logic ever at work here.
The main Dev may have added a feature in the algorithm to detect obvious 50-move draw stuff, if so very OK ...but this is human intervention stuff and it makes no difference to actual playing level which is maybe 99,9% above Carlsen level anyway. Only way engine evaluation ever changes is from a Dev re-programming the said engine, which was not needed anyways!!!
This position is a joke anyway, as no way to achieve in real life. Computers have zero perception of logic too, nothing exists bar the entropy that computers need to process randomness. The higher the level of entropy, the better that any computer plays Chess (or better stated how it computes the numbers game).
Entropy btw defines the ability for computers to design and designate dedicated randomness parameters within the pre-cpu algorithms (and on main RAM stuff) and all kinda RAM based too, to better perform in terms of choice management choices. Seeding randomness is preferred, but extreme high cost here (entropy wise).
High Entropy will kill at Chess, but comes at massive cost, computational-wise.
Maybe you should just talk about chess.
And maybe laugh a little less.
You have a lot of confused ideas about cybernetics.
Speaking of chess, today a trivial chess program recognizes that the position in question is 0.0.
As shown above.
That's all, man.
This topic has been archived and can no longer be replied to.