Can Humans Still Beat Computers in Any Games?

[Dr Wu]

Hi

Aside from one form of poker (whose name escapes me at present) what games of any kind can humans still stand a reasonable chance of beating computers?

Many thanks.

 

[jedishrfu]

I found one at Wikipedia:

https://en.wikipedia.org/wiki/Arimaa

It has rules a 4 yr old can understand but for the computer its a lot of processing to find a good move.

And here's a list of others:

http://senseis.xmp.net/?OtherGamesConsideredUnprogrammable

So your challenge is to select a game and get your computer to play it.

 

[Svein]

Aside from one form of poker (whose name escapes me at present) what games of any kind can humans still stand a reasonable chance of beating computers?

You do not compete against computers. You compete against an application, designed by someone who has found an applicable algorithm for handling the game in question.

So, in reality you are competing against another human being through a computer medium.

 

[GTOM]

Go is very hard for compus.

 

[Hornbein]

Hi

Aside from one form of poker (whose name escapes me at present) what games of any kind can humans still stand a reasonable chance of beating computers?

Many thanks.

Charades. Humans rule Charades.

 

[Chronos]

You can forget rodeo poker, the computer never blinks or does anything provocative, like running when the bull approaches.

 

[Czcibor]

Grand strategy wargames?

I'm not sure whether there are any formal studies for this, but assuming that game have interesting complexity, which is intuitive to human, should be a nightmare for a programmer.

 

[Hornbein]

You do not compete against computers. You compete against an application, designed by someone who has found an applicable algorithm for handling the game in question.

So, in reality you are competing against another human being through a computer medium.

With the advent of machine learning, this is no longer the case.

 

[Hornbein]

Hi

Aside from one form of poker (whose name escapes me at present) what games of any kind can humans still stand a reasonable chance of beating computers?

Many thanks.

I have no doubt that computers will be better than humans at poker before long, if not already.

 

[Svein]

With the advent of machine learning, this is no longer the case.

I expect that "machine learning" is programmed by a human based on an algorithm created by a human.

 

[Dr Wu]

What would be the case if computers started creating their own algorithms?

 

[Greg Bernhardt]

How about rock paper scissors?

 

[Hornbein]

How about rock paper scissors?

The New York Times website had a software rock paper scissors software. I gave it a whirl and was soon down twelve. Dang, this thing knows my tricks.

So, I would come up with a move. I assumed the computer could predict my move then choose the move to beat it. So I made the move to beat THAT move. It worked. I won twelve in a row to get even, then quit.

But I'm sure a really serious rock paper scissors program would figure out what I was up to and react.

 

[Hornbein]

What would be the case if computers started creating their own algorithms?

That's pretty much what machine learning is. Sure, machine learning is directed by an algorithm, but no one knows what the computer is going to do once it has learned. It is difficult to understand why it makes the choices it does.

Two computers given the same training set but in a different order will learn differently.

 

[phyzguy]

I expect that "machine learning" is programmed by a human based on an algorithm created by a human.

I don't think this is the case. True machine learning, also called "deep learning" involves setting up a neural network conceptually similar to the ones in a human brain, and then training the network with a training database. In this case, even the person who designed the network doesn't know exactly what the computer is doing or why it makes the decisions that it does. It is not algorithmic in the sense that we usually think of it.

 

[Macroego]

Good Call phyzguy "True machine learning, also called "deep learning" involves setting up a neural network conceptually similar to the ones in a human brain, and then training the network with a training database."

Looks like Deep Learning Machines has been created DeepMind.com by Google http://deepmind.com/alpha-go.html . AND

It already beaten European Top Player (5-0) of Intuitive game GO- https://en.wikipedia.org/wiki/Go_(game) and set to play world top player in March2016 http://www.nature.com/nature/journal/v529/n7587/full/nature16961.html

It appears to be true machine learning because none of the programmers are more than amature GO players.

 

[Hornbein]

Good Call phyzguy "True machine learning, also called "deep learning" involves setting up a neural network conceptually similar to the ones in a human brain, and then training the network with a training database."

Looks like Deep Learning Machines has been created DeepMind.com by Google http://deepmind.com/alpha-go.html . AND

It already beaten European Top Player (5-0) of Intuitive game GO- https://en.wikipedia.org/wiki/Go_(game) and set to play world top player in March2016 http://www.nature.com/nature/journal/v529/n7587/full/nature16961.html

It appears to be true machine learning because none of the programmers are more than amature GO players.

Amazing! Artificial intelligence has arrived, suddenly and unexpectedly. I think that this is even more amazing. Computers learned to play computer games from raw pixel feeds. It's hard to believe.

 

[Boolean Boogey]

I don't think anything concerns me more than AI backed by quantum computing.

 

[fluidistic]

Good Call phyzguy "True machine learning, also called "deep learning" involves setting up a neural network conceptually similar to the ones in a human brain, and then training the network with a training database."

Looks like Deep Learning Machines has been created DeepMind.com by Google http://deepmind.com/alpha-go.html . AND

It already beaten European Top Player (5-0) of Intuitive game GO- https://en.wikipedia.org/wiki/Go_(game) and set to play world top player in March2016 http://www.nature.com/nature/journal/v529/n7587/full/nature16961.html

It appears to be true machine learning because none of the programmers are more than amature GO players.

The last sentence is wrong. Aja Huang which worked (and maybe still work, I don't know) for DeepMind is a strong amateur go player ranked 6 dan.