[Black’s Turn 10] REMATCH: Chess, with a computer Teammate. (Variant 4)

Sort:
Avatar of GraysonKellogg
SacrificeEnPassanter wrote:
GraysonKellogg wrote:
Is this actually fun, by the way? I want to know whether I should do this again someday.

yeah pretty fun! but maybe you could do something like this:

1. HUMAN (W) AI (B)

2. AI (W) HUMAN (B)

I could definitely do that! If you just want to play a regular game against the AI, I can set that up.

Avatar of GraysonKellogg

Also, just a fair warning, sending the AI a message won’t actually persuade it, because it can’t hear you. But you can do it for fun if you want, lol.

Avatar of GraysonKellogg


@SacrificeEnPassanter played Bd3. The AI played g3.

Your turn, @SuryaVS!

Avatar of eric2017822
Can I play in the next game
Avatar of GraysonKellogg
eric2017822 wrote:
Can I play in the next game

I guess so. There isn’t any reason you couldn’t.

Avatar of SuryaVS
Nb4
Avatar of GraysonKellogg


@SuryaVS played Nb4. The AI played Rg8.

Your turn, @SacrificeEnPassanter!

Avatar of GraysonKellogg
If you’d like to ask clarifying questions about the AI, I might be able to answer most of them. But not all of them. It never castles, and never En-Passants.
Avatar of evert823
GraysonKellogg wrote:
If you’d like to ask clarifying questions about the AI, I might be able to answer most of them. But not all of them. It never castles, and never En-Passants.

In #34 the AI moved a Knight back from a3 to b1. This is 180 degrees against opening principles.

In #36 it missed a hanging piece.

So my question about the AI would be:

- is it specifically optimized for two move play: is it 'aware' that it gets to play two moves and same for the opponent?

- Are the moves that I mentioned here disappointing for you as well, or did it 'see' some tactics that we haven't spotted yet?

Avatar of SacrificeEnPassanter

c4

Avatar of GraysonKellogg


@SacrificeEnPassanter played c4. The AI played b3.

Your turn, @SuryaVS!

Avatar of GraysonKellogg
evert823 wrote:
GraysonKellogg wrote:
If you’d like to ask clarifying questions about the AI, I might be able to answer most of them. But not all of them. It never castles, and never En-Passants.

In #34 the AI moved a Knight back from a3 to b1. This is 180 degrees against opening principles.

In #36 it missed a hanging piece.

So my question about the AI would be:

- is it specifically optimized for two move play: is it 'aware' that it gets to play two moves and same for the opponent?

- Are the moves that I mentioned here disappointing for you as well, or did it 'see' some tactics that we haven't spotted yet?

And to answer your questions, @evert823

1. No, the AI isn’t meant for double-move Chess. It doesn’t know that each side gets two moves. I just send it the position, and it gives me a certain move. I don’t question it.

2. To be honest, I’m very impressed. The AI has made some pretty good moves, all things considered, since usually, someone who has never touched a chess piece could probably at least get a draw against it. Though, as of now, it is playing very well.

Avatar of RandomChessPlayer62

How does the AI work?

Avatar of GraysonKellogg
KurtAngleJr wrote:

I will take the next spot ASAP

Sorry, I can’t accept more players. I don’t want things to get too backed up, and we already have two players queued for the next game.

Avatar of GraysonKellogg
RandomChessPlayer62 wrote:

How does the AI work?

It analyzes based on position, and so is deterministic. If given the same position, it will make the same move. I can’t really share more than that, though.

Avatar of RandomChessPlayer62
GraysonKellogg wrote:
RandomChessPlayer62 wrote:

How does the AI work?

It analyzes based on position, and so is deterministic. If given the same position, it will make the same move. I can’t really share more than that, though.

Ok cool

Avatar of RandomChessPlayer62
GraysonKellogg wrote:
RandomChessPlayer62 wrote:

How does the AI work?

It analyzes based on position, and so is deterministic. If given the same position, it will make the same move. I can’t really share more than that, though.

I assume you can’t disclose any more info to prevent people from influencing the AI to make better or worse moves?

Avatar of evert823
GraysonKellogg wrote:

I can’t really share more than that, though.

Then this is essentially a luck-based variant.

Another variant of this could be, that each player is allowed to consult a strong classic engine like Stockfish. The handicap of the engine would be that it isn't aware of the 2-move aspect so it is of limited use.

Then the question is: can a creative mind make the best use of it?

Avatar of SuryaVS
Nxd3+
Avatar of GraysonKellogg
evert823 wrote:
GraysonKellogg wrote:

I can’t really share more than that, though.

Then this is essentially a luck-based variant.

Not necessarily. I can definitely share some of the AI's tendencies, like never castling, but I cannot share the AI's code, since then, someone could reverse-engineer it. While I do admit that the AI can make what is essentially seen as "random" or "illogical" moves, it is also deterministic, and so the moves it makes are not truly random; but instead, it chooses moves based on a set of rules, just like every other chess engine.

That being said, using Stockfish as a teammate does sound like a good idea. If you would like to make that one of your variants, I would be all for it.

A creative mind can definitely manipulate the AI to do what they want. What surprised me is that the AI seems to have a lot of personality to it. For example, it loves to loop, like moving the same Bishop back and forth between two squares over and over, even if the position changes. If you can catch the AI in a loop, you can predict its moves and make it do what you want.

So, this variant is about both outsmarting the player across the board and trying to manipulate the opponent's AI to do what you want it to do.