I SAVED HUMANITY
Everybody says that Chatgpt is impossible to beat,so I took that to the test
I played against Chatgpt and In move 29,it played an illegal move by counterchecking without blocking my check so I captured the king and won the game
I’m literally 9 years old with an elo of 582
congratulate me pls
if you want to be kind you can sub to my youtube channel,it’s called @BrilliantAdam14TalCAG
This is so wrong; you didn't actually win against Chatgpt, you never pointed out that it made an illegal move. Imagine you made an illegal move when you were playing with me and hung your queen - I would definitely point it out! You should continue that game by asking it to make another move. Capturing the king is not the same as checkmating.
Well guys to be fair ChatGPT can summon chess peices, a thing I doubt you can
good thing that ChatGPT cant summon in real life
be nice he's a kid you bully
ChatGPT is a chat bot, not a chess engine.
When you ask it to play chess, it will search in its database of texts, that includes chess publications, articles, books, magazines, and will try to find, not a chess move, but a likely follow up to the ongoing conversation it's having with you.
ChatGPT doesn't know what chess is, doesn't even know what a game is. That software just can't play chess, because it does posses no code lines for it.
If, and that's some of a big if, some day, it happens ChatGPT can actually play a whole game of chess (that would not be a game he has find in its database), it would mean a chess engine has been added to it, along with instruction to use that chess engine whenever someone ask it to play a game of chess.
Asking ChatGPT to play chess, is about the same than to ask your dishwasher to go sail on the seven seas.
You'll find tons of videos on Youtube explaining in all details how ChatGPT and other so called "AI" do function. You'd better forget the idea it's be anything close to an "intelligence", because it's not, and never will be.
The best shot you have got, to get ChatGPT to play legal moves against you, is to re-type the whole of the moves you already both played, every time it's your turn to play.
Like, for an example:
You: 1.e4
ChatGPT: 1...-e5
You: 1.e4 - e5 2. f4
etc.
Eventually, ChatGPT will fail to find a pre-existing game that fits your ongoing game.
It will then pick some "credible" move in some other pre-existing game, and post it to you, with chances it will be an illegal move.
That's how it works.
In other words, ChatGPT does not "think". It does a look up in a huge database of texts, Google like, and post what it finds in texts written by others (most of the time, humans) that seem to fit what has been said before in your conversation.
I just beat it too. It played the Mikenas Defence against me, I was expecting something more solid like , , 1.Nf6, . I carried on with 2.Bg5 and I couldn't believe what it did it just took the d4 pawn with it's c6 Knight. From there it was all down hill losing the Queenside Rook the a pawn as well. Then eventually stalling I'm guessing that means it's resigned ? Here's a screen shot of the final position.