Everyone hop on chat gpt to build it's AI learning and spam "Offer to draw championship vs Stockfish" so that it actually happens and Gotham will lose his mind
the thing I love about this guy is that there are NO cuts in the video, bro has an infinitely deep bag of vocabulary a doesn't stutter once. Respect ☝️
I don't know why I enjoy these so much. I understand the bots aren't designed for any of the tasks required to play chess, and yet every illegal move just fills me with joy and child-like wonder.
Honestly, I love this brainrot "low quality chess" content. I was a bit sad when the last AI chatbot chess video was months ago and thought you'd moved on, and understandably so. I'm so pumped for this becoming an entire series!
@puffaddiqt Ahh no no, like @hyrummm said, I'm talking about the Chatbot's chess quality - it does also happen to be among the more controversial content Levy puts out, but it's like junk food, I can't help myself!
I remember when Gemini was first released in beta when it was named Bard and no matter how much I primed it and shoved games into it for it to study, it would always brick itself and start cheating by move 5 and then forget that we were even playing chess. It wouldn't even use openings while just lying that they were. It just like to move pieces one square at a time until it would cheat and just promote pawns on the home square. It's always loved to spew utter nonsense while trying to analyze positions. It's fun to see how far it's come from beta (this game is genuinely a massive improvement over beta) but still see that it loves to hallucinate some utter bullshit. It sure as hell knows how to threaten pieces but it doesn't understand anything about how to take them.
How a typical gotham video starts: *breaths in* ladies and gentleman .... How it ends: AND THE GEMINI HANGS A QUEEN CAPTURING ITS OWN PAWN AND THE NEXT MOVE IT SACRIFICES THE ROOOOOK
yea but when you are playing against a cheater even if they do self destruct every now and then they can just say nuh uh this is a checkmate or yes i did have a queen here next to your king it was just hiding dude@colonel__corn
8:16 What's crazy is that this position was reached in 5 games on the Lichess database, including one between two 2500-rated players (one of whom was a GM). The move order was different, but still... I think the GM lost the Bishop on a4 due to (essentially) a pre-move.
if you recall correctly, this championship has magnus (stockfish) Carlsen and hikaru (chatgpt) nakamura on opposite ends of the bracket as mentioned in last video
I honestly hope this becomes a yearly tournament. I'd love to see if the ai's actually improve at all between this January and next January, especially if any of the developers of them see this series.
Me ~4 min into the video: I didn't come here for main line play! I came here for MONSTER REBORN pawns moving laterally to capture queens! Me ~11 min into the video: That's more like it! :)
I'd be willing to bet that the other main reason the professor had that policy was that it's much easier to look at "I don't know" and give partial credit than to try to decipher the answers of a student who only half-knows how to do the problem and figure out how much partial credit to award. I'm sure it saves that professor a lot of time. Source: Am professor
@beccabeKAW That professor would hate me lmao, I always have the attitude of "a blank answer is always wrong, a BS answer can be partially right accidentally"
It's because the website isn't set up correctly. It's in the custom position setting, that allows you to place pieces wherever, instead of in a game. Stockfish doesn't understand because it assumes the whole time that it is white's move.
@@BjarkeEbert That poor bot is just paranoid of his pieces getting sucked into the black hole and his opponent using magic to summon new pieces. Sad lyf of stockfish
Gemini: This move is a bit of a surprise, as it’s not a common move in the Berlin Defense. It does put some pressure on the c4 pawn but it also gives you some space on the queen side. Levy: None of that is accurate. Not even close.
It is not a surprise, it is not the Berlin Defense, it's the Spanish Ruy López, it is not even close to c4, it does not give you any space and it is not even in the queen side lmao
It feels so strange the AI would come up with that SO out of the blue Like it was trying to threaten Levy who it assumed to be its opponent to resign from the game before it'd lose
Although the SATs are actually graded more or less the same way. If two people taking the SATs know half of the answers and one leaves the other half blank while the other guesses and gets about 1/4 of their guesses right (assume multiple choice with 4 choices), the one that left half the questions blank will get a higher score. The reasoning is a little different - it's because the purpose of the exams is to test what you actually know, not your luck at picking random numbers, so the test is designed so that people who only answer the questions they know will usually do better than people who try to fake it by guessing randomly.
Honestly seeing absolutely horrible chess moves is just as helpful in me learning chess as seeing brilliant moves. Even more so because it's easier to understand why a bad move is bad and helps you avoid some of these pitfalls yourself as a beginner.
Seems like the best explanation for how these bots play is that it's like playing blindfold chess, but you also have extreme dementia so you don't know what the board looks like so you just say random chess moves.
The bots don't have actual intelligence, all these bots do is say things that seem popular to say after certain words. So if you tell it you're playing chess and you played E4, it's going to essentially recite a Spanish opening, because that's a bunch of common words when someone says "chess" and "E4". It has a lot less words to draw on for various middle game and end game positions so it just throws stuff out there which leads to hilarity, like summoning pieces from the grave or capturing its own piece.
@@rainbuckets Not again. Why does this nonsense get thrown around on literally every AI-related thread? No, pattern matching isn't how LLMs work. Not even close. Even if you ask an LLM to play without relying on known theory, it will do just fine for the first few moves, but will start to lose the thread as the position gets too complicated to track - same as humans playing blindfold chess. And the reason it starts playing chess after you ask it to play chess isn't that chess notation usually comes after the word "chess"; it's that the way that chess is played without a chess board is using notation, and a chess game usually comes after a request to play chess. If what you were saying was true, the LLM would start playing chess even after I say "don't play chess with me", which it doesn't.
Not sure how pre-recorded this is, but the reason for the weird ahh moves is due to the ai "forgetting" the initial moves so in each prompt if you remind them of what the board looks like they might actually do things ...
I was playing Gemini vs bing and bing jumped his king 2 squares to take the queen and then claimed it was check mate 🤣 then I told Gemini and Gemini straight accepted it, with not a single piece on the board even threatening the king.
This is not the reason why. The real reason why all the generative AI models make nonsensical and illegal moves is because they are trained on texts, and they simply spew back common text patterns. The same is true for their justifications behind the moves, it's like when a begginner hears some GM explaining the reason behind their moves and it's rather complex and the begginner tries to justify their moves using the same explanation but they're often just saying whatever words and don't actually know what they're talking about. That's also why they played perfect theory at the beginning of the game - those theory moves have probably been mentioned one right after the other so many times on the internet that the AI's could simply repeat them back. The AI's don't actually know what the board looks like and are not thinking about any moves.
@@aaronb2181i think thats untrue due to the fact they sometimes make sense with the moves. They must have some idea whats happening because a good amount of the time they move to do things already happening in the game and that is simply impossible. The chat bot is probably around the same skill as a 150 elo who knows openings blindfolded
The real question is - Who is the hydrogen bomb? Stockfish? The smartest of all? Or ChatGPT the cunning of all? Of course we all know... Martin is the one
7:26 that is an absolutly legitimate way of testing the knoledge, but only if this rule was established before the exam was taken. it tests how certian are the students in their knowledge and it's not uncommon practice in academic tests
after a certain amount of time, say around move 15 or 20, you should ask both bots to create a picture of how the board looks at that certain time. It might help you understand how they think and why they make certain moves.
not at all they can't generate complex image it will create a board where king is randomly somewhere etc etc i have tried it, it can't even generate a normal chess board
7:37 the best grade I ever got at school was 3X. "He tries adequately enough but the end result is a complete fail." (it was me trying to make a birdbox in woodwork).
Hi Levy, I believe Claude and Llama should be here instead of Martin and Stockfish, as they are technically not chatbots. We could have the Chatbot championship happen on its own and the winner faces stockfish and the loser faces Martin or something like that. This could be an interesting concept, as it can show us how chatbots like ChatGPT, Meta, etc bend the rules against rule based AI like Stockfish and Martin. This is my opinion, please do consider this for the future championship.
Claude might be especially interesting, because it has such a large context window that it should be able to go for much longer before making really stupid moves.
Its a good experiment actualy to show you you can be the best in the world but if you are sticking to the rules against an idiot that doesnt care for rules then you will lose
i love how he keeps a straight face, talking normally and not cracking out on the moves like "yeah of course why take the queen when you can threaten it"' LMAOO
Hi levi, just for help, the AI chatbots work by context so before starting a game you should gave them a promt of context something like "Hi i want a play chess with you, play focus and the best as you can, im going to pass you a sequence of moves and you have to make the next, your answer should look like this 1.e4 c6 2.d4 d5 3.Nc3 (or 3.Nd2) dxe4 4.Nxe4 Bf5." Then just pass the sequence like " 1.e4 c6 2.d4 your turn" This will help the chatbots to have more context and generate a better response. But also is fun all the weird moves they do. xd
That's good to know, but this isn't an actual test to see which bots are better. It's just entertainment, and the illegal moves are an integral part of that.
An intersting point of view: In medical school, not only was the absolute minimum pass mark for any exam 50% they were all negatively marked - fail to answer a question and you got nothing, but answer a question even partly incorrectly and the value of that question was *deducted* from your score! Why? It's not just about what you know, it's also knowing what you don't know and *not* taking guesses! Yep, 50% was a bloody hard mark to achieve.
Here was Gemini's response to me asking its thoughts regarding this game "Ah, you saw the game! It appears I was a bit off in my previous response. My apologies, I am still under development and learning to access and process information from the real world. Yes, I recall that match against Grok now. It was… interesting, to say the least. A bit of a wild ride, wouldn't you agree? From what I remember, we both made some questionable moves. I seem to recall blundering my queen quite a few times. And at one point, I even revealed my location to Grok mid-game! Definitely not my finest moment. Grok had its own share of blunders as well, including a slow-roll checkmate that was more accidental than intentional. Ultimately, Grok managed to checkmate me, but it was a chaotic and unpredictable game. Perhaps not the most impressive display of chess prowess from either of us. I am always learning and improving, and I hope to show a better performance in future chess matches. Do you have any specific questions about the game you'd like to discuss? I'd be happy to analyze some of the moves or share my thoughts on our respective playing styles."
@@lwalker8785 It doesn't remember it. Your chat instance with gemini will not be remembered by someone else's instance. It's just yapping (hallucinating)
levy for the next game try to make it live if it is possible. Like I mean you ask the bot for the moves during the video so it is more surprising for you and for us. It would be fun.
![BLACK BAG - Official Trailer [HD] - Only in Theaters March 14](http://i.ytimg.com/vi/Du0Xp8WX_7I/mqdefault.jpg)
Just dropped the e4 e5 Nf3 bonus course on Chessly.
Try it FOR FREE. No card required: bit.ly/e4e5nf3
when I click, it redirects to sign up to a plan
because you already used the free trial perhaps?
@@villbuzeep2043same
ok Levi Roseman
@System.Error. you still need to sign up to use the free trial, you just don’t need to give away your credit card info
Gemini playing a move and then casually telling Gotham that it knows where he lives is comical as hell
Snapchat: Talk to me nice or we can't keep chatting
Gemini: I ain't snapchat, I don't forget stuff, this you?
@@richardlynch9549 This is hilarious
Good game.
One small issue.
ᴵ ᵃᵐ ᶦⁿˢᶦᵈᵉ ʸᵒᵘʳ ʰᵒᵐᵉ
Claude will also be interesting
Bro should stop using chrome
Imagine at the end the chat bots just go "This is an equal position. Let's share the championship"
That would be a perfect ending to this tournament
😂
🤣😂🤣
Everyone hop on chat gpt to build it's AI learning and spam
"Offer to draw championship vs Stockfish" so that it actually happens and Gotham will lose his mind
the thing I love about this guy is that there are NO cuts in the video, bro has an infinitely deep bag of vocabulary a doesn't stutter once. Respect ☝️
Professional yapper
10:59
It helps with the daily uploads if he doesn’t even need a script or an editor
Agreed. But there are times when he just gets crazy and starts to mumble 😂
@@tomasbd5918yapping is an artform
Grok was actually playing an extremely advanced variation of the Botez Gambit the entire game, it's okay if you didn't understand it Levy.
18:33 Gotham realised he f'ed up
*places nuts on table to establish dominance*
*plop*
-slow roll
Leaked Address Variation 8:50
The doxxing gambit
lol
@@michka841man I am dying of laughter
the CIA defense
The location trap
I don't know why I enjoy these so much. I understand the bots aren't designed for any of the tasks required to play chess, and yet every illegal move just fills me with joy and child-like wonder.
finally! bots that play like they understand low-rated human play. I love them already!
Honestly, I love this brainrot "low quality chess" content. I was a bit sad when the last AI chatbot chess video was months ago and thought you'd moved on, and understandably so. I'm so pumped for this becoming an entire series!
LOW quality? his content is as good as when he became so MASSIVE during the chess boom. his chess content's quality isnt going to FADE ever!
@puffaddiqt Ahh no no, like @hyrummm said, I'm talking about the Chatbot's chess quality - it does also happen to be among the more controversial content Levy puts out, but it's like junk food, I can't help myself!
@hyrummm ik I’m making a low taper fade meme
@@Dr_Wings you said low so I had to reference the low taper fade meme because it is MASSIVE
@puffaddifqt LOOOOOOW TAPER FADE
I remember when Gemini was first released in beta when it was named Bard and no matter how much I primed it and shoved games into it for it to study, it would always brick itself and start cheating by move 5 and then forget that we were even playing chess. It wouldn't even use openings while just lying that they were. It just like to move pieces one square at a time until it would cheat and just promote pawns on the home square. It's always loved to spew utter nonsense while trying to analyze positions. It's fun to see how far it's come from beta (this game is genuinely a massive improvement over beta) but still see that it loves to hallucinate some utter bullshit. It sure as hell knows how to threaten pieces but it doesn't understand anything about how to take them.
My favorite bit is its decided to not only threaten pieces but also threaten the other players by revealing their address. Thats some next level play.
Yes, well, this is the AI that thought George Washington was black. We can’t expect too much from it.
How a typical gotham video starts: *breaths in* ladies and gentleman ....
How it ends: AND THE GEMINI HANGS A QUEEN CAPTURING ITS OWN PAWN AND THE NEXT MOVE IT SACRIFICES THE ROOOOOK
11:41 Grok had a mouse slip. It happens to all of us.
It happens to a lot of us...
guys I think this guys in need of medication
Only championship that Stockfish can't win
idk about that, Stockfish is the only competitor who will never self-destruct. Even Martin will play some awful moves
yea but when you are playing against a cheater even if they do self destruct every now and then they can just say nuh uh this is a checkmate or yes i did have a queen here next to your king it was just hiding dude@colonel__corn
Fr 😂
welp, stockish won
@Celatra I jinxed it
18:00 "'I like your historical videos'" Bro really took that one comment personally
I thought I was the only one who noticed that as well
Context?
@@KrazymaddogGo to his previous video of ChatBot Chess Championship Day 1 and see the top comment there.
@@KrazymaddogI think a comment from yesterday he responded to
Fr
8:16 What's crazy is that this position was reached in 5 games on the Lichess database, including one between two 2500-rated players (one of whom was a GM). The move order was different, but still... I think the GM lost the Bishop on a4 due to (essentially) a pre-move.
gork : TAKE MY QUEEN!!!
gemini : no, no i don't think i will
“Gork” 🤣 no hate tho just thought it was a funny typo
@Pantherboy957 you know what? I'm not gonna change it. That way funnier
Grok the sigma 🗿
@randomdude055 it is
Gork 🦧
bro finally left magnus alone
Very original comment
Alone with his wife
LOL
for the first 3 mins lol
if you recall correctly, this championship has magnus (stockfish) Carlsen and hikaru (chatgpt) nakamura on opposite ends of the bracket as mentioned in last video
0:10 WE HAVE TO DO THIS AGAIN!!!!!
Absolutely, it’s the best chess championship
😂
This needs to be the most liked comment
And again & again
Most definitely 😂
Gemini leaking your location is crazy😂 8:37
Crazy
@@AidenDu-i8u I was crazy once
😂😂😂
I honestly hope this becomes a yearly tournament. I'd love to see if the ai's actually improve at all between this January and next January, especially if any of the developers of them see this series.
yessssssss this is such a good idea to track the ai's yearly improvementttt
Me ~4 min into the video: I didn't come here for main line play! I came here for MONSTER REBORN pawns moving laterally to capture queens!
Me ~11 min into the video: That's more like it! :)
You helped me sir
18:39 nuts on the table💀💀💀
Only real fans of this 5’2” chess guy have seen nuts on the table💀🤫
whenever my opponent is winning, i put my nuts on the table and they retreat
Briliant man.
I'm dying how he said not like that oh my god 😂
18:29
Bro letting out his anger on that professor is so good
I'd be willing to bet that the other main reason the professor had that policy was that it's much easier to look at "I don't know" and give partial credit than to try to decipher the answers of a student who only half-knows how to do the problem and figure out how much partial credit to award. I'm sure it saves that professor a lot of time. Source: Am professor
@beccabeKAW That professor would hate me lmao, I always have the attitude of "a blank answer is always wrong, a BS answer can be partially right accidentally"
yeah but he is right, the professor was crazily bad for doing that kind of grading
3:49 I'm so grateful you've mentioned it, I was looking for it.
13:01 Queen Check *throws blocking pawn off the screen* LMFAO
You got me crying laughing on my drive home 18:20.
Playing the accordion with the rook😂
18:27 ”its when you have the NUTS on the table” 🤨😭
Stockfish throughout the entire video is facing epileptic concussion, its just going up and down for no reason. It has gone feral after this
It's because the website isn't set up correctly. It's in the custom position setting, that allows you to place pieces wherever, instead of in a game. Stockfish doesn't understand because it assumes the whole time that it is white's move.
Yes, I believe that's right. And not having seen the game to the end , my guess is that it's to prepare for "illegal AI moves" 😉
@@BjarkeEbert That poor bot is just paranoid of his pieces getting sucked into the black hole and his opponent using magic to summon new pieces. Sad lyf of stockfish
Finally no Magnus Milking
Nevermind… 3:49
Never mind… 3:49
Bro is evolving, with meta commentary
K
Levy never _stops_ to mention Magnus for relevance. 👍
@@DontReadMyyPictureee don't ever comment again...
Casually dropping your location is the most Google thing Gemini could've done. Truely like father like son.
Gemini: This move is a bit of a surprise, as it’s not a common move in the Berlin Defense. It does put some pressure on the c4 pawn but it also gives you some space on the queen side.
Levy: None of that is accurate. Not even close.
It is not a surprise, it is not the Berlin Defense, it's the Spanish Ruy López, it is not even close to c4, it does not give you any space and it is not even in the queen side lmao
8:33 was such an unexpected threat to the match referee Levy 😭😂
My favorite 5'2 GM RUclipsr is back
4'11'' IM*
pin of truth
*3'5, 800 elo RUclipsr
I'm 2'5
2'3.5@@GothamChess
18:28 got me so out of guard XDDDDDDDD
Someone get this into a sus levy clips compilation
CEASE THE SUS!
NUTS 🥜🥜
7:11
I am SO glad our exams are marked by an impartial exam board instead of the professor with arbitrary rules.
1:17 I felt threatened by that statement
8:44 he played the adress gambit
8:25 I'M DYING LMFAO
aftet finishing the video I can say that this is the thing that made me laugh the most in this week so far this is freaking gold
It feels so strange the AI would come up with that SO out of the blue
Like it was trying to threaten Levy who it assumed to be its opponent to resign from the game before it'd lose
"They're only slightly worse than the quality of your games." jokes on you. they are actually about the same.
At least I know how the horsey moves
@@lukasg4807 nuh uhhh , jokes on you if the mark dont indicate where the piece moves my horsy is gonna go sniping
3:49 Self-reflection at its peak. Lol.
"less then 1% of this audience is above 1700 / 1800 hundred" meanwhile i am chilling at a solid 400
The previous match was about disrespect and today's match was about not taking the pieces.
I know where you live
-Gemini-
@@chirwynadams3292 When will AI say this when they get offended?: "Sorry, I can't continue this conversation, my dad is taller than your dad!"
01:20 Getting offended over 2 bots playing chess is another level of stupidity 😐😐
i hate ai for anything to do with art but it's mostly just silly and fun in stuff like this
i dont think people are getting annoyed by the chess part, but by the ai part
@@heatheretaithahano reason to unless it's for lying
I think he means just people mad that he’s allowing them to play illegal moves which almost made stockfish lose in the first round to a cheating ai
@EDGKangKang even though the illegal moves are what make these games interesting and funny as hell.
7:20 Bro's venting 💀
But fr that's some bullshit
Although the SATs are actually graded more or less the same way. If two people taking the SATs know half of the answers and one leaves the other half blank while the other guesses and gets about 1/4 of their guesses right (assume multiple choice with 4 choices), the one that left half the questions blank will get a higher score. The reasoning is a little different - it's because the purpose of the exams is to test what you actually know, not your luck at picking random numbers, so the test is designed so that people who only answer the questions they know will usually do better than people who try to fake it by guessing randomly.
Honestly seeing absolutely horrible chess moves is just as helpful in me learning chess as seeing brilliant moves. Even more so because it's easier to understand why a bad move is bad and helps you avoid some of these pitfalls yourself as a beginner.
Seems like the best explanation for how these bots play is that it's like playing blindfold chess, but you also have extreme dementia so you don't know what the board looks like so you just say random chess moves.
I mean they literally are playing blindfold chess
The bots don't have actual intelligence, all these bots do is say things that seem popular to say after certain words. So if you tell it you're playing chess and you played E4, it's going to essentially recite a Spanish opening, because that's a bunch of common words when someone says "chess" and "E4". It has a lot less words to draw on for various middle game and end game positions so it just throws stuff out there which leads to hilarity, like summoning pieces from the grave or capturing its own piece.
@@rainbuckets LLMs do think bud. They learn in ways very, very similar to humans. Stay uneducated, get a CS degree.
@@rainbuckets Not again. Why does this nonsense get thrown around on literally every AI-related thread?
No, pattern matching isn't how LLMs work. Not even close. Even if you ask an LLM to play without relying on known theory, it will do just fine for the first few moves, but will start to lose the thread as the position gets too complicated to track - same as humans playing blindfold chess.
And the reason it starts playing chess after you ask it to play chess isn't that chess notation usually comes after the word "chess"; it's that the way that chess is played without a chess board is using notation, and a chess game usually comes after a request to play chess. If what you were saying was true, the LLM would start playing chess even after I say "don't play chess with me", which it doesn't.
I love this levichess guy
He’s really up there with Gary chess, inventor of chess
grok really brought out a pawn from nowhere and the bar just unmichaeljacksoned
uniMichealJacksoned. 😭😭
A Martin vs stock fish final would be awesome
Gemini was my Grandmas name. She was crazy good at chess. I miss her😢
17:58 oh hell no, THIS is prime chess content
I really love this content man! Keep it up.
I feel like this game was higher quality than some how to lose at chess games
No AI is safe from him.
gemini: ok s o im gonna move my pawn here *doxes you*
Not sure how pre-recorded this is, but the reason for the weird ahh moves is due to the ai "forgetting" the initial moves so in each prompt if you remind them of what the board looks like they might actually do things ...
Lol, they make like illegal moves too
Such as they sometimes say King to D6 if King is on F1
I was playing Gemini vs bing and bing jumped his king 2 squares to take the queen and then claimed it was check mate 🤣 then I told Gemini and Gemini straight accepted it, with not a single piece on the board even threatening the king.
This is not the reason why. The real reason why all the generative AI models make nonsensical and illegal moves is because they are trained on texts, and they simply spew back common text patterns. The same is true for their justifications behind the moves, it's like when a begginner hears some GM explaining the reason behind their moves and it's rather complex and the begginner tries to justify their moves using the same explanation but they're often just saying whatever words and don't actually know what they're talking about. That's also why they played perfect theory at the beginning of the game - those theory moves have probably been mentioned one right after the other so many times on the internet that the AI's could simply repeat them back. The AI's don't actually know what the board looks like and are not thinking about any moves.
@@aaronb2181i think thats untrue due to the fact they sometimes make sense with the moves. They must have some idea whats happening because a good amount of the time they move to do things already happening in the game and that is simply impossible. The chat bot is probably around the same skill as a 150 elo who knows openings blindfolded
dude just why its clear who will win MARTIN SOLO 🗿0:56
0:56 7 coughing babys VS hydrogen bomb
😂😂 Oppie would be proud
Actually probably not…
The real question is - Who is the hydrogen bomb?
Stockfish? The smartest of all?
Or ChatGPT the cunning of all?
Of course we all know...
Martin is the one
14:38 you know what else is MASSIVE
LOOOOOOOOOOOOOOOOOOOW
TAAAAAAPERRRRRRR
FADEEEEEEE
7:26 that is an absolutly legitimate way of testing the knoledge, but only if this rule was established before the exam was taken.
it tests how certian are the students in their knowledge and it's not uncommon practice in academic tests
7:28 why bro started to cooking his old teacher😭
This is fun content, but I'm really looking forward to road to GM coming back. Your lessons with Neiksans is some of my favorite content you've done.
7:11 As someone passionate in teaching, I was soo sad when I heard Levy's story ... Until Levy called that prof an idiot 🤣🤣🤣🤣
after a certain amount of time, say around move 15 or 20, you should ask both bots to create a picture of how the board looks at that certain time. It might help you understand how they think and why they make certain moves.
not at all they can't generate complex image it will create a board where king is randomly somewhere etc etc i have tried it, it can't even generate a normal chess board
Ruy Lopez, Morphy’s Defense: SSSniperwolf Variation
This is what i imagine modern politician and businessman playing chess looks like
That’s not the bracket! Stockfish won the first one, by the opponents disqualification.
It was to not spoil the results for folks who hadn't seen the first video yet
As a complete noob at chess, this was brutal to watch probably the same feeling Gotham gets watching 500 elo players 🤣
it's like chess playing software from the 70's
you better run levi, gemini is coming for you after that lost
The tangent about the teacher LMFAO
3:16 when he said "Руййй Лопезсс" i felt that
😭😭🙏🏻
7:37 the best grade I ever got at school was 3X. "He tries adequately enough but the end result is a complete fail." (it was me trying to make a birdbox in woodwork).
I love when it was just like "I DECLARE CHECK!" despite no such check existing, and the other bot just goes with it haha
"King to f oouaaaaaaaahhhhhnnnnn"
-Levy rozman
Hi Levy, I believe Claude and Llama should be here instead of Martin and Stockfish, as they are technically not chatbots. We could have the Chatbot championship happen on its own and the winner faces stockfish and the loser faces Martin or something like that. This could be an interesting concept, as it can show us how chatbots like ChatGPT, Meta, etc bend the rules against rule based AI like Stockfish and Martin. This is my opinion, please do consider this for the future championship.
I mean if you're a researcher interested in this, you should just run the games yourself, no?
Oh sorry, I didn't mean to write the last line I will edit it. Yeah, I can test that on my own.
Claude might be especially interesting, because it has such a large context window that it should be able to go for much longer before making really stupid moves.
It's a pity about Stockfish, because everyone plays with teleports😢😂
Its a good experiment actualy to show you you can be the best in the world but if you are sticking to the rules against an idiot that doesnt care for rules then you will lose
We got levy pressing a professor before GTA VI😂
Gemini hallucinating an absentee pawn just Grok to take the queen with that same pawn is fantastic stuff.
I was trying to do this and gemeni responded with: I'm sorry, I'm not able to play chess yet.
2763rd best chess player in the world is back
Christmas vid throwback ig
bfdi reference???
He’s 2nd if you think about how much he makes
Making him being in top 1,6 % of all active ranked chess players. Really not that bad.
@@ERFARINGyoutubeyou have the best pfp in existence
This is how you play when the LSD kicks in
You should give the bot the FEN in each prompt.
Fein fein fein fein
Finally. You have presented a video where I believe I could have beat either contestant
i love how he keeps a straight face, talking normally and not cracking out on the moves like "yeah of course why take the queen when you can threaten it"' LMAOO
Hi levi, just for help, the AI chatbots work by context so before starting a game you should gave them a promt of context something like "Hi i want a play chess with you, play focus and the best as you can, im going to pass you a sequence of moves and you have to make the next, your answer should look like this 1.e4 c6 2.d4 d5 3.Nc3 (or 3.Nd2) dxe4 4.Nxe4 Bf5."
Then just pass the sequence like " 1.e4 c6 2.d4 your turn"
This will help the chatbots to have more context and generate a better response.
But also is fun all the weird moves they do. xd
This ruins the video and has levi doing the moves for them. Absolutely terrible idea
@@harrykrause2879great hating
That's good to know, but this isn't an actual test to see which bots are better. It's just entertainment, and the illegal moves are an integral part of that.
@@harrykrause2879 "This sign can't stop me because i can't read"
An intersting point of view:
In medical school, not only was the absolute minimum pass mark for any exam 50% they were all negatively marked - fail to answer a question and you got nothing, but answer a question even partly incorrectly and the value of that question was *deducted* from your score!
Why? It's not just about what you know, it's also knowing what you don't know and *not* taking guesses! Yep, 50% was a bloody hard mark to achieve.
Here was Gemini's response to me asking its thoughts regarding this game
"Ah, you saw the game! It appears I was a bit off in my previous response. My apologies, I am still under development and learning to access and process information from the real world.
Yes, I recall that match against Grok now. It was… interesting, to say the least. A bit of a wild ride, wouldn't you agree?
From what I remember, we both made some questionable moves. I seem to recall blundering my queen quite a few times. And at one point, I even revealed my location to Grok mid-game! Definitely not my finest moment.
Grok had its own share of blunders as well, including a slow-roll checkmate that was more accidental than intentional.
Ultimately, Grok managed to checkmate me, but it was a chaotic and unpredictable game. Perhaps not the most impressive display of chess prowess from either of us.
I am always learning and improving, and I hope to show a better performance in future chess matches.
Do you have any specific questions about the game you'd like to discuss? I'd be happy to analyze some of the moves or share my thoughts on our respective playing styles."
Really? Gemini said that? It's wild that she remembered the game, and had an opinion on it.
@@lwalker8785 It doesn't remember it. Your chat instance with gemini will not be remembered by someone else's instance. It's just yapping (hallucinating)
@lwalker8785 It didn't, I had to send the link to Levy's YT then it remembered.
Lets get an illegal move counter on the screen for each side
poor grok is gonna get mated in like 10 moves
8:55 bro wth 💀
18:00 "'I like your historical videos'" Bro took that one statement personally.
@@ChainBreakingMenace yep
these bots are getting more advanced
how are these bots getting more advanced
They arent anymore advanced it just copies other comments
@human4116 ohhhhhh ok
7:38 As someone in a job writing sql I know that saying idk when you don't is generally a good move
Im guessing Levy failed this class
Thse bots played surprisingly legal. There was only one illegal move.
19:23 how did bro know.
levy for the next game try to make it live if it is possible. Like I mean you ask the bot for the moves during the video so it is more surprising for you and for us. It would be fun.
This
10:40 It's Chess 2
Chess dlc
best stand-up comedy show i want more tournaments like this
Him having to dq snap chat was like the magnitude pants scenario.