Chess Pro Explains Chess in 5 Levels of Difficulty (ft. GothamChess)
Director: Maya Dangerfield
Director of Photography: Charlie Jordan
Editor: Robby Massey
Host: Levy Rozman
Guest: Level 1: Riyae Park; Level 2: Jeffrey Kim; Level 3: Tani Adewumi; Level 4: Irina Krush; Level 5: Siddartha Sen
Creative Producer: Wendi Jonassen
Line Producer: Joseph Buscemi
Associate Producer: Paul Gulyas; Kameryn Hamilton
Production Manager: D. Eric Martinez
Production Coordinator: Fernando Davila
Casting Producer: Vanessa Brown; Nicholas Sawyer
Camera Operator: Brittany Berger
Gaffer: Rebecca Van Der Meulen
Sound Mixer: Lily Van Leeuwen
Production Assistant: Ryan Coppola
Hair & Make-Up: Yev Write-Mason
Post Production Supervisor: Alexa Deutsch
Post Production Coordinator: Ian Bryant
Supervising Editor: Doug Larsen
Assistant Editor: Billy Ward
Released on 12/18/2023
I'm Levy Rozman.
You might know me as Gotham Chess.
Today I've been challenged with the task of explaining chess
in five levels of increasing difficulty.
Chess is an absolutely fascinating game.
Think of it as a language. It's completely timeless.
It's been around for thousands of years.
It's also completely ageless.
[contemplative music]
A lot of people who are gonna watch this video
have never played chess before.
How would you explain chess to them?
I would start off with the basics
by teaching them how to move the pieces,
make the pawns, knights, queens, bishops, kings.
What's the goal of the whole game?
Well, the goal of the game is trap the king
and attack it at the same time.
Yes. So it can't run away.
Yeah. Exactly.
And also take all of their pieces.
Yeah. For me,
I would describe chess as a lot of what you said,
and also, it's a mix of really quick plans like tactics
and then things that can take four, five,
six, seven moves like a plan, like a strategy.
So let's practice.
In this position, I had the white pieces.
You're playing on my side.
My opponent was playing with the black pieces.
So we switched a little bit.
So if I said, Hey, you get three moves in a row,
can you come up with a plan? I think I could.
If I have three moves in a row,
I'd probably checkmate them with rook H3,
queen H4, and then queen takes H7, checkmate.
I'm very happy that's the first plan
that you came up with, right?
You can bring your queen out this way or the rook.
You can start with both,
and then you come here and then you checkmate.
But I did say white gets three moves in a row, right?
Yeah. But that's
how we have to think about plans, right?
We have to think about, Okay, well if I get a couple
of moves in a row, what would I like to accomplish?
What does my opponent want to do?
So I'm not just gonna sit here and get checkmated hopefully.
Right? So what do you think I want to do in this position?
I think my king is on a bad scale right now,
so maybe you are trying to like trade
and then somehow get an attack on my king.
Oh, you're saying like if I take this pawn,
maybe in the future I'll try to open up this diagonal.
See, that's very good vision.
There's some pieces in the way, but yes.
In chess, you have to always keep an eye
on what can possibly open.
Do you know how you can tell
whether a position is closed or open?
Usually when a position is closed,
pieces are more squished together.
The easiest way by far to figure out if a position
is closed is how many pawns do we each have?
Right now, it's eight. So think of it like a fence, right?
So if you and your neighbor have a full fence,
you can't really see each other,
but if there's only four pawns,
that means somebody
probably knocked down half of the fence, right?
So if we start picking up all these pawns,
then the position is more open.
So this would be a closed position.
This type of strategy would be called an attacking strategy
'cause you found a way to get your queen
and rook down this line against my king.
Before we even talk about what tactic is here,
who does it look like is winning this game?
I think white's winning. Okay. Why do you think so?
Well, the black king is very exposed
into the white side of the board,
and like the white king is a bit more safe.
Yes, you're 100% right 'cause in chess,
the two most important things we have
to think about are who has more material,
but then, like you said, who has the weaker king?
And I would say my king has gone way too far, right?
So there's a tactic here for what?
Maybe. Is it bishop takes upon on E4?
So that's like immediately you get a reward,
but I'll give you a hint.
The best way to start looking for a tactic,
always look for a check first.
Do you have a way to attack my king?
Yes. Rook, H1.
So exactly. You would give me that check.
Now where can my king go? Let's just try to think.
Your king can go to king G4.
Yes, that's my only move. So I would go here.
Now what else is on that diagonal?
The rook. Yes.
And which of your pieces can-
The bishop. The bishop.
So do you know what this type of tactic is called?
A skewer? Yes. Well done.
And then I would take the pawn.
And then I would take the rook.
And then you would take the rook.
Now the funny thing is you are now totally winning, right?
But as we know in chess,
maybe one day white will make a mistake in the future.
So there's no point for black to surrender here,
but this is definitely a really good start.
[contemplative music]
How long have you been playing chess?
I played a little bit of chess growing up
against my brothers,
but ever since the Queen's Gambit show came out,
that's I think when I downloaded the app
and started playing against friends for fun.
So chess is separated into three games, obviously.
You have your opening, which is the early stage of the game.
Then we have the middle game, right?
The most confusing, mysterious side of the game,
and the end game
where there's no pieces remaining on the board.
You gotta have to know how to convert positions,
defend positions, understand how to play the game
with only five pieces remaining or something like that.
So for you, what do you struggle with the most?
Yeah, I would say I really struggle in the mid game.
I feel like I've developed most of my pieces,
try to control the center, but at a certain point,
I'm a little bit stuck on like, okay, now what?
What do I do?
So I'm going to look up your account.
We're gonna pick a game and we're gonna analyze it.
Sounds good.
In all middle games where we have this mix of tactics,
which are immediate gain or immediate loss,
depending on if we're getting hit
with a tactic or playing one, and strategic chess.
I just want you to tell me,
who do you think has a better position here and by how much?
To me right now, it feels pretty even.
Like there's no major vulnerabilities from both sides,
but I guess the white pieces,
the pieces are a little bit more developed, I would say.
Everything you just said is 100% accurate.
So that's one of those baseline pieces of information
that we need for the middle game.
I have the two bishops.
How can I activate one of my bishops?
You have a little plan you could actually implement
right now that would make one
of your bishops extremely strong.
You think you can find it? If I had to guess,
I think it'd probably moving this bishop here.
Yes, you definitely want to get the the bishop out,
but you don't wanna move it just yet.
So how do you do that? I am not sure.
This is how we unlock the intermediate level,
which is this move.
That move looks like it's disconnecting
from the whole board, right?
It's a move totally on the edge,
but nobody is gonna stop your bishop
from going to that diagonal,
and then you're threatening a tactic.
You're threatening a skewer.
You would hit my queen and my rook.
Stronger piece in front, weaker peace in back,
and then you would be guaranteed to win some material.
Just like little pattern recognition in the middle game.
So for the future, you're gonna have a game
where your opponent makes the same trade early, right?
They trade their dark-squared bishop,
and you can activate it like that,
but in this game, you focused on the center,
which was totally fine also.
Maybe one question I have is like,
as I'm also looking at this position,
just trying to understand like,
okay, what is my opponent trying to do here?
One thing to always monitor in every chess middle game
is does my opponent have a pawn break?
Can my opponent strike somewhere
with a pawn that would open up the board?
If you have a low risk position,
your position's just nice and comfortably better,
you don't want chaos.
[Jeff] guess maybe they would try to advance their
middle pawn. Yes. Yes.
And this move is actually very annoying for white
because not only am I now targeting things in your center,
what else have I changed just by moving this pawn?
Now there bishop is activated.
Yes. Another thing for middle game.
So we just keep adding to that bucket of knowledge.
Pawn breaks for the opponent,
and then, every time something moves,
like a pawn, we have to look around.
What else is now opening up?
And we wanna get to a situation
where you knew this move was gonna happen,
and the next I'm gonna come here,
and then now what can I do?
I can bring both my rooks,
and now I have no problems, right?
Now I have both knights, queen, rooks.
So we need to get to a point where we see that coming,
and we're not just focusing on our own plans,
but that's not always easy.
So with a pawn here, that disrupts my plan, right?
Because now this move fails to-
Yeah, respond to the skewer. Yes.
As you're trying to improve your middle game,
one question you always have to ask yourself is,
Is that trade good for me?
Is it good for my opponent? It's not checkers.
You don't have to take me, and I don't have to take you.
So we can do something called keeping the tension.
We're just having a staring contest here among the pawns.
So I'm gonna ask you, do you think it's better for black
to capture this pawn or keep the tension?
Just slide the rook over
and threaten to push the pawn here,
which would be a tactic and a fork.
If I was black, I would advance the pawn
because I think knowing that the black has two knights
and I have two bishops,
they would want more of a closed game.
Yes, but before we advance, we need to prepare it, right?
Because you push it now, these two would win that trade.
So yeah, rookie 8 is a very good move.
You used kind of like a heuristic, right?
Well, black has two knights.
So black prefers the position closed
because decide what the bishops wants
to be able to see the whole board.
Yes, exactly, but your opponent disagreed with you
and took this pawn
because that's just what intermediate players do.
They see they can take something. Pawn for pawn.
What the heck? Like, sure, and you did this.
Now what does this move change in the white position?
So that opened up the diagonal for my dark square bishop.
Yep, here, yeah, and what else?
What else did that move open up?
It opened up the E file for us.
Yes, exactly.
So now in the snap of your fingers,
you can get your bishop out and the rook,
two things that a move ago
were just totally blocked in by a pawn, exactly.
So in the middle game,
trades are either good or bad for you and for your opponent.
You know, I'm in the mid game right now,
but what do I need to do or learn
in order to take it up another level
to a more advanced player style?
It's a lot of practice and analysis.
Really, an advanced chess player
was not born an advanced chess player.
They have probably hundreds, if not thousands of more games
in their mind, in their past, in their history
that they've analyzed, that they've studied.
It's like any athlete, you know?
I put my weight on this foot,
and so I wasn't able to hit the shot back that well.
So the next time that that happens,
I'm gonna be more prepared.
[contemplative music]
I'm wearing a black hoodie, black pants.
You definitely have more style than me today.
So in chess, people have style, too.
What's your playing style?
I wanna say quite aggressive.
Like I really wanna win the game fast.
So as we move up the levels today,
we wanna start introducing slightly more advanced terms.
One of them is called initiative,
but what exactly does that mean to you?
So like if you're in a battle,
you win a certain like spot, right?
A really important spot. That gives you initiative.
When you get something, right, that you want,
then it can start building up.
It can start building up to be better
and better and bigger and bigger.
When you find a target
and you have a way to create threats,
attacks against that target, against that goal,
but you do it without sacrificing your resources, right?
So you can imagine two boxers, right?
One is being pushed into a corner.
He could still win, he could totally still win,
but for now, one side is definitely the aggressor.
And has the initiative. Exactly, has the initiative.
You're a tactical aggressive player.
You're always looking for the initiative. Me, too.
Why don't we play a blitz game?
Good luck. Good luck.
Okay.
Okay. That's not the opening I remember you playing.
[tense contemplative music]
We've played in the park, we've played online.
Wonder how many games we've played against each other.
We've played, I think, probably like,
including over the board and online,
maybe like nine, 10 games.
Okay, so we have like, just like a regular Fiancetto.
Let's see.
This is not the Tani I remember.
Tani- It's quite slow.
It's a very solid, you know, steady approach.
There's nothing wrong with it.
Okay. Very balanced. Let's see.
Actually, I had this chess opening really recently,
and I lost that game, and it was very painful and very sad.
You've have like black has more space.
Would you say black has more initiative?
Not yet. Not yet,
but considering the way I said it,
it's like you follow up with something.
So you just attacked me with that, and I move back.
So you kind of, you have the initiative.
The tough thing about chess
is you can have a good position,
but you actually have to move pieces.
You know, like a piece of art,
you just say, Oh, it's done.
So chess is different from art.
[Levy] Chess is different from art. Let's go here.
I wanna go here. Okay, let's go here.
Okay. Let's see.
I'll go here just in case.
[Levy] I don't know.
Okay, I'm just gonna like take some more space from you.
This is quite like a, just a,
Wow. just a position.
You know what? We've been talking about initiative.
I think I just gotta go for it.
I just gotta like sort of of push all my pawns.
So that could be good. That can be bad.
I have to deal with it at some point.
No, you don't. Wow! Just in case
because I feel like my position is just like a fortress.
So it feels like I can do moves like rook H1.
Doesn't help my position if you don't attack,
but think it's just good to be just steady.
I'm stunned. Okay, you know what?
I'm gonna go bring more pieces to help.
More pieces to help.
Okay. Good.
How are you just ignoring all my,
my whole squad is over there.
[Tanitoluwa] Don't wanna take space.
Wow. I mean, I gotta go for it, right?
I feel like- You have to. Okay.
I really feel like this is a nice example
of black has an initiative.
[Tanitoluwa] Yeah, black definitely has an initiative.
Black has an initiative.
But can black capitalize on this initiative? I don't know.
The problem with having a position like this
is you feel a lot of pressure to do something.
If I was black, I want to end the game like right now.
[Levy] You wanna end the game right now?
Yeah.
Okay, let's see. What can I do?
[tense dramatic music]
Okay, I'll do that. Okay, try and roll up the sleeves, okay.
Yeah, that's the chess player's sign of focus, just-
Yeah, when you roll up the sleeves.
I guess I have to defend my pawn.
I mean, don't really want to, but fine.
[Tanitoluwa] Now I'll be annoying to you.
Take as much pieces off, and I will take back-
Would've been nice if you didn't, but.
Okay, now maybe we go up.
[Tanitoluwa] And I'll take that pawn.
That was all part of the plan.
It was a very greedy move of you.
[both laugh]
[Tanitoluwa] All right, let's see, here.
Now I'll take a pawn. Oh!
All right, let's see. What?
[whimpers] I'm scared.
Okay, I can do that. Yes.
[Tanitoluwa] Can do that.
[Levy] The board looks different in this building.
[Tanitoluwa] Are you up a pawn? No.
[Levy] Six, six each. We good.
Um, blah blah blah blah blah blah blah blah.
[Levy] Aha!
That king of yours, just saying.
Just saying that, king of yours.
Come back. Come back. I'll go forward.
No. Ah! Ooh!
Ooh. It's a block with my rook.
This is the Tani game in the tournament all over again. No!
[Levy mumbles] [both cry out]
I almost lost on time.
[Tanitoluwa] Yeah, I have no idea what to do.
Me too, by the way. Ah!
But then that. Is you a rook trapped
but like I can't actually take it?
That's so weird! Check.
Wait, what? [laughs] Okay.
[Levy] Oh, I like having more time. That's nice.
Think I'll take those pawns.
Oh, can I like go eat the other pawns? No, I can't.
Ee! Ah.
Keep going. Whoa.
Yes, keep going. No, I can stop you.
No, you can't. I mean, yes you can.
Can you? Wait, can you?
Ah! Okay.
Oh my gosh. No. Oh.
My bishop. Nice game. All right, good game.
Oh, that was crazy. Ah!
[dramatic electronic music]
Generally in five levels,
level five is the ultimate master,
the person that's played the game 20,
30 years, has the highest pedigree.
Actually, that's you in this case,
and we're still on level four because computers
have been better than humans at chess for like 30 years.
How has it changed the way you approach the game?
Yeah, I mean, I think chess players
are using computers intensively all the time.
Obviously, with material odds,
it's gonna be one part of the training,
and other part of the training that I find interesting
is like, you know, to find positions from,
let's say, real life tournaments.
And I've been doing that actually
from this recent tournament that's going on in Isle of Man,
playing out some of those fascinating positions
and just seeing what the computer does,
and he's like coming up with resources
that you would never think of, you know?
Today, we've been talking a lot about how chess
is kind of a mix of tactics versus strategy.
Let's elevate that discussion. So what is chess to you?
Well, the way my coach explained chess to me
that really seems intuitive is that chess
is a game of time, space, and harmony.
You have one turn to get things done, right?
Obviously you try to make it as efficient as possible.
We have the saying in chess that, you know,
you lost by one tempo, by one move.
Space. You know, we have a game of 64 squares, right?
So control over these 64 squares is what the game is about.
You can get control in moving up your pawns.
You can get control by winning opponent's material, right?
So obviously, that's another big element
that we're fighting over,
and harmony is basically like you got this army
of pieces, right?
Your job is to make them work well with each other,
coordinate them, and build that beautiful picture of harmony
that, you know, all chess players aim for
and are so pleased when they can finally achieve.
So I've never thought of chess that way.
This is very interesting to me.
So of those three, what would you say is the one
that you have more mastery over than the other?
Harmony trumps everything.
Like you could be winning when it's not your turn,
and you're like totally behind on material.
So it shows you that harmony is the end goal.
Chess is a miniature life.
I always like to say that chess exposes all
of your flaws on the board. Oh yeah.
in your training habits,
in the way you approach certain positions.
So you say you can see yourself on the chess board.
Who is Irina on the chess board?
Yeah, well, I think Irina is a fighter.
You know, Irina is not someone who likes to give up.
Irina is someone who wants to leave everything
on the board, right?
Like every last drop. That's the idea.
How do you see yourself as a player?
I think I like to be creative.
So I find fun in chess by learning 10 different openings,
trying different positions.
I like chaotic positions.
I like positions where I can attack.
I like positions where I'm getting attacked
and I can maybe defend and find some creative way
to turn things around, and I really, really struggle
when I need to sit there and just do nothing.
So, you know, we have opposite styles, Levy.
I think our blitz game is gonna be a lot of fun.
All right.
If I had to guess, I'd guess you're a King's Indian player.
No.
See, that's an opening that requires a lot
of time and practice and dedication,
but I'll play Oh wow.
something also still fun. All right.
Okay. Dutch defense.
[tense contemplative music]
Is that the Leningrad? I think G6 is Leningrad.
I've tried to find a name.
I don't actually know the name of this one.
I guess there's also like the Stonewall, right?
You can play. Yeah.
All right, let's go for the most solid move.
That's not fun. We don't want double pawns.
No, we don't. I would've liked double pawns.
Okay, let's complete that bishop Fiancetto.
So you actually play the Dutch.
That's not just like today's surprise.
I play like a lot of openings poorly.
That's my approach to chess.
I just play things that I think will be fun.
Then the game starts going, and I'm like,
Why did I play this?
I should learn real openings.
Okay, I guess I'll...
Yeah, this is nice and boring.
Yeah. Prefer it wasn't boring.
Hmm, I think I gotta do something like
on the queen side here before you get your E5 move in.
Yeah, I was debating playing
A5. A5, yeah.
But question is, is E5 a good move or a bad move?
There's only one way to find out, so.
[Irina] Yeah, you're gonna go for it, yeah.
All right, well I gotta take you.
Can I take with the pwn?
Probably. E5
looks a little dangerous,
but I guess we can analyze that later.
All right, so what are you threatening?
At the moment, not much, yeah?
If I take an E5, it gets very boring.
Now it's a good thing that like my queen
is not gonna hang when you take my knight.
Yes. Well, actually, it's not a good thing.
I would've preferred that it might, you know, it might hang,
but oh, we only have like 90 seconds.
We have to make conversation
and also play some semi-decent moves.
That's what streamers do, right?
Like Naka. [laughs] Awesome.
I do YouTube, you know?
Okay. I think we can-
Yeah, I gotta save my queen.
So we're gonna recapture like that.
What if I do this?
What if I just don't acknowledge what you're trying to do?
That's interesting, right? 'Cause I don't have A3.
I can play B5.
Five's kind of interesting. That's a good point.
Yeah, hmm. I have C6, but you can play B6.
I can take on D6. Looks a little boring but possible.
[Levy] The average chess player would want
like five more minutes now. Yeah.
I think this move's kind of interesting,
so we're gonna go for that.
[Levy] Okay. It's a pawn sacrifice.
Yeah, sort of temporary,
but I'm trying to keep your rook closed.
[Levy] Yes, I take, right?
Yeah. I feel like that's-
So we're trying to get the pawn back.
I think this idea to keep talking
during the game was really good by me.
Yes. I play-
I strategize before the game started.
I play best when I have no time on the clock.
So if I take an F7, you take with the king,
so I don't really get anywhere with that,
so I guess I gotta take this pawn.
So I was going for this position. I kind of like it.
Okay. I think my plan here was to go here.
Then your C7 pawn is kind of stuck forever
unless I get the tempo on your rook.
Oh, bishop D5 is always an idea, too.
Oh. Wow.
I think you got some issues here, Levy.
I think you should just think for like 47 more seconds,
and then you should make your decision.
All right. Hmm.
I mean, I have some really tempting moves here.
[Levy] This was just information overload.
You just give too many possibilities and then-
All right, all right. It's kind of hard to decide here.
Let's not try not to win the game right away.
Wow, that surprised me. You're stacking the exchange?
All right. Yes, I will.
[Irina] Okay.
Do we just go for counter play?
Maybe going for counter play was not very smart.
Maybe it was like very not smart.
Hmm, wow. Okay, so I guess we should bring the rooks in.
This is the most fun part of the game,
when you have four seconds.
You just have to- Very fun.
Okay. Does this one look really terrible?
[Levy] Let's just do it.
Now the game could theoretically go, oops, sorry.
[Irina] It's okay.
It's no fun.
No!
Took a really important pawn there.
No! Oh.
And that pawn is also really important.
Why would you take both of my really important pawns?
[Irina] Yeah, yeah, yeah. Okay.
[sighs] And I lost on time. Double whammy.
That was exciting though.
This was probably my best move of the game, right?
In taking my rook, yes. Yeah, yeah.
And B5, I think so- Well, this is
the subtle stuff, Levy, that we, you know,
build a game out of.
Yeah, but I think- Yeah, but it's ugly.
It's ugly because just in general,
like this position, I think, is kind of ugly.
Yes. You know?
You're stuck with this pawn. You're stuck with this pawn.
But question. Weaker king.
Should you play A4?
Because if you don't play A4 now,
can I play A4? Can you play A4?
Yeah, that's a question.
I mean, okay, like, so like,
you mean if I let you play a move like that,
I still think like you're gonna-
'Cause I still feel like now I-
You're gonna have like more problems.
Like even if I have like a pawn here, it's still annoying.
[Levy] But then if I start generating some-
Some play. Yeah, you can try.
A little trap. Yeah, yeah.
Do you see it? Do you see it?
Do you see that bishop? This position is not fun.
It's tough because you got some issues with the king,
some issues with the rook, with the pawns, you know?
But I think, yeah, like in terms of-
It just exposes who you are as a person.
I just have many issues. That's what-
Your move was interesting,
and it kind of posed this big decision in front of me.
Right? Like over here.
So I think in terms of like strategy,
this was really the key moment strategically
because I have to like make this big decision
with the pawn structure, and so this one kind
of prevented your rook from opening up on this file,
and I keep my pressure on the queen side.
So this is a perfect example of time, space,
and harmony all in one move.
Yeah, that's right, that's right.
We're fighting for space.
We're fighting to not let the opponent build harmony.
Yeah, B5 I didn't even consider.
I just thought you would either take or push or-
But yeah, this game went more in my style, Levy.
I listened to you carefully talk
about all the things that you liked about chess
I know. and I tried.
I thought, Okay. It always happens.
Now I know what to do against him.
That's what always happens.
Okay, I have a personal question now.
Yeah. You are
the eight-time United States women's chess champion.
You also have the highest title in all of chess grandmaster.
I'm an eternal runner up.
Like what is the championship mentality,
and how have you succeeded at it so many times?
In terms of the grand title, that's like,
I think, everyone's personal journey, right?
It can take some people longer.
I actually did it very late in my life.
I don't know, Levy, if you know this,
but I made grandmaster when I was 29.
Most people, if they don't do it
by the time they're like 20, they'll probably never do it.
In terms of becoming a grandmaster, really,
that's the time and the commitment.
I think if you're an international master like you are,
like everything is within reach.
Like you've already played chess long enough in your life.
You're close enough to that level that, really,
it's gonna be little things like putting in the time,
going to the best tournaments,
and getting the best competition,
and just maybe getting a good coach to help you as well.
So it's not too late, Levy. Okay.
We are gonna look forward
to Gotham Chess making grandmaster.
Yeah, then I'll retire and, I don't know,
do horse racing or something.
[soft bright music]
Tell me a little bit about yourself. Do you play chess?
So I do play chess. I played as a kid.
I remember learning from my grandfather
when we lived in India,
but I got recently more interested in it
when my kid started playing chess in school.
In New York, kids play chess at a very young age.
Warms my heart. And, you know,
it gets more and more intense, and he's playing tournaments,
and I thought, Let me try to keep up with him.
So I realized that I could kind of do it in my work as well
because I study AI, and I study computer science in general.
What type of AI? Are we talking chat bots?
I don't know the terms. Yeah, there's all kinds.
So I try to create what I call human-oriented AI.
So I try to use AI to understand human behavior,
what makes humans decide
and make the decisions that they do.
And we thought, Wouldn't it be interesting to try
to create an AI that actually tries
to play like humans at different levels
and tries to mimic their behavior?
So that's what we set out to do with Maia.
Generally, the chess board does not care how you feel,
I think is how AI plays, right?
It's sort of just figuring out the best move.
You're right that AI is trying
to play the best way possible,
and it's combining this evaluation function
with this search.
It's powering it with brute force.
Like it can do millions of searches a second.
I think some of the components
of the evaluation function map
to some of the things humans do
when they look at a chess board, right?
When they're thinking about the king being unsafe.
Even that emotion you describe, I feel uncomfortable.
Like I feel uncomfortable whenever I've castled
and I push that G pawn or something.
But that feeling translates to some of the components
of the evaluation function that Stockfish uses.
Humans don't have this luxury
of searching millions of nodes a second.
I can't even finish one node without going,
Oh, what about the second node?
Right, as you get deeper and deeper the tree,
it just grows and grows, expands like combinatorially,
and it starts getting very confusing,
and your evaluation starts getting less and less accurate.
You might make mistakes 'cause you might miss something.
And less and less confident,
which is an emotion which the computer doesn't feel.
One thing that we noticed when I was talking
with a colleague of mine, his name is Ashton Anderson.
He's a professor at University of Toronto.
He was in my lab, and we both had this passion for chess,
but we realized no one had really used AI
to play like a human would play.
I think my audience once pulled 92%
or below the level of 1,200,
but it's really hard to learn chess
because you don't get to experience learning content based
on people who are your level.
Like how did you really deduce how a 1,200 will play?
We used state-of-the-art neural networks
like Alpha-Zero and Leela, but we instead try to predict
what move a human would make given a board position.
These deep neural nets are kind of like black boxes.
So we've come up with interesting ways
to probe at them to get an idea of,
okay, what is actually is going on
that is different about this person
or different about that person?
We had like a style vector for each person.
We can compare those, see where you defer,
and try to map those back
to human understandable chess concepts
because that's the key, right?
If I just tell you, Oh, they defer
in the third number and the fifth number,
you'll be like, What does that even mean, right?
But if I can tell you, Oh I think this person worries more
about their king's safety or sacrifices their center
at the cost of protecting their king,
or something like that, some kind of trade-off,
then you might be able to understand what's going on there.
So that's what we're trying to do now
with these style vectors,
and the fun thing about having numbers representing people
is that you can add them together.
You can mix them. With something like Maia,
do you think it's possible to develop some sort
of accurate measure?
Let's say I'm a 1,200, I have an opening course.
I just wanna practice my opening.
Can I do it some way against Maia or-
I think that's actually a good idea,
and I agree with you that it's not something you
can easily access today like out with the online tools
and the chess platforms that are out there.
I play the Caro-Kann, and when I learned it,
I started getting afraid of, so when you go,
I started getting afraid of
Yeah, the advance. the advance,
and even more so, at some point,
like I always like to get my bishop out
'cause I feel very uncomfortable if I don't,
and then eventually something happens.
This pawn making its way over here used to scare me,
and I never experienced this in any, you know,
looking at the openings in
because I would not know what to do.
I would take it, I would get in trouble,
and even if I let them take, I would-
[Levy] Yeah, you'd have to move your king.
Get to my king,
and so I remember hitting all these variations
that just scared me. Onto to E5 on the third move
is the most challenging move against the Caro-Kann.
It restricts the knight and the main line.
So the move is bringing out the bishop,
Bringing out the bishop. so you don't get locked.
I play that move,
but the move that I also added to my toolbox
and is the only move I teach now is this move
to try to undermine the center.
It's not commonly played at the highest level.
If you study master games,
this is probably one in every six games
or something compared to, right?
So now we go down a totally different path.
We're taking a totally different train to our destination,
but just this undermining of the position,
you know, splitting up these pawns
and then slowly capturing them both,
that does wonders at intermediate and early advanced level.
This move, despite it being move three,
according to databases,
already gives black like a 53% chance of winning,
which is crazy because normally white is about 52%, right?
White just doesn't know what to do about this move.
Is that like human empirically, though,
based on the games that people are playing?
Yeah, like I would say-
Is the Stockfish eval like weaker for black?
Computer, Computer will say-
after like 12, 13 moves, says white is better.
White is better, okay, okay.
But what does that really mean
Right, in practice. in practice?
Even in grandmaster level chess, okay, move 13,
white is better, and move 20, black is equal.
I love this because this is an example
of the kind of advice, as you said,
in the intermediate and advanced level,
that could be appropriate because you're thinking
about the human opponent that you're playing.
And maybe you wouldn't recommend this
at the highest, highest levels, but in that range,
I think it could very well be very good advice,
and as they get more comfortable with it
and understand it, your students, at some point,
they might realize, You know what?
I've progressed to a point
where I don't wanna play that anymore,
and they'll change, but for a good part of time,
it could be very useful for them.
In 1997, Kasparov lost to Deep Blue.
We can say for about 26 years now,
AI has been better than humans in chess,
but where do we go from here?
Thinking about chess, first of all,
I have to hand it to Alpha-Zero
and Leela and these other engines because I think they
have sometimes given us interesting ideas.
They've shown sparks of creativity
that have inspired the way humans play,
but I agree with you that, right now, you know,
you look at the the TCAC competitions,
and they're just kind of getting better and better
and a little bit better, a little bit better,
a little bit more data, maybe a deeper neural network,
and, you know, these advances I don't find that interesting.
I'm more interested in trying to get better and better
at creating human-like presence and human-like avatars
when it comes to chess or when it comes to other games.
So I feel like, you know, within five years,
we'll probably have a lot of human-like entities
that we can interact in in virtual settings,
not just because of the rise of large language models
and the ability to converse and speak
and produce text like humans but hopefully the ability
to produce actions like humans, too.
I hope you enjoyed this episode of Five Levels,
and, before you go, I just wanna say that it's probably one
of the most unique episodes because our ages
did not go up from low age to a higher age.
Chess doesn't really care about any of that.
Just give it a try.
It is ever evolving,
and the learning journey is really never over
when you play chess.
[bright soft contemplative music]
Musician Explains One Concept in 5 Levels of Difficulty
Expert Explains One Concept in 5 Levels of Difficulty - Blockchain
Oculus' John Carmack Explains Virtual Reality in 5 Levels of Difficulty
Biologist Explains One Concept in 5 Levels of Difficulty - CRISPR
Neuroscientist Explains One Concept in 5 Levels of Difficulty
Astronomer Explains One Concept in 5 Levels of Difficulty
Laser Expert Explains One Concept in 5 Levels of Difficulty
Sleep Scientist Explains One Concept in 5 Levels of Difficulty
Physicist Explains One Concept in 5 Levels of Difficulty
Astrophysicist Explains One Concept in 5 Levels of Difficulty
Hacker Explains One Concept in 5 Levels of Difficulty
Nanotechnology Expert Explains One Concept in 5 Levels of Difficulty
Physicist Explains Origami in 5 Levels of Difficulty
Computer Scientist Explains Machine Learning in 5 Levels of Difficulty
Neuroscientist Explains Memory in 5 Levels of Difficulty
Computer Scientist Explains One Concept in 5 Levels of Difficulty
Astrophysicist Explains Black Holes in 5 Levels of Difficulty
Computer Scientist Explains Fractals in 5 Levels of Difficulty
College Professor Explains One Concept in 5 Levels of Difficulty
Quantum Computing Expert Explains One Concept in 5 Levels of Difficulty
Computer Scientist Explains One Concept in 5 Levels of Difficulty
UMass Professor Explains the Internet in 5 Levels of Difficulty
Mathematician Explains Infinity in 5 Levels of Difficulty
Theoretical Physicist Explains Time in 5 Levels of Difficulty
MIT Professor Explains Nuclear Fusion in 5 Levels of Difficulty
Harvard Professor Explains Algorithms in 5 Levels of Difficulty
Chess Pro Explains Chess in 5 Levels of Difficulty (ft. GothamChess)