Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 40 original podcasts with new episodes released every day.
Towards Beating the Engine
This post is about chess, but chess as it bears on a larger question. How do you play against a computer that’s smarter than you? As time passes that question seems more and more important. The general case is vague enough to make things difficult to answer, so we’ll take a good bit of time to look at playing against chess computers. In the grand scheme of things it might not matter whether we win or lose over the board, but it matters a great deal how raw humanity can compete with silicon intelligence.
In 1997 then-reigning world champion Garry Kasparov lost a chess match against the IBM supercomputer Deep Blue. In 2006 Kramnik (who had taken the title from Kasparov) lost a match to a program running on ordinary computer hardware. Since that time the engine has reigned supreme. I’ve asked my local chess club whether a human might ever beat the engine ever again. I’m the only one who sees even a glimmer of a possibility.
How did We Get Here?
Thousands of years of human dominance. A few decades of weak competition, a few years of struggle for supremacy. Then, game over. -Garry Kasparov, who, on the whole, really is an optimist about technology.
Alan Turing, the famous computer scientist, invented the first chess engine. Simple as it was (only evaluating the moves currently available) it was still too complex to run on the computers of his day, so he played a game with it, working out the calculations by hand. It wasn’t a great game — as far as chess goes — but it was at least recognizable, which makes it a stunning achievement. As Samuel Johnson noted of a dog walking on its hind legs: “It’s not so much that it does it well as it does it at all.”
I bring up Turing’s engine because the logic is easily comprehensible to us humans. It works sorta like this:
For each move on the board, if you make that move:
- Does it capture a piece? If so, assign points for the value of that piece.
- How much room does it allow your pieces to maneuver in the future? Assign points for potential activity.
- How well protected is your king? Assign points for good defense.
Sum up the points for every possible move you can make, and then choose the move with the highest score.
Even Turing’s original engine was more sophisticated than that — weighing more considerations — but that’s enough to go on. (If you’d like the specifics you can find them at the linked videos here and here.) Later chess engines also run an evaluation for the moves your opponent makes, to see how good they are, and so on. For the interested nerd, the phrase “to Google” is “alpha-beta pruning“. The rest of us will blithely march on and assume the computer knows its business. We’re also going to skip over most of the history of human-computer chess. If you’d like to know more about that, I highly recommend Kasparov’s 2017 book Deep Thinking.
In 2016 a new challenger appeared. At that time the best engine in the world was known as Stockfish. Google trained an algorithm called AlphaZero using machine learning techniques. It wrote a program to play chess against itself over and over and over again until it got good. How good? Good enough to feed Stockfish its own pants. Google then diverted into other areas of research. The engineers behind Stockfish rebuilt their engine to incorporate machine learning principles (and how that works, I haven’t the foggiest), and these days the top-level computer versus computer chess is played between Stockfish (who reigns once again) and an open-source project called Leela. I’d love to get Kasparov’s thoughts on this but publication dates being what they are, I’ll have to wait for a sequel.
Thinking about the Computer’s Thinking
If you know your opponent and you know yourself you will be victorious in a hundred battles — Sun Tzu, who’s never been down a rook and a bishop and forced to trade queens.
During the decades between the advent and the triumph of computer chess, engineers were caught in a dilemma. Take another glance at that algorithm. You can refine it two ways. Either you evaluate a given move more precisely or you calculate one move deeper. Call it a sniper versus machine gun approach. A sniper will target one bullet exquisitely, being pretty sure he’s going to hit the right thing. A machine gunner will spray hundreds of bullets at the target, also being pretty sure he’s going to hit it with at least one. It’s hard to do both.
A chess algorithm has a calculation it does to evaluate a given board position. This is a single shot in our analogy. The more precise the calculation, the more accurate the shot is. However, the more accurate each individual shot is, the slower it is to calculate. Alternately you could make the calculation very simple and not very accurate, but you can run it over and over again for a deeper set of results, machine-gun-like. If you increase the accuracy of any given shot, you decrease the rate of fire. If you calculate more positions, you need a more efficient and less accurate calculation.
An engineer can happily spend an entire career arguing with that dilemma, but we’ll keep things simple. You know what else goes up exponentially? Computation power. Without denigrating the hard work put in at every level of these problems, part of the solution required waiting for the right hardware. We’ve gotten further than any human can calculate. But are you sure it’s correct?
The only way to check that an engine is correct is to feed the position to another engine and see what it says. If they disagree, how do you know which is right? If they agree, are you sure they’re correct? Since the engine is stronger than we humans there’s a tendency to imagine it’s infallible. You can’t think that way if you’re hoping to beat it.
The engine excels at tactics. Given a position, it can accurately calculate dozens of moves forward, but it can’t calculate everything. On the first move of the game, white has twenty legal options. Black has twenty options as well. After each player has made a move that means there are 20×20 = 400 possible board states. Extend that out to, say, thirty moves, and people start throwing around phrases like “the number of atoms in the universe”. The engine can’t calculate every single possible move, so it has to cut it down. How sure are you that you’ve cut out only the inferior options?
Weaknesses in the Engine
No; not a chance – Magnus Carlsen on whether he could beat his phone.
Practically speaking this leaves two weaknesses in the engine. The first is that, despite its mastery of tactics, it lags behind humans in strategy. Suppose the engine evaluates the next twenty moves. If a given maneuver will only pay off twenty-four moves down the line, the engine will overlook that. Kasparov used an “anti-computer strategy” in a brilliant game against a chess computer in 2003, deliberately stalling the game where he could improve his position slowly while the machine couldn’t. Here’s a recent game where the engine may have misevaluated a strategic position, showing things to be dead even where white has a definite advantage. Or maybe the engine sees something we don’t.
The second is that, to deal with the exponential options, the engine is forced to prune off certain branches, ignoring moves that don’t look like they’ll have a good outcome. Here’s an example video from a couple years back, where the engine misses the winning move because it pruned off the wrong branch.
There’s a fundamental problem with trying to exploit these resources. Computation power is cheap. The default you-haven’t-even-signed-up-yet engines on Chess.com or Lichess will calculate twenty moves deep right in your browser. While it may still be possible to find small advantages that build up over the long term, the further out the computer can push that long term the harder it gets. At some point you run out of long term; for better or for worse most games are over by move one hundred.
The better the position evaluation, the less likely the machine is to ignore whatever small advantages you might see. In the final game against Deep Blue, Kasparov laid a trap for his opponent, figuring the algorithm wouldn’t weigh long-term problems accurately against a knight gained now. Deep Blue avoided that trap, which cost Kasparov the game. In the decades when man was still superior, the engines made steady progress because they were advancing both in power and sophistication, in accuracy and rate of fire. Maybe they still are; it’s kind of hard for a human brain to tell.
Should a human ever be able to challenge the machine in a fair match again, that gives us an idea as to how he’d have to do it. He’d have to start with world champion-level tactics to at least stay alive against the traps the machine would lay. Then he’d also have to have a brilliant grasp of strategy, making moves that give him benefits far down the line, such that he can use those tactics not only to stave off defeat, but to slowly build advantage. Then ideally he’d also be on the lookout for wild and crazy moves that the engine might overlook. I said up top that I see the glimmer of a possibility of it happening, but even that dim hope marks me as insanely optimistic. On the other hand, that doesn’t leave us entirely without options.
If You Ain’t Cheating you Ain’t Trying
If at all possible play chess with your opponent sitting in front of a mirror. That way you can see his pieces. -Carlin. Not the comedian you’re thinking of, the one in my local chess club.
I’d prefer humanity to win in a completely fair match. There are other options. It used to be common for stronger players to offer an odds match to their opponents; starting down a piece to even the game. We could do something similar for machines; restricting them to calculations within a certain time, or using a limited number of cycles, or even to only burn as much energy as a human brain. I like this idea because it creates interesting problems for the engineers, but I’m less sanguine about it because victory is kind of hollow when it comes to setting the dial.
Two more quick ideas I gleaned from Kasparov’s book: One is to remove the computer’s opening book. Human chess players have figured out the best first couple moves of the game (depending of course on what your opponent does). These are the moves where strategy matters most and tactics the least. By denying machines access to recorded memorized sequences, you’re forcing them to make calculations where their calculations are the weakest. The other idea is to take the knights off the board. From novice to grandmaster, the tricky ways knights move seem to mess with human computation. A more linear game of chess is easier on our brains.
Instead of making things easier on us, you can make things harder on the machine. The engine is a computer program. Humans input data and the program spits out a move. Suppose I insisted the machine play like any other player. The designers have to build a robot capable of imaging and reading the board, doing the computation, and then gripping and placing pieces themselves. Then the problem designers have is several times more complex than merely building a better algorithm. You could insist the machine was forfeited due to mechanical failure. The difficulties I’m adding here are solvable — similar things are routinely done in an industrial context — but the principle is worth remembering: The more accommodating the human is toward the machine, the easier it is on the machine designer. The more the machine works to accommodate the human, the harder it is to build. Though I suppose you could even things out again by insisting the human play from inside a Mechanical Turk.
The Hippy-Dippy Solution: Can’t We All Just Get Along?
AI deserves to lose. -James, the Bartender
Much as I hate to indulge in that kind of talk, there really are some interesting things that happen when you redefine the question. Do you know what happens when you pair a human and a computer versus a computer? I don’t. Kasparov talks about an open tournament where any combination of human and computer teams could enter. The strongest computer didn’t win. The strongest grandmasters with their engines didn’t win. The winners were a pair of ordinary dudes running three off-the-shelf engines. They were better at working with the technology. That’s an interesting result, but does it still apply with machine learning engines? I’d like to see the experiment run again. And again. There’s plenty to explore.
On the other hand the idea of using a computer to cheat against a computer puts me in mind of my junior high football career. At some point you cross the line from providing questionable “help” to actively screwing things up for the better players.
There are ways we can learn from chess computers. We’ve run the great games of the past through the engines to see what we’ve missed. One surprising thing we learn here is that the moves made over the board, under tremendous pressure, are better than moves favored by analysts studying at their leisure afterward — even when it’s the players themselves doing the analysis. Simply having a machine capable of doing more accurate calculations than us gives us an opportunity to learn about how we ourselves think.
Some of the engine’s ideas can be stolen and used by the hypermodern Prometheus. Computers like to push their H-pawn. This was considered a bad idea by right-thinking strategists for centuries, but you see it creeping more and more into human play. When I’m sitting across the board from another player, what does it matter that a computer can beat us both? Maybe if I steal some ideas from the machine I’ll be able to crush the man in front of me.
The game Go has resisted the same sort of machine dominance, partly because it’s harder to make a good evaluation function for a given position, and partly because the 19×19 board requires the engine to calculate more options each move. In our gunpowder analogy it’s both harder to snipe and requires more machine gun bullets. Google’s AlphaGo beat the world champion in 2016, almost twenty years after Kasparov lost. It did so because the machine learning algorithm it used deployed brand new tactics, things that the masters of the game hadn’t even considered before. That’s cool and all, but the story gets really wild in 2023.
Kellin Pelrine, an American Go player and not even a top-rated professional, beat a top-ranked engine. He did so by using a computer to analyze the engines for weaknesses, discovering a strategy that might work against them, and then deploying it entirely with his human brain. Perhaps the models can be trained to defeat this strategy, but there might be more chances besides that out there. Perhaps I should go back to the top and rewrite this whole post that way, but I’m a rank amateur at Go. The most I can say is that it validates my idea that computer dominance is not inevitable.
Playing a Different Game
I don’t mind losing to my computer at chess because I still win the rematch at kickboxing. -Jack Rysider, a guy whose podcast I listen to only partially for dumb jokes.
We can make it impossible for the computer to win by changing the game. You may beat me at chess but I’ve decided to play Tic-Tac-Toe instead. Yeah, I know Tic-Tac-Toe is a solved game, and that in principle I can never do better than a draw against a computer, but it seems you weren’t programmed to play that game. Guess I’ll win by default. Any five-year-old employs this strategy under similar circumstances. When she decides she can’t win, she’ll say, “Let’s play a different game.”
Suppose the engineer comes back with a chess engine that also has a Tic-Tac-Toe solution stapled on. I challenge it to checkers, but he’s thought of that, and it beats me at checkers too. And Go (remember, rank amateur), and Risk, and Monopoly, and even when I try Candy Land, an unlucky shuffle of the deck means the computer won even though neither of us, strictly speaking, made any choices. Board games as a class are relatively easy for a program to solve because the real-world complications are necessarily abstracted away. If you lined up real-life soldiers like chessmen, the resulting donnybrook would look nothing like a game. In real battles no one takes turns. By stating my original goal as beating the chess engine, I’ve already conceded most of my chance because I’m facing the computer on terms that are incredibly favorable to the computer.
This needn’t be true in all circumstances, even in all games. Last I’ve heard the best human players of Starcraft II can still beat the best AI competition. You’d think a video game — especially one reliant on precise timing and superhuman reaction speeds — would favor the machine, but apparently not enough. I have my doubts that we’ll ever see a computer able to play consistently professional Magic: The Gathering. Once you’re dealing with questions that haven’t been deliberately abstracted down, the computer’s advantage diminishes rapidly. Granting that SkyNet thought of nuking its enemies back home, it’d have a hard time winning the resulting war because war is a game that greatly favors those who innovate.
As fascinating as I find games (almost certainly more than my audience — thanks for sticking around), we should look at the rest of the world. The computer from War Games is right; the only winning move is not to play. Why do we really care if the machines are better at chess? Even if the computer’s dominance ruined the fun of me losing to other humans, I could leave that field behind and do something else with my life. But for how long? That’s the lingering fear behind all worries about automation; what do you do when your job is automated away? When every job you can do is automated away?
Marc Andreessen is fond of reminding us that software is eating the world. It’s cheaper to solve a problem in software than by getting a man to do it. However difficult the problem is, once it is solved you can copy and copy the solution. Hence, anything that can be solved by software will be solved by software. This worry has grown more acute since large language models burst into the public consciousness. The picture might have six fingers on the right hand, but I’m already seeing better pictures. Are you sure you can tell the difference between AI-generated art and stuff made by humans? If machines can take over the art world, what else is there?
I’m honestly not that worried. Because the Large Language Model that’s spitting out — let’s not say art, because I’m not going to be competitive in the drawing world absent a last-man-on-Earth scenario — computer code might do an amazing job at replacing me. But that AI is going to share some DNA with the chess engine. It’ll determine what to put where based on a large calculation of rules it derived from its training set. That calculation is going to have weaknesses, weaknesses I can potentially exploit.
Perhaps I can out-compete it directly, by applying my superior strategic understanding to write code that’s actually adapted to the problem at hand, not just the computer’s understanding of the problem. Perhaps I can use the LLM to write some of the code, trusting it to handle the tactical problem-solving while I focus on the larger goals. Perhaps I can just write a better algorithm than the AI because it’s pruning the wrong branches when it’s optimizing it’s decision tree. Perhaps I can switch to a field the machines are worse at: gardening, or baseball, or con artistry.
It may be that a world is coming where people are automated out of all usefulness. That they won’t be able to get a job because there are literally no jobs that they can do profitably. That world isn’t here today. It won’t be here tomorrow either. If they announce tomorrow that they’ve got a new algorithm that can do everything I can do, but better, I still won’t trust it. Algorithms, even unbeatable ones, have their weaknesses. And that’s why I find it worth considering how I’d beat the chess engine.
Published in Technology
Great stuff. I’ve been very interested in seeing how AI will be adopted into emergency medical care. I assume the major weakness is that the software can only perceive what we tell it; weak or missing inputs will deny it the ability to do patient assessments better than humans can. I wonder how much intuition really helps humans; of course the problem is also defining what we label as intuition. Is it just preconceived notions we bring to situations or subtle cues we unconsciously receive?
Anyway, great post.
Outstanding post – and the right length too!
I’ve been reading Nate Silver’s book The Signal and The Noise, and its chapter on Kasparov and Deep Blue is most instructive. I will have a post about the book in a few weeks. What I say about that chapter will complement this post: there’s more than one way to program a computer to play winning chess, and there’s more than one way for a human to mess with its mind, ha! (My words, not Silver’s or Kasparov’s or IBM’s.) For now, I’ll just remark on a couple of things I came across. One is that there are many endgames whose outcomes are inevitable – only you have to have a midgame that gets you to that threshold first. Even from there, it may take over 200 moves to checkmate. The other – which was in a footnote, but Deep Blue’n’me, we read those things – is that there is a rule of chess that states that if there is a 50-move stretch wherein no piece has been taken and no pawn has been advanced, the game is declared a draw.
I did not know that. I have never come close to knowing that!
I liked this quote by Kasparov:
“It turned out that making a great chess-playing computer was not the same as making a thinking machine on par with the human mind,” Kasparov reflects. “Deep Blue was intelligent the way your programmable alarm clock is intelligent.”
I used to play against my Atari. On the easiest setting, I could beat the machine. On about medium difficulty, the machine started winning frequently. On the hardest setting, not only would I lose, but the lights on the Atari would blink and it emitted an eerie “Ye haw!” sound . . .
Anyone here play Battle Chess on the computer?
Sure you can beat the computer. Early in the match accidently drop your glass of water on it. Game over – You win.
I always think about the Marines who were tasked to sneak up on the robot guardian and went all Solid Snake on it.
I’ll grant that you could probably teach it to open fire on giggling boxes but even so it still makes me laugh.
One could also take Zapp Brannigan and the Killbots as a way to overcome the machines but it’s not ideal.
I have a good deal of faith in human intuition. It’s certainly not always right, and people can be fooled, but it’s also solved problems that brute calculation can’t.
Magnus Carlsen in his Rogan interview says that if he were cheating with a computer you could never tell from how he played. He’d play normally at world champion levels, and then when a doubtful move came up he’d consult the engine. That would be enough to make him unbeatable.
I think there’s that potential for doctors. Ignoring for the moment all privacy and regulatory concerns, if you churned everything you knew about a patient through a large language model then the machine gives you a diagnosis. Is the diagnosis correct? Probably. (Also assuming here the engineers know their stuff.) Some times the box is going to be wrong. Sometimes the doctor is going to be wrong. Those won’t be the same times. A doctor, consulting with (and not blindly following) the machine will have a better chance at accurate assessment.
Kasparov talked about that some; an end game book that, having once solved a perfect ending, keeps the moves in a database so that it can replay that without thought. I think as of writing they had one that could accurately finish any game so long as there were no more than six or seven pieces on the board. After that you’re running up against an exponential curve once again.
In the world of human chess there are plenty of known end games where one side will win, or can’t be pushed for more than a draw. High level players will generally agree to a draw, or one player will resign when his position is hopeless Of course, that causes frustration in lower level players (like me) when someone resigns a game and you can’t see why. “He’s just down a pawn!”
Yeah, that puts a cap at the longest possible game of chess, at 3,500 moves. Both players have to be colluding to get it that far. It works like this. You stick your right knight in, you take your right knight out, you stick your right knight in …
Right, the essential details. At move fifty one player has to place a knight such that the other player can take it with a pawn, doubling two pawns on the same rank. At move one hundred the other player has to double his opponent’s pawns opposite the gap. So on until both sides have all eight pawns that can advance without knocking into the other guy’s pawns. After that it’s fifty moves per piece, fifty moves per pawn advance, promotions all around, and fifty moves until each of them gets sacrificed in turn. On move 3,500 a king takes the last promoted pawn, and the game is drawn for lack of material.
There’s one wrinkle in here. The game is drawn if a position is repeated three times. I don’t think that’ll be a problem with both players colluding, but I haven’t played one of these games out to be sure.
This is a thoroughly useless fact, but I worked it out earlier this year because I was curious and now you have to share.
I played the chess program built into the seat back computer last time I took a flight. I could consistently beat it, which tells me that it isn’t very good.
I’ve done Russian-English translation on the side for a while, and AI has really eaten into that business. Google Translate or DeepL can do an understandable translation of many language pairs, so many people use it for non-serious translation. But AI can’t produce human-quality text in the target language. AI-produced translation is not pleasant to read, is often not smoothly worded, and sometimes misses the point of a sentence completely.
But it takes real skill for a human translator to produce a target text that doesn’t read like it was translated.
The story for the rest of y’all, so far as I remember it, was that DARPA had an automated sentry to try out. What do you do to see how something breaks? You hand it to the marines. A group of marines was tasked with trying to evade robot detection. As it turns out the robot was trained on normal human behaviors, like walking. It didn’t detect alternate means of locomotion, like one marine approaching in a series of somersaults. Or yes, two marines crawling in a box, audibly giggling the whole time.
There’s another, less funny example I heard about. DARPA had posed a question to a LLM; does this photo contain a tank? They managed to get real high accuracy out of their training model, but failed abysmally when they tried it out on a larger scale. As it turns out the training photos were taken on two different days; one day with the tank, one day without. The AI was keying off of different lighting in the photos, not the presence or absence of tanks.
The lesson here in both cases is to be very careful what you’re training your LLM on, because the real world won’t always mimic your training set. Though I doubt machines will ever really be able to predict marines.
To go back to your chess example, what if someone played a totally off-the-wall game against the computer to force it to examine all the branches that it would normally prune off. Could Magnus Carlsen beat the computer with such a strategy?
It certainly has gotten better over time. Question for you; do you ever use machine translations in your work? As in you read the document once in Russian, then use Google Translate to turn it into English, then polish the unfortunate spots?
I don’t use machine translation for an entire document if I want a polished translation as the end product. I find that I spend as much time editing machine translation to polish it up as I would if I just typed the translation myself. A lot of the editing of machine translation requires you to rewrite whole sentences, so it’s quite a bit of work.
Then they started adding APCs and mobile artillery and it all went to pieces.
I don’t think so. The first problem is that, however weird the position on the board, it’s something the machine has the ability to calculate. Pieces, positions, moves, whatever the board looks like you could write out a list of legal moves and check each one. To do something truly unexpected to the computer you’d have to define something other than ‘checkmate’ as winning, like @tigerlily who’s happy to settle for a machine forfeit.
If you tried a crazy game after that, then we’re still running up against Moore’s Law. On average a chess board has 40 legal moves (not sure how that’s calculated, but let’s run with it.) If the machine has to evaluate every one of them, then calculating five moves on each side you’re looking at roughly 10^16 positions, a number so large it dwarfs our national debt. We’re not going to let the computer think any longer than that, so let’s restrict it’s future vision to five moves for each side ahead. That’s still a tremendous amount of calculation to throw at even wild and unpredictable moves.
The next problem comes from what makes crazy moves crazy; they don’t generally work. A good move is definitionally good because it advances your position or gains material. An off-the-wall move can be good, but more likely it’s just bad. Hikaru Nakamura can make the bongcloud attack work, but I doubt he could against the engine.
On the other hand, an engine isn’t a player either. I don’t know how prone engines are to caching previous results. If it calculates five moves deep once, and you make one of those moves, does it start calculating from four moves in, or does it have to start all over? If you’re playing an odds match (as described in the post) forcing it to recalculate every time might waste enough CPU cycles that it’ll blunder.
Interesting. I wonder if that’s a solvable problem or not.
The solution (for Now) to defeating AI is to engage it in 2nd or 3rd order analysis. Hearken to Sung Tzu: and don’t engage the enemy on the ground of his choosing. I admit I am not exactly sure what this would look like with a chess match. Maybe something like playing chess with each player only moving the his/her own players that start out on the left side of the board.
You start with what you can expect the AI to know, and go beyond it. Particularly in the area of best practices, lived experiences, etc. AI’s are like the genius 6th grader in college, like Sheldon, or Doogie: Very intelligent, but no experience.
Essentially. If you’ve conceded most of your advantage by confining yourself to sixty four squares, then why confine yourself to sixty four squares?
Three of my spouse’s books have been translated into Italian and German, with hopefully more to come.
I know very little Italian and German. He has HS German, but no Italian. So we were fortunate that outside forces in those 2 countries were interested and undertook the translating job.
But after it became necessary to view Mark’s books as potentially being translated into other languages, as the copy editor, I had to go through each of his 13 books and clean them up. Every modern language speaker and reader has their own vernacular way of structuring statements. If a book is written for us English speakers, but ends up in the translation process, prior to that process, it is necessary to delve into each sentence and make sure that the ways that we sometimes “skip” into a meaning of a statement has that “skip” spelled out quite clearly. That way it will be less likely that the meaning of the text will be lost when translated.
One other way to think about the big job AI will have with regards to translation: in 1979, when I spent the summer in Norway, one errand I had was to go to an auto supply store and purchase WD-40. Now Norwegians were big on American stuff. In grocery stores it was often the case that a case of Campbell’s soup was available.
So my initial hope was that I’d walk in the store, browse the shelves, and see WD-40 just sitting there. But that was not the situation.
My second hope was that someone in the store spoke English. But unlike people closer to the local university area, this more remote store had only Norwegian-speaking workers. It took a long time to try and explain in my faltering beginner’s Norwegian that I wanted an oil that wasn’t an oil, but more than that. Had cell phones been available with translation abilities, I could have looked up words I did not know in Norwegian like lubricant, rust preventative, moisture displacer. Luckily, my goal was, after some pantomimes, accomplished: one worker figured out that a can of a very popular and necessary “oil that was not a motor oil” possibly would meet my needs.
AI will have to stumble through this type of thing. It is said to learn quickly so it should eventually get there, depending on how well whoever is doing its programming understands the processing protocols for translating written text.
I wonder if writing out “WD-40” would have worked. Perhaps the locals would recognize the brand name and find you a substitute. Impossible to tell at this late date, of course.
I think AI will have different stumbles. One strength of a computer is the ability to remember a great many things. I’d imagine that the machine would have all the words for lubricant and so on but wouldn’t know to use them, at least without some coaxing. But I don’t know. It’s not a field I’ve looked in to all that much.
I think that computers do an amazing job with “art.” But there are some limitations. I’ve been working with a researcher in England, Professor Hassan Ugail, on artificial intelligence learning to distinguish between the paintings of different Old Masters. In my estimation, the computers can tell the difference between artists to a reasonable degree, but not as well as a highly-trained human expert, or a top-notch artist (like me). For instance the computer has little difficulty telling a Rembrandt from a DaVinci, but runs into difficulty telling a Rembrandt from one of his students who paint like him.
Professor Ugail ran a control by having me paint a copy of Raphael’s self portrait in the Uffizi Gallery and see if the computer could tell if it was by Raphael or not. The score turned out to be 1/3 – yes, this is by Raphael, and 2/3 – no, this is not by Raphael. Considering I had to learn Raphael’s actual painting technique quickly within a few days (even though I had been studying him generally for some time), and was under a time restraint of three days to complete the painting from start to finish, I figure I could have fooled the computer a little better had I more time. The results can be found embedded in this academic paper.
Screenshot
Another project I got roped into was recreating a lost masterpiece by the Renaissance artist Titian, that was destroyed by a fire 300 years ago. A company that is trying to recreate lost masterpieces in general, hired me to paint this picture. The company (whose name I will not reveal due to circumstances I choose not to explain), was trying to recreate works of lost art using Artificial Intelligence. They tried to recreate the images of 12 Roman Emperors, a complete series by Titian that had been lost in the fire. The series was so popular in Titian’s day that numerous artists painted copies of them before they were destroyed. The only trouble was that those painters were not as good as Titian and their copies did not reflect Titian’s particular technique.
After a year of trying, the company reached a point of diminishing returns. The computer’s digital images looked pretty good, but not good enough to convince anybody that Titian had painted them. So they hired me to do it the old-fashioned way, like an art forger(!) I did the first painting in the series, Julius Caesar, here:
Now I’m going to show a close-up of the shoulder and compare it with a close-up from one of the main copies from the Renaissance, and the best digital image the artificial intelligence could come up with after hundreds, if not thousands, of iterations.
As you can see, the computer chose an altogether different pattern to put on his shoulder than the Renaissance copy and mine, and it even put laurel leaves on his shoulder where there is supposed to be a knot. Aside from these errors of subject matter, the computer failed to make it look like a Titian painting in the detailed close-ups that I compare here:
Missing is any hint of Titian’s canvas texture which is considerable because he painted on very heavy weave canvases. I had no trouble because i was painting on actual heavy weave canvas. The AI generated patterns look somewhat flat close up, and although the computer tried to emulate some brushstrokes, they look nothing like Titian’s paintings up close.
Now in my “other life,” I happen to be a nationally ranked chess master. The computers that us chessplayers used to laugh at in the 1970’s and 80’s, are now monsters incapable of being beaten by a human. I don’t share your optimism that a human will eventually be able to beat the best computer program, but hey, it’s worth trying! Who knows?
A friend of mine is a Chess Grandmaster and four-time U.S. Champion who learned by studying with former World Champion Mikhail Tal in Latvia. I think it was close to ten years ago (when the computers were not as strong as they are now) that he told me “Ya know, Steven, I found out that all my openings are bad, according to Stockfish (best computer chess program). I have to change all my opening preparation.” (Grandmasters and even lesser players spend a lot of time on “opening preparation”, that is, memorizing lines of play in the openings that have been analyzed and tried out and refined by top players for at least 150 years)
That he was going to revamp his whole opening repertoire blew me away! I use Stockfish to analyze games that I play against a lesser computer program built-in to my Apple computer. The moves and resources it sometimes finds are mind-boggling. You mentioned that it has a penchant for playing pawn to h4. I just recently found that it thinks the best move against the Winawer French Defense is to play h4 after these moves –
1. e4 e6
2. d4 d5
3. Nc3 Bb4
4. e5 c5
5. a3 Bxc3+
6. bxc3 Ne7
7. Pawn to h4!…
Ten years ago people would have thought you were either an amateur or a coffee-house player if you played such a move!
I think that the computers are much better at chess than they are at art. That is because art is many times more complicated than the game of chess. You mentioned the exact number of possible opening moves in the first two moves being in the hundreds, and then it quickly reaches thousands and millions and gazillions. In art, the average human eye is capable of distinguishing between roughly one-million different colors and tones. That is the starting point for just one aspect of art. Then there is subject matter that has to be more than a million times that. And then how you arrange that subject matter (composition) which has got to run into the trillions and then all the different combinations of color, subject matter, and composition together in unison. And I forgot to add particular technique and style of painting or drawing, too. It gets very tiring………..but there is still yet hope for the Human Race!
You have quite a collection of talents!
Thanks! I can also make fart noises with my clasped hands but nobody has written a post about that subject for which I could add a comment or two………….
It came up this weekend that my sister-in-law uses AI to translate some materials for church. She’ll use ChatGPT (preferentially to Google Translate) to translate into Spanish, then correct mistakes. This is the opposite approach to what Mr. Fast was describing, I think because she’s more willing to tolerate mistakes. I wondered up above if truly accurate translation was a solvable problem. I wonder the same thing about picking up Titian’s brush.
I’ll wager you could fool me. Show me an AI copy and say that this is the real one, that the other is a forgery and maybe I’d believe you. I imagine it looks like a chess game in the movies does; the pieces and the movements are enough to fool anyone who doesn’t play but a glance for you and I is enough to see that the board is sideways. An AI copy of a painting will make errors that no human forger would.
If I’m making a board game I could see using AI to produce generic art for the cards. “Draw Andrew Carnegie as if he were illustrated by Jack Kirby”. Pick one I like, and we’re good to go. If I’m trying to reproduce the masters I’m going to need a much more accurate result. I suspect your painting and Mr. Fast’s translation are butting up against the same problem.
This might be a problem with the AI training set. I know they scrape up thousands of comics and such off of the internet in order to train the model. “In the style of Studio Ghibli” is relatively easy; you’ve got hours and hours footage to feed in to the machine.
I wonder if you painted thousands and thousands of paintings on different canvases and fed that into the model, then might it be able to differentiate that in the input and mimic it more faithfully? You wouldn’t be saving yourself any effort.
Neither do Garry Kasparov and Magnus Carlsen. You’re in good company.
A friend of mine was reading a book on an opening. He got to the end where the author said “however I’ve had to stop playing it because the engines have refuted it” and wondered what the point was. I’m guessing the work was mostly done before the refutation.
At the level I play I know very little opening theory. Perhaps it’s something I ought to study more, but I think I’m better off studying how to play well. There is something to be said for all those lost openings; if they worked for a century before computers then they’ll work just fine against humans now, at least humans who aren’t preparing at the GM level. Contrawise, if you look at the Fried Liver attack, the computer says it’s dead even even though the position looks so much better for white. Perhaps there are openings that have been previously bypassed that are in fact viable.
We’re also seeing a push towards Fischer Random (or Chess 960, or Freestyle) chess at the highest levels. The kind of thing which takes away opening preparation. Here’s a game from a tournament that just concluded. I’m leaning more and more towards trying that out because of the importance of that opening book.