I.
In 2019, artificial intelligence pioneer Rich Sutton published an observation that wounded the pride of his entire field. After studying 70 years of AI research, he found a pattern nobody wanted to admit: the simple, brute-force approaches always won. Not the elegant algorithms. Not the clever shortcuts. Just dumb methods of throwing lots of data and compute together.
He called it “the bitter lesson” because it stings to admit that just throwing raw scale at a problem beats human brilliance.
But the more I’ve thought about it, the more it feels like everything can be explained by the bitter lesson - that just meditating upon the implications of it feels like unlocking a profound truth about the universe.
I used to think the world ran on genius- that clever people shaped the world the way a potter shapes clay.
You know the story: brilliant mind has breakthrough, world changes forever. Newton under the apple tree. Einstein in the patent office. The romantic image of someone thinking really hard and unlocking reality’s secrets. I’ve often imagined myself in that place, and I’m sure that everyone who has experienced the thrill of finally figuring something out that’s difficult- have often imagined the same.
Then I started paying attention to how things actually work.
My phone or event the most popular apps everyone uses? Product of millions of A/B tests. The language I speak? Nobody designed it. The food I eat? Bred through thousands of generations of trial and error. Even my favorite songs probably topped the charts because an algorithm noticed people replayed them, not because a record executive recognized genius.
The pattern is everywhere: blind repetition beats careful planning. Massive trial and error beats brilliant design. Systems that try everything and keep what works outperform systems designed by experts.
This essay is an exploration of this “Bitter Lesson” phenomenon that seems to invisibly influence everything around us- and more importantly, it’s about figuring out how we can leverage it instead of fighting a losing fight raging against it.
Let us start with the most obvious and glaring example where the bitter lesson dictates everything: life itself.
II.
Nobody sat down with a blueprint and said, “Let’s put the retina here, add some rods and cones, maybe throw in a blind spot for character.” The human eye evolved through millions of years of random mutations. Most failed.
The tiny fraction that helped animals survive got passed on. No meetings. No design reviews. Just endless experimentation with death as the divine “error correction” filter - something we’ve now termed as “Evolution”.
This process created structures so complex we still can’t fully understand them. We’ve built computers that can beat any human at chess, but we can’t definitively explain how the three pounds of meat in your skull generates the experience of reading these very words that you’re reading right now. Evolution, through pure trial and error, built something our greatest scientists can’t reverse-engineer.
The irony is very apparent if you think about it- we’re using evolutionarily-developed brains to build artificial ones, and we don’t fully understand how either works. Nobody knows exactly why scaling up language models makes them seem intelligent. We just noticed that if you make them bigger and feed them more text and train them for longer, they get better at everything. Does that sound familiar?
Language itself is a glowing example of the bitter lesson. No committee announced, “Let’s adopt the word ‘cool’, and let’s carefully consider whether the word ‘sybau’ should be immortalized for all time”
Millions of tiny conversations acted like mutations; the phrases that felt fun or useful survived and the rest are quietly dying.
Carefully invented tongues such as Esperanto never took off, while emoji, rap lyrics, and “LOL” spread like wildfire. One can predict that the most popular languages - English & Chinese - will be the language of most of the world in a thousand years. English being fairly decentralized might end up being the most popular one- but the English of tomorrow may not resemble the English of today, in the same way that the English of the past is hard to read today. Viral phrases and words - words which universally capture some emotion or moment perfectly - will continue to be added - while words which once signalled status and sophistication (including the word sophistication, probably) will die out.
The bitter lesson is perhaps bitter, because you can’t do anything about the above fact. You just have to accept it.
It’s only once you accept it that ideas start to emerge of how you can leverage it - like if you want to write a novel that lasts a centuries beyond your life - your best bet is to write in a language that’s the most popular right now, and use words which are considered too simple even by today’s standards.
III.
Thus, accepting the bitter lesson is better than fighting it.
The first time this was made apparent in the realm of tech was in chess.
For decades grandmasters poured wisdom into computers, adding rules like “put rooks on open files.”
Then DeepMind built AlphaZero, told it only the legal moves, and let it play millions of games against itself overnight. The result crushed the previous world-champion engine and discovered knight sacrifices no textbook approved.
Centuries of human lore lost to raw experience. Some believe that Chess might be fully “solved” in the same way as Tic Tac Toe is solved, but even if that’s not the case - the fact that we got machines to be superintelligent at Chess far before we even know “how” it learned to go there - shows another bitter truth about the bitter lesson - it doesn’t matter if you don’t understand the field where the bitter lesson is being deployed, all that matters is that you respect it and let it run its course.
This same bitter lesson has been internalised by TikTok & Facebook today - where every interaction is captured while trying to answer the question of “What video makes you stay on this app longer?” - and this process is repeated for billions of videos daily. Thus, by the time you’re opening one of these apps - the algorithm has already internalised exactly what videos you’re likely to watch (based on the data they have on cohorts of people who are extremely similar to you) and also videos you’re likely to skip or comment on, and they present all of them to you in the hopes that they’re able to retain you for hours. If you go against this - swipe past things you’re not supposed to be swiping past, or close before the algorithm’s predictions, that’s new data it uses to more efficiently try to hook you next time.
And none of that is nefarious - you’re being hooked by content you want to see (otherwise why would you be opening the app?) and it’s a sign of the machine minds working correctly when you feel yourself addicted to such mindless scrolling as much as everyone else around you.
Short form content thus becomes news and entertainment and everything in between - and it’s only those who ride this wave and internalise this bitter lesson will end up thriving, while those who don’t will get lost in obscurity.
Such a world where virality matters more than truth also means that what’s likely to be viral - sensationalistic, rage-inducing or divisive content spreads much faster than the boring and nuanced truth. This means everyone from influencers to politicians - everyone with better memes - catchy slogans and divisive attitudes will fare much better than those who aim at the signal and the truth over trusting the noise.
All this paints a grim picture - that perhaps we should just accept the inevitable conclusion that it’s all about repetition and scale. That it’s better to never fight the bitter lesson and accept its ways. That in a world that’s birthing super intelligence, maybe we shouldn’t question the very law that seems to define everything from evolution to social media to AI to culture and politics.
But that’s not the case.
IV.
Or at least, not the whole case.
In cases like evolution, we should all be thankful of the bitter lesson for letting us be the thinking monkey species that we are. Even in politics, the bitter lesson might mean that the most viral - not the most competent get elected into positions of power, but it also means that the masses can error-correct and throw them out of office in healthy democracies - thus leading to better outcomes and wiser societies as a whole.
But what about lone inventors and geniuses themselves?
Henry Ford famously said if he’d asked people what they wanted, they’d have said faster horses.
Steve Jobs didn’t focus-group the iPhone into existence.
Einstein didn’t crowdsource relativity.
Sometimes a single brilliant insight really does change everything.
But here’s the important part about even these examples- even these triumphs of individual genius end up proving the bitter lesson in a different way.
Ford’s Model T wasn’t actually the first car, or even the best car of its time. What made Ford revolutionary wasn’t the invention itself but his assembly line - a system that enabled massive scale and iteration. The Model T succeeded because Ford could produce millions of them, learning and improving with each one. His genius wasn’t in defying the bitter lesson but in being the first to apply it to manufacturing.
The iPhone? Revolutionary, yes - you could argue all phones would still have keypads if it wasn’t for Jobs. But look what happened next. Within years, Android phones - built by countless manufacturers trying endless variations - captured most of the global market. The iPhone remains premium and profitable, but the bitter lesson won the numbers game. Today’s smartphones, whether iOS or Android, are shaped more by billions of user interactions and A/B tests than by any single designer’s vision.
Even Einstein’s relativity, that pinnacle of individual genius, followed the bitter lesson’s pattern. His breakthrough built on countless failed attempts by other physicists. And more importantly, it only became accepted truth through endless experimental validation. One brilliant idea meant nothing until thousands of experiments confirmed it worked.
The real insight isn’t that genius doesn’t matter - it’s that genius and the bitter lesson aren’t opposites. They’re dance partners.
The geniuses who truly change the world aren’t the ones who fight the bitter lesson but the ones who see it first and harness it best. They’re not sculptors imposing their will on marble; they’re surfers who spot the wave before anyone else and ride it perfectly.
Consider how the most successful “genius” companies actually operate. Amazon’s Jeff Bezos didn’t try to predict what customers wanted - he built a platform that could run millions of experiments. Netflix didn’t guess what shows to make - they analyzed viewing data from millions of users. Even Apple, that supposed temple of design genius, tests countless prototypes and iterations for every product.
The bitter lesson doesn’t eliminate the need for human creativity and insight. It just changes what kind of creativity matters. Instead of trying to design perfect solutions, the new genius designs perfect systems for finding solutions. Instead of being the smartest person in the room, you build rooms where collective intelligence can emerge.
This is actually liberating. You don’t need to have all the answers. You just need to be smart about setting up the questions.
Want to write a novel that lasts centuries? Don’t try to predict what future readers will love. Write many things, see what resonates, and double down on what works. Want to build a successful product? Don’t trust your intuition about what customers want. Build something simple, release it fast, and let user behavior guide your iterations.
The bitter lesson is only truly bitter if you’re attached to the idea of being the sole author of success. If you can let go of that ego, it becomes a superpower.
You get to leverage forces far greater than your own intelligence - the collective creativity of crowds, the relentless optimization of evolution, the surprising discoveries that emerge from large-scale experimentation.
We’re entering an age where AI can try a billion possibilities while we sleep, where global platforms can test ideas on millions of users in minutes, where the feedback loops that once took generations now happen in seconds. Fighting this is like trying to hold back the tide.
But surfing it? That’s where the magic is waiting to happen.
The bitter lesson isn’t that human ingenuity doesn’t matter. It’s that human ingenuity needs to evolve. The old model of genius - the lone inventor in the garage - is being replaced by something more powerful: the genius who knows how to orchestrate systems that discover solutions we never could have imagined. Where AI isn’t Artificial Intelligence but rather Amplified Intelligence, and whoever can amplify themselves the greatest- wins.
So yes, the bitter lesson is everywhere. Yes, it means accepting that brute force often beats brilliance.
But it also means we have access to a force multiplier unlike anything in human history. We can be gardeners in a garden that tends itself, conductors of an orchestra that writes its own music, surfers on waves that carry us further than we could ever swim.
The lesson is only bitter if you wanted to be God. If you’re content to be human - but a human who understands the rules of the game - then it’s the sweetest lesson of all.
Because once you truly understand that the bitter lesson is everywhere, you realize something profound: you don’t have to fight the universe to change it.
You just have to know which way it wants to flow, and gently guide it toward better outcomes.
It doesn’t always have to be about trusting the crowds or refusing to be contrarian, but truly leveraging the bitter lesson is understanding how everything works in the first place - and then starting from there.
You can be original, you can be creative, you can change the world against all odds - and you can do all of this without being bitter about the bitter lesson.



