After Move 37
Why the rise of AI won’t replace us — it might just reveal our deepest strengths
Go, developed over 2,500 years ago in China, is the oldest continuously played board game in the world. Two players compete to fence off more territory than the other—using just black and white stones.
Go is particularly unique because despite its simple rules, it’s quite complex. The number of possible game positions exceeds the number of observable atoms in the universe.
To put it in perspective: Go has 10^126 times more possible positions than chess—a number so vast it's almost meaningless. That’s why Go was long seen as too complex for AI. It was the ultimate stronghold of human intuition.
Move 37: The machine has arrived
This all changed in March 2016, when long-time Go world champion, Lee Sedol lost 4 -1 to DeepMind’s AlphaGo in Seoul.
The rise of the machine was pointedly apparent in the second round of the tournament. In that round, AlphaGo performed a move that observers agreed no human ever would - with Sedol himself describing it as touching the divine.
Move 37 as it is infamously known “perfectly demonstrated the enormously powerful and rather mysterious talents of modern artificial intelligence”. God-like intelligence was here, or at least on the way.
Today it seems that every facet of life, especially our work, is around the corner from a “Move 37” moment. If the doomsayers are correct, most of us will be out of a job in no time at all, with the most acutely affected in the short-term being those in entry and low-level jobs.
Why would you hire a fresh graduate or an intern when ChatGPT’s Deep Research can do the job without the need for a desk, let alone work-life balance?
The real impact of AI on work
There is little doubt that AI, even more so than previous technological waves, will change how we work. Nevertheless, the current rhetoric around this AI wave (LLMs) is off the mark. Our work as humans will change but will not be eliminated. There are three key factors missing from the debate.
First, AI will likely create more knowledge work—not less. By expanding what’s possible, it raises expectations and generates new kinds of tasks and roles.
It’s tempting to believe that “this time will be different” but looking at past productivity revolutions says otherwise. Consider the launch of the personal computer: it was expected to destroy jobs in law and finance. Instead, it unlocked entirely new opportunities in both.
Second, AI doesn’t learn like us. In other words, LLMs don’t improve over time the way a human would. It is tempting amid all the hype about LLMs to overvalue raw “intelligence”, but what is irreplaceable today about humans is “their ability to build up context, interrogate their own failures, and pick up small improvements and efficiencies as they practice a task”.
Third, despite all the current hype on AI “reasoning” it appears that AI reasons very little. It turns out you can do quite a bit with rote learning - especially if you’ve memorized every bit of text on the internet. In fact, you can do more than most humans on tasks within your “training distribution”, but you eventually hit a “wall”.
As education reformers have known for a long time, rote and rooted are two different things. You can’t memorize your way to deep understanding. Current LLMs are poor at dealing with novel use cases as complexity rises. Again, this reinforces the need for deeper human expertise and unique taste.
The real work ahead
Of course, this is not a forgone conclusion, and is not an argument that the transition will not cause dislocation and distruption.
For us to be able to ask workers (new joiners and old) to do more we need to change how we educate them. We’ve known that current education systems are broken but done little about it. LLMs are just finally forcing us to face that reality.
This is no easy task and will entail finally making the transition away from rote memorization to a focus on deep understanding and true expertise using better and more project-based assessment. Thankfully, this is a move that the current AI wave itself can help enable.
In parallel, we need to do this while being vigilant about the permeation of AI in all facets of our lives and the negative impact it can have on us. Although the research is still early, it is clear that cognitive offloading can be quite harmful.
Move 78: God-like intelligence in all of us
Even though Sedol lost the tournament with AlphaGo, it marked a turning point for Go players around the world. Sedol’s loss counterintutively catalysed a leap in human gameplay.
Since 2016, the average “move quality” in Go has continued to improve after largely plateauing from the 1950 to the mid-2010s. In fact, there were already signs of this in Game Four of the tournament, the only game that Sedol won.
During that game, Sedol made a move (move 78) that completely baffled AlphaGo and expert observers. The AI put the odds of a human playing that move at 1 in 10,000.
Move 78 won Sedol the game and proved that as humans we have not lost the ability to generate our own transcendent moments. When Sedol was asked after the match about move 78, he simply answered that it appeared as the “only logical move”.
It turns out, there is a little bit of God-like intelligence in all of us. We might just need AI’s help to unlock it.
I hope you enjoyed the last edition of Nafez’s Notes.
I’m constantly refining my personal thesis on innovation in learning and education. Please do reach out if you have any thoughts on learning - especially as it relates to my favorite problems.
If you are building a startup in the learning space and taking a pedagogy-first approach - I’d love to hear from you. I’m especially keen to talk to people building in the assessment space.
Finally, if you are new here you might also enjoy some of my most popular pieces:
The Gameboy instead of the Metaverse of Education - An attempt to emphasize the importance of modifying the learning process itself as opposed to the technology we are using.
Using First Principles to Push Past the Hype in Edtech - A call to ground all attempts at innovating in edtech in first principles and move beyond the hype
We knew it was broken. Now we might just have to fix it - An optimistic view on how generative AI will transform education by creating “lower floors and higher ceilings”.