# What Go and AI tells us about revolution
Date: 2025-05-19

---

---

### **What Go and AI tells us about revolution**

The advent of AI in board games demonstrated that human players exhibited [Simon’s bounded rationality](https://en.wikipedia.org/wiki/Bounded_rationality):

> In *Models of Man*, Simon argues that most people are only partly rational, and are irrational in the remaining part of their actions. In another work, he states “boundedly rational agents experience limits in formulating and solving complex problems and in processing (receiving, storing, retrieving, transmitting) information”.

In [Lee Sedol’s famous Go game against AI](https://www.nytimes.com/2024/07/10/world/asia/lee-saedol-go-ai.html?unlocked_article_code=1.6U0.ZaaY.hihR1DAlgQY4&smid=url-share), the AI differed from the standard strategies that humans were playing. Human Go players had been taught and never challenged many moves that were wrong.

**What’s more ditching standard strategies to try to play at the AI level only made the humans players worse.** Humans lacked the compute necessary to rethink every move while playing. Lee Sedol and others played better against the AI when they continued to more quickly play moves they knew were wrong IE bounded rationality was the best they could do.

Bounded rationality means we all have intellectual assumptions that we do not have the time and compute to introspect. Without these assumptions are decision making would be even worse.

If one of these bounded rationality assumptions is wrong it will take us a long time and a lot of compute to correct it — unless we are given an AI that shows us better decisions.

Chess is roughly 1500 years old and underwent a transformation after AI also. Go is 2500 to 4000 years old and yet after AI [started beating humans](https://www.amazon.com/Fuseki-Revolution-How-Has-Changed/dp/4906574890):

> Opening patterns previously favored by professionals of all levels have lost popularity and some have disappeared altogether. Large moyos have lost out to the thoroughgoing preference of AI for actual territory and its skill at reducing moyos. Josekis have been transformed, with ‘standard’ moves disappearing and their place being taken by new techniques invented by AI. Even some moves that were previously considered taboo, as being crude or ineffective, have been reassessed by AI and have earned places in the standard repertory.

Three thousand or so years is a very, very long time for great human thinkers to be carrying around misconceptions on the basics of Go.

We live in a time where current economic practices have led to outcomes that are unacceptable — ranging from environment destruction to enormous inequality and the danger of fascism. Our “standard” economics is clearly very, very wrong.

Furthermore humans competing with each other for better solutions may not be the way out in any reasonable time frame. [Bounded rationality](https://thedecisionlab.com/biases/bounded-rationality) means less innovative approaches may outperform experimental economic decision making unless the experimentation gets a lot of long run effort.

AI training isn’t competitive because the AI freely experiments without worrying about trying to maintain a win ratio.

![Image](https://disrael.github.io/blogs/assets/what-go-and-ai-tells-us-about-revolution-ea36c43f48.png)

What Go and AI tell us about revolution is that our only way forward is to be less conservative and less competitive— even if it doesn’t immediately work out. At the same time the revolution can’t be random as that won’t teach us anything.