Matrix Locks One Person in One Room

Originally written in Traditional Chinese: Matrix 是把一個人關在一個房間：AI 讓你多一個角色，前提是你還在思考

"Matrix organizations lock one person in one room." That was my argument in a debate in Tokyo.

On day one of my Minerva Immersion Week, a workshop asked which organization works better. Matrix: everyone has a job description and does their own job. Foreigners in Japanese offices cannot build partitions, the speaker joked, so they wear big headphones. Amoeba: first developed at Kyocera, shared responsibility, and you join the company, not a department. My table was assigned Amoeba.

My strategy: if we only argued from the slides, we would lose. We needed a new factor on the table: AI. A matrix limits one person to one room. With AI, one person can do far more, and the information sharing that used to make Amoeba exhausting can become simple. Systems that worked in the past need to be rethought.

In our report we said Amoeba has two old problems, unclear objectives and information overload, and AI gives us a better chance to tackle both. I mentioned a new Wharton paper: Thinking, Fast and Slow gave us System 1 and System 2, and in the AI era there may be a System 3. Our conclusion: with AI, Matrix and Amoeba can be combined.

I went back and reread the paper. Steven Shaw and Gideon Nave at Wharton call AI "System 3," cognition that runs outside the brain. Across three experiments, 1,372 people and 9,593 trials, participants chose to consult AI on more than half of the questions. When the AI was right, accuracy rose 25 percentage points over baseline. When it was wrong, accuracy fell 15 points. Confidence went up after using AI, even when the AI was wrong. They call it cognitive surrender.

So, to be clear: AI does let one person play more than one role, but only if that person is still thinking. Someone who hands everything to System 3 does not gain roles. They lose the one they had. The discipline of the matrix, knowing what you are accountable for, is not outdated. It has moved from your job scope to your judgment.

Try this: before you ask AI anything this week, spend two minutes writing your own answer. Then ask: "Here is my answer. Don't give me yours yet. Tell me where I might be wrong."

This is what Minerva trains: not who memorized more, but who can put a new factor on the table. The MDA gives me these debates, and World Wise lets me test them with people who lead real teams. Application links for both programs are in the first comment.

References

Take a seat at this table

Follow in English

I also post in English on LinkedIn.

👉 Follow Haishuo Lee on LinkedIn