A day of intense input

I set out to see the company’s new office, a long trip away. Halfway there, the person in charge said I need not come. I turned back, had lunch, and settled into a café with cup after cup of coffee. In the afternoon I finished an hour of recorded material and reviewed eight lessons.

One lesson compared a computer with the mind: working memory to RAM, longer-term memory to a hard drive. It described the mind as a kind of Bayesian inference machine using remembered concepts and their connections. The teacher argued for memorizing core ideas directly so there is more to reason from. I take that as a model for why a richer store of knowledge can support better judgments, not as a complete description of the brain.

I understood the analogy more clearly while researching leading AI companies. Wanting to write well about them pushed me to study how AI works and learn what hard drives, memory, and GPUs do. Teaching through writing made the lesson real. Li Xiaolai’s idea of learning by teaching has stayed with me.

Technical terms intimidate me less now. I installed OpenClaw myself after seeing others do it, and I research companies by looking things up rather than waiting for someone to explain them. Having used paid AI tools for years and tried to make real things with them, I can judge some claims more directly.

I want to keep taking in good material so I can understand faster and make more. I am replacing some entertainment apps with AI tools, building a note system, and connecting ideas across its parts. Restating what I learn in different contexts has become its own pleasure.

• Learning can be a form of making.
• Making can be a form of learning.
• A popular view is not necessarily the result of independent thought.