I write about the AI transition: what it costs, who pays for it, and what physics allows once it is done.
Most of what gets written about AI is either a warning or a sales pitch. I would rather work the numbers out and show them, so you can check them. Often the limit everyone talks about is not the one that actually holds things up. It tends to be something physical: power, transformers, heat.
Why I am writing it
- To see what a good outcome looks like. What is possible is set by physics, not by opinion, so it is worth working out where the limits are.
- To help it go at a pace people can prepare for. The outcome could be fine and the way there still a mess. The difference is mostly speed, and speed depends on things you can check: power plants, money, and whether people agree to it.
How the articles work
Each article asks one question. Next to the text is a chart that changes as you scroll, to match what the text is talking about. You can turn the chart off with the switch at the top of the article.
The chart has buttons for different assumptions. One of them shows the setting where my argument stops working. If there is a flaw, I would rather you find it than take my word for it.
Under every article is a list of the numbers it uses, where they come from, and what I found when I checked them. Some of them say I had it wrong.
Ground rules
Assumptions are labelled as assumptions. These are scenarios, not predictions, and there are no dates in them, because delays are political.
None of this is financial advice.
Me
Stijn Verlinden. I am not a physicist, and some of these articles are me finding out that something I said confidently for years was wrong. I write by sending voice notes to an AI and arguing with it.
If you think a number is wrong, send me the number and where you got yours.