Optimism, abundance, and the ambition to build a new world

Every automation I have ever built had the same shape. Find the ritual nobody wrote down. Write it down. Hand it to a machine. For twenty years the limit was the second step: you could only automate what you could describe precisely, and most of what a business does is not precise. It is judgment, applied a thousand times a day, by people who cannot tell you how they do it.

That limit is gone. A model can be shown ten examples of a judgment and make the eleventh. The ritual no longer has to be written down. It has to be demonstrated. I have watched this happen on my own desk: automations that used to take a weekend take an afternoon, and automations I would never have attempted, because the judgment inside them was too fuzzy to encode, now take a conversation.

The case for optimism

Here is the fact I keep coming back to. The cost of a unit of intelligence has fallen by orders of magnitude in three years and it is still falling. More than a billion people use a frontier model every week. The price of a piece of reasoning that would have cost a consultant an afternoon is now a fraction of a cent, and it is available to a kid in Orlando and a founder in Lagos on the same terms as to a bank in New York.

Every time a fundamental input gets cheap, the world gets bigger. Cheap steel gave us cities. Cheap energy gave us the twentieth century. Cheap computing gave us everyone a phone. Cheap intelligence is the biggest of those because it is the one that makes the others cheaper too: it designs the reactor, finds the molecule, writes the software, and runs the company. We are standing at the beginning of that curve, not the end, and I think that is the most optimistic place anyone has stood in a generation.

Abundance is a choice

Abundance does not arrive on its own. Cheap intelligence makes it possible; it does not make it happen. Somebody still has to decide to build the clinic that diagnoses in minutes, the school that tutors every child one-on-one, the power plant that gets approved in a year instead of ten, the company that does with three people what used to take thirty. The technology lowers the price of ambition. It does not supply the ambition.

And that is the thing I worry about more than any of the usual worries. Not that the machines will do too much, but that we will ask them to do too little. We will use the biggest lever in history to write slightly better emails. The failure mode of this decade is not catastrophe. It is timidity.

What it looks like to be ambitious

Three things, and they are not the three most people talk about.

• Aim at the problems that were too expensive to try. Drug discovery for diseases that affect ten thousand people. Legal help for people who could never afford a lawyer. Personal medicine. Personal tutors. The things that were obviously good and obviously unaffordable are now just obviously good.
• Start more companies, with smaller teams. A three-person company can run a support desk, a sales motion, a content operation, and a finance function that would have taken thirty people. I have lived this. It is not hypothetical. The number of problems a small group can take on just went up tenfold, so take on tenfold more.
• Build for a bigger world, not a defended one. The instinct when something powerful arrives is to protect what you have. The better instinct is to ask what you would build if the cost of trying were nearly zero, because it is.

The honest caveats

A model is confident when it is wrong. An automation built on it inherits that confidence. The discipline I follow is the same one I would want from a new hire: narrow scope, checked work, a human on anything irreversible, and logs for everything. The tools are getting better at policing themselves, but the fence is still mine to draw. Optimism is not the absence of care. It is care with a direction.

And the bottleneck moves. When building is cheap, deciding what to build is the whole job. Taste, judgment, and the nerve to pick a problem that matters become the scarce thing. That is a job I like, and it is the job I am inviting you to.