I named this blog Living Unscripted because I've never been one to follow the default path. For most of my life I thought that was just a personality thing. I now think it's becoming one of the most valuable traits in the economy.
Here's the argument.
As intelligence becomes abundant, agency becomes scarce.
Everyone is worried about superintelligence and that there will be no work left for the rest of us. The assumption hiding under that worry is that intelligence was the scarce resource all along.
I don't think it was.
The world has never suffered from a lack of intelligence. It suffers from a lack of execution.
My theory is that having superintelligence is like having a building full of brilliant people waiting for direction. They can generate options, analyze tradeoffs, and execute fast. But there have always been more brilliant people than missions worth pointing them at, and the scarce thing was someone with the agency to aim them. That doesn't change when the intelligence comes from a model instead of a team.
Think about any codebase. There's always a pile of work everyone agrees should be done — tests, lints, security audits, documentation. None of it is a mystery. Nobody disputes the value. It just never gets done, because the cost of doing it is higher than the return, so it sits on the list for years. Cities have the same list with infrastructure. Companies have it with projects. We know the work is worth doing. We just never do it.
AI changes that math. The most important thing about it isn't that it can answer questions. It's that it collapses the cost of actually doing things. Writing comprehensive tests for a mature codebase used to take weeks of someone's attention. Now most of it can be generated, reviewed, and shipped in hours. The value didn't change. The cost did. And when the cost drops, a massive backlog of work that was never worth doing suddenly is.
This is already happening. If we had today's models in 2022, we would have concluded that software engineers were finished. Instead, there is more software being built than at any point in history.
The doomsday math assumed the amount of worthwhile work was fixed. It was never fixed. It was rationed by the cost of turning intelligence into execution.
The constraint didn't disappear. It moved. That's the agency economy: a world where intelligence is abundant, execution keeps getting cheaper, and the scarce resource is judgment about where to aim both.
That judgment has two parts.
Agency is the willingness to push an idea into reality. To start before you're ready. To act before consensus forms. To take responsibility for the outcome.
Taste is the other one. The word gets thrown around a lot right now, and it usually means aesthetics — good design sense, knowing what looks right. That's not what I mean. I mean the ability to recognize what matters before consensus forms around it.
This is where models are least useful without human direction. A model learns from consensus — the accumulated record of what we've already written down, labeled, and agreed was worth saying. That makes it extraordinary at every problem the world has already named, and weakest on the ones it hasn't. It can't tell you what the world hasn't realized matters yet.
That part is on you.
Taste without agency is preference. Agency without taste is motion without direction. The future belongs to people who have both.
In the agency economy, the value isn't in what you know, or even in what you can do. It's in what you decide is worth doing, before anyone agrees with you, and whether you actually go do it.
The world is full of things that should exist but don't. Not because nobody thought of them, but because between the idea and the thing stood a fundraise, a hiring plan, and months of convincing other people to care. Most ideas died in that gap. That gap is collapsing, and what's left is the bottleneck that was always underneath it: deciding what matters and taking responsibility for making it real.
Which brings me back to the name of this blog. The scripted path — school, safe career, work your way up — is made of work everyone already agrees is valuable. And the clearer and more agreed-upon the work, the faster AI improves at it. The safest-looking route just became the most automatable one. The further you step off the script, the more your taste and agency are worth.
Which means this may be the best time in history to live unscripted.