The Last to Be Automated
People keep pointing to OpenAI's $500,000 software engineer salaries as evidence that claims about automating programming are overblown. How can they talk about AI replacing programmers while paying such astronomical sums to hire them?
But they have it exactly backwards. OpenAI is a company that's massively bottlenecked on talent. They need the world's top talent and they're willing to pay whatever it takes to get it.
This is the top few percentiles of human ability and productivity we're talking about. And crucially, these are the last percentiles that will be automated. What's more, on the path to being automated, their value will actually skyrocket. Why? Because having them on your team might be the difference between getting to AGI first or not.
This is exactly what you'd expect to see in a world where software engineering is on track to being automated. The clearest signal wouldn't be top companies stopping their pursuit of elite talent. It would be a sharp drop in hiring for more basic roles: the kind needed by companies developing standard websites or apps that don't push the boundaries of what's possible.
We're seeing a bifurcation in the market for programming talent. At one end, there's a fierce bidding war for engineers who can work on the bleeding edge of AI. At the other, we're beginning to see the first effects of automation on routine development work.
In short: in any field being automated, the people who are last to be replaced are those working on the automation itself. Their value tends to increase as automation progresses, because their work accelerates the whole process. The better the automation tools become, the more leverage these top performers have, and the more valuable they become to companies racing toward the finish line.
The real gauge for automation isn't what's happening with elite talent at companies like OpenAI. It's what's happening to hiring for conventional programming roles. That's where we should be looking for the first signs of change.