AI Is Not Coming for Your Job

AI is not going to overtake our jobs. Or at least it's not going to happen soon.

You can get decent deliverables with AI, but they're far from finished work. Getting what you actually envision takes a lot of trial and error. And if you want higher-quality outputs from better models, you need to pay a lot.

For example, I tried using Claude to build a swim lane diagram using Mermaid syntax. You can get a decent first draft, but making small adjustments after that is painful.

An investment banker can get a tidy and organized deck. But if you care at all about how it looks — and anyone who's worked in corporate finance knows how particular people get about this — you're going to spend a lot of time going back and forth with AI to get it exactly right.

That dynamic hits hardest early in your career. If you're a junior analyst in the financial world, it'll be harder, because AI amplifies what you bring to it. If you don't share the right context — business rules, conditions specific to your situation, decisions that aren't public knowledge — the output is going to be partial and not necessarily correct.

You are its manager. If you're a bad manager, you're going to get bad work from AI regardless.

For young analysts, the learning process will work the same way it's always worked: trial and error. Like everyone else, they'll have to build on what others figured out before them. That's how we've always learned. Apprenticeships work because they're structured trial and error.

AI just makes the feedback loop faster, but the fundamentals are the same. Some knowledge gets lost along the way, and some gets passed on. It's the same pattern with weapon-building, craftsmanship, or any trade. Just much faster.

So what will actually happen?

This AI boom is going to look a lot like the internet revolution. And career changes are also going to look a lot like the internet era.

When the internet arrived, did companies offer generous compensation packages to employees who weren't tech-savvy? Not necessarily. They mostly used the same methods they always had: restructuring, attrition, sometimes just being blunt about it. That's what I see happening here. The methods won't change.

The difference between companies will come down to who's willing to take the risk of changing fast. Some will move quickly, a lot will die in the process, and many will keep doing things the way they did them decades (even centuries) ago.

It's also not going to be as extreme as people fear. Robotics was supposed to replace most of the workforce, and it didn't. Because it's too expensive and there are things machines still can't do. There's a similar ceiling here. We're nowhere near AI doing everything, and I genuinely don't think we'll be living on this planet when that happens.

The landscape will shift, but the changes will be minimal. Yes, there'll be more knowledge work and more contract workers. But I don't see it being a 180 degree transformation. Even after the explosion of IT jobs, they're still not even close to half the workforce.

I do believe AI tools will make it easier to spin up something on your own. We're already seeing this happening. But the barrier is not the tools. It's the worldviews and beliefs of people:

1. Many are willing to trade financial freedom and creative freedom for the perceived security of having a 9-5 job.
2. Most people are risk-averse.
3. Not everyone is motivated to do something on the side. A lot of people just want the peace of mind of leaving all problems behind at five o'clock.

As long as that doesn't change, the work environment won't change much either. It'll come and go in waves. Like after COVID, when a lot of people went freelance for the flexibility, and then some burned out on finding clients and went back to employment. That cycle will keep happening.

People will keep rioting to secure jobs and income. Most riots in recent history come from economic reasons, like the price of bread, transportation, or housing. And as living standards rise, so do expectations. The poverty line gets higher, the minimum people will accept goes up. So they'll keep making demands.

The people confidently predicting either total transformation or everything staying the same are probably both wrong. History suggests it'll be messier and more gradual.

PS: This article was written as a response to questions raised by Paul Millerd on Twitter about AI, jobs, and the economy. If you haven't read his post, it's worth it.