What Happens When AI Can Do Our Jobs?

I've been thinking about this a lot lately.

Not in the usual "AI is going to change the world" way. We already know that.

I'm thinking about something much more personal.

What happens to us when AI can actually do our jobs?

I'm a programmer, so naturally I started thinking about programming first.

A few years ago, if you wanted to build a website or an application, you needed developers. Someone had to sit down, write the components, connect the APIs, design the database, debug everything, deploy it, and maintain it.

Now I can describe what I want and AI can generate a surprisingly large portion of it.

And it's only 2026.

That's the part that scares me.

I don't think coding disappears. I think writing code becomes less important.

There's a difference.

Programming used to mean telling computers exactly what to do.

We went from machine code to assembly.

Then C.

Then higher-level languages.

Then frameworks.

Each time, we moved further away from the actual machine.

AI feels like another abstraction layer, except this one is enormous.

Maybe twenty years from now, manually writing thousands of lines of application code will feel a little like writing assembly for a website today.

You can do it.

Some people will still need to do it.

But most people won't.

Instead of writing:

const map = new Map()

and carefully constructing the implementation ourselves, we might describe the properties of the system we want.

Customers should never be charged twice.
European customer data must stay in Europe.
Optimize for availability.
Keep infrastructure below $20,000 per month.

And AI figures out the architecture, writes the implementation, tests it, deploys it and monitors it.

The programmer hasn't necessarily disappeared.

But programming has changed.

And this isn't just about programmers.

My first instinct was actually:

Fine. If programming gets automated, maybe sales and marketing become more valuable.

But then I thought about it.

AI can write cold emails.

It can research prospects.

It can generate advertisements.

It can create videos and images.

It can analyze sales calls.

It can write SEO content.

It can answer customer questions.

It can translate between languages.

It can analyze spreadsheets.

So moving from programming to marketing doesn't really solve the underlying problem.

If AI becomes sufficiently capable, almost every job that happens entirely through a laptop is exposed in some way.

Accounting.

Design.

Marketing.

Programming.

Writing.

Analytics.

Recruiting.

Consulting.

Even parts of management.

The question isn't really:

Which white-collar job will AI not touch?

There might eventually be very few.

Physical jobs are different — for now.

An AI can already generate code.

But ask it to walk into a random 40-year-old house, figure out why one room keeps losing electricity, open the wall, discover that somebody wired something incorrectly twenty years ago, repair it, explain the problem to the homeowner and guarantee that the house isn't going to catch fire.

That's much harder.

The physical world is messy.

Every house is different.

Every patient is different.

Every construction site is different.

This is why electricians, plumbers, nurses, construction workers, mechanics and other physical professions may actually be harder to automate than many prestigious white-collar jobs.

It's kind of funny.

For decades we assumed robots would replace manual labour first and knowledge workers would be safe.

Instead, AI learned to write poetry and code before robots learned how to reliably fix your plumbing.

But robotics will improve too.

Which leads to the uncomfortable question.

What happens if machines eventually become better than us at almost everything?

This is where things get weird.

Our society basically works like this:

You work → someone pays you → you use that money to live.

That system works because human labour is scarce and valuable.

But imagine intelligence becomes almost free.

Then imagine robotics becomes cheap too.

A company that once needed 100 employees might need 10 people and thousands of AI agents.

Eventually perhaps five people.

Maybe one.

At the same time, starting companies becomes ridiculously cheap.

A single person could potentially operate something that previously required an entire corporation.

So we might get both things simultaneously:

far fewer people required per company

and

far more companies being created.

Nobody really knows where those two forces balance out.

I think the next few years will look something like this.

Right now we're entering the AI coworker era.

A programmer doesn't disappear.

A programmer with AI simply becomes much faster.

Then we'll probably move toward agents.

Instead of asking AI:

Write this function.

We'll say:

Build this feature.

Then:

Fix this bug.

Then:

Maintain this service.

Eventually:

Here's the business objective. Figure it out.

The human moves further and further away from execution.

So maybe the progression looks something like:

2020

Human does the work.

↓

2026

Human + AI assistant.

↓

2030?

Human + multiple AI agents.

↓

2035?

Human supervises autonomous systems.

↓

2040?

Very small groups of humans control enormous productive capacity.

↓

???

And I think that last question mark is important.

Nobody actually knows.

Anyone confidently explaining exactly what employment will look like in 2045 is guessing.

Including me.

Maybe "having a job" eventually stops being the center of life.

This is the part I find most interesting.

Suppose AI and robotics eventually become capable enough to produce most of what society needs.

Food becomes cheaper.

Transportation becomes cheaper.

Software becomes almost free.

Manufacturing becomes heavily automated.

Energy becomes cheaper.

Education becomes available to everyone through incredibly capable AI tutors.

At some point, tying everyone's survival completely to employment starts becoming strange.

You can't have billions of people with no economic value and simply tell them:

"Well, good luck."

Governments would have to respond somehow.

Maybe UBI.

Maybe shorter working weeks.

Maybe AI taxes.

Maybe public ownership of certain AI systems.

Maybe some kind of national AI dividend.

Maybe something nobody has invented yet.

And strangely, that future doesn't necessarily sound terrible.

For almost all of human history, most people have spent a huge portion of their lives doing things they had to do in order to survive.

If machines eventually handle much of that work, perhaps the question changes.

Instead of:

What job should I get?

It becomes:

What do I actually want to spend my life doing?

People could build things.

Study mathematics.

Paint.

Travel.

Research obscure subjects.

Make games.

Start businesses.

Write.

Compete in sports.

Take care of their families.

Or spend six hours trying to understand some completely useless thing simply because they find it interesting.

Maybe work doesn't disappear.

Maybe mandatory work becomes less important.

Of course, there's a darker version of this future too.

If a tiny number of companies own all the AI, robots, infrastructure and capital, inequality could become absurd.

The technology itself doesn't decide which future we get.

Society does.

So what am I supposed to do right now?

This is the question I eventually came back to.

I'm 23.

I'm learning DSA.

I'm learning system design.

I'm learning AWS.

I'm trying to become a better programmer.

And occasionally there's this thought in the back of my head:

What if I'm becoming good at something that won't matter anymore?

I don't think the answer is to stop.

Understanding computers deeply still matters.

But I don't want my entire economic value to be:

I can type code.

That's becoming a dangerous place to stay.

I'd rather move through something like this:

person who writes code

↓

engineer who understands systems

↓

person who uses AI to build systems

↓

person who understands which problems are worth solving

↓

person who can build, communicate and sell solutions

↓

person who owns things

That last one might matter the most.

If AI makes labour cheaper and cheaper, then spending your entire life selling your labour might become increasingly risky.

Ownership becomes interesting.

Own a company.

Own software.

Own equity.

Own intellectual property.

Own distribution.

Own something whose value increases when technology becomes more productive.

I don't want to compete against AI by trying to type faster than it.

That's obviously a battle I'm going to lose.

I want to be the person deciding what we should build with it.

And eventually, hopefully, owning some of what gets built.

I'm still scared.

I don't have a motivational conclusion here.

AI could create incredible prosperity.

It could destroy huge categories of employment.

It could create completely new industries we can't imagine yet.

It could dramatically increase inequality.

Most likely, it will do some combination of all of those things.

But I think I've changed the question I'm asking myself.

I used to think:

"How do I become good enough that AI can't replace me?"

I'm starting to think that's the wrong goal.

There may eventually be very few things AI can't do.

A better question might be:

"How do I put myself in a position where increasingly powerful AI makes me more capable instead of making me irrelevant?"

I don't completely know the answer yet.

But I think learning deeply, building things, understanding business, learning how to communicate and sell, and eventually owning things is a better bet than trying to predict the one profession AI will never touch.

We're probably going to live through one of the strangest economic transitions in human history.

Might as well learn how it works while we're here.