AI Is Going to Create More Jobs, Not Fewer

AI is going to create more jobs, not fewer. I think it is going to bring us into a golden age of work and create a better life than most of us can imagine.

That does not mean every current job survives. It means the category of work keeps expanding.

Changing a job is not ending work

Every technological revolution makes some jobs less useful, removes parts of other jobs, and changes what people are paid to do. For the person caught in that change, the distinction can feel academic. A lost role is real. A skill that suddenly earns less is real. Companies can handle those transitions badly.

But none of that proves human beings run out of useful things to do. It proves that a particular arrangement of work was temporary.

The mistake in most AI job panic is treating today's job list as the final inventory of human possibility. Count the tasks a model can perform, subtract them from the current economy, and declare that the remainder is humanity's future. That is an accounting exercise pretending to be imagination.

Software is getting rearranged first

A lot of people expected AI to hit trucking, factory work, and manual labor first. What I see changing fastest is software. AI was built by people who code, and code is one of the things it tends to be extremely good at. The people who understand the technology best are also watching their own work change at an insane pace.

That does not look like the death of software engineering to me. It looks like a change in how software gets made.

Engineers who learn to work agentically can direct much more activity than they could before. The culture around vibe coding can be ridiculous, manic, and overconfident, but it is also productive energy. People are staying up all night building apps and trying businesses. Some of those businesses will be bad. Some of the apps will never find a customer. That is normal. The important part is that more people can make a serious attempt.

The barrier between an idea and a first version has dropped. Work that used to require learning several systems, coordinating specialists, or raising enough money for a team can now be attempted by one determined person using agents. The tools do not guarantee judgment, demand, or taste. They make action cheaper.

That is how new work begins.

Lower costs create a larger field

When people talk about AI replacing tasks, they usually stop the analysis at substitution. A system performs something a person used to perform, so one human task disappears. Fine. That happens.

But AI also makes capabilities affordable. It lets a small team attempt work that previously belonged to a larger one. It lets a person test an idea that would have remained a note in a phone. It lets companies offer things they could not economically offer before.

Once the cost of trying falls, the number of worthwhile attempts rises. More attempts mean more products, more businesses, more services, more failures, and more unexpected successes. We cannot describe all of the jobs that follow because the capabilities are arriving before the vocabulary.

That is not a reason to ignore transition. Businesses should train people to direct agents, inspect work, make decisions, and combine tools into real workflows. They should measure what new value a team can create, not only how many existing tasks can be removed. Leaders who use AI only as a head-count knife are choosing the smallest possible interpretation of the technology.

Humans do not run out of wants

The deeper reason I am optimistic is that people always want more. We want better homes, businesses, tools, entertainment, experiences, care, education, and lives for our families. Satisfy one want and another one appears. Remove one constraint and somebody notices a possibility on the other side of it.

Give people more capability and they do not become furniture. They start reaching.

I also reject the idea that human beings are not meant to work. I do not mean everybody needs a miserable job or that a paycheck is the only source of meaning. I mean doing, giving, providing, building, and becoming useful are part of how we understand ourselves. Family matters. Hobbies matter. But contribution matters too. What is a family if you cannot be the one who does, gives, and provides? They are just roommates.

Work is bigger than a current title. A title can vanish while the need to contribute remains.

Let the tools take the grind

I want AI to get better at the boring, shitty, hard work—the repetitive, annoying bullshit that drains human attention. That is not a tragedy. That is the point.

If a tool can carry more of the grind, people have more time for judgment, creativity, relationships, courage, taste, and care. Those are not decorative leftovers. They are where a huge amount of value begins.

For an operator, the implication is straightforward: do not ask only, "Which person can this replace?" Ask, "What can this team attempt now that it could not attempt before?" Then redesign the work around that answer. Keep people close to decisions that require accountability. Teach them to direct, review, and improve agentic systems. Move the repetitive load off their backs and expect more creative responsibility in return.

That is a much bigger opportunity than using AI to make the same company slightly cheaper.

The doom story has no imagination

I keep coming back to a point I heard Marc Andreessen make: the idea that AI takes today's jobs and leaves us with nothing shows a severe lack of imagination. That is my paraphrase, but the point lands.

The doom story assumes we invent no new products, businesses, needs, experiences, or forms of contribution. It assumes human ambition freezes at the exact moment our tools become dramatically more capable. It imagines machines changing and human beings staying perfectly still.

I do not believe that is how this goes.

Some roles will shrink. Some will disappear. People will have to learn new ways of working, and companies will have to decide whether they use AI only to cut or to create. The transition will not be painless or evenly distributed.

But I believe the direction is expansion. More people will be able to build. More ideas will become affordable to test. More boring work will move onto machines. New capabilities will create new expectations, and those expectations will create work we cannot name yet.

AI is not the end of human usefulness. It is a force multiplier for human ambition. If we have enough imagination to use it, that is the beginning of a golden age.

Full episode transcript

AI is going to create more jobs, not fewer. I think it is going to bring us into a golden age and create a better life than most of us can even imagine. The fear says AI will take the jobs we have today and leave everybody standing around with nothing to do. My argument is that this starts with the wrong assumption. It assumes work is a fixed pile, human wants have an ending, and once a machine handles today's tasks, history just stops. Every technological revolution makes certain jobs less useful. A tool augments a job, replaces part of it, or makes the old way of doing it look ridiculous. That can be brutal for the person living through the transition. I am not pretending otherwise. But changing particular jobs is not the same thing as eliminating work itself. We keep confusing those two ideas, and then we use the pain of a real transition to justify a fantasy where humans somehow run out of needs, ambitions, and ideas forever. The funny part is that the first people everybody thought were in danger were truck drivers, factory workers, and people doing manual labor. What I am actually watching change fastest is software engineering. AI was created by people who code, so of course code is one of the things it tends to be very, very good at. The people who built the tool, understand it, and use it constantly are also the people whose work is being rearranged most aggressively. Meanwhile, I am not seeing manual labor, trucking, or factory work change at the same speed. There have been attempts to put driverless vehicles on the road for years, but the physical world is still hard. The contrast matters because the popular picture of automation had the order backward. The screen-based knowledge work looked protected, while the physical jobs looked disposable. Right now, software is taking the harder first hit. That does not make physical work immune. It means this revolution is arriving in a different order than people expected. Business software and software as a service are getting hit in insane ways right now. What it means to create software is changing. Teams are being forced to ask how many people and how much time a product really needs when one capable person can direct agents through a huge amount of the build. But I do not think this is the death of the software engineer, the founder, or technology as a whole. It looks more like a shift in how the work gets done and who learns to work the new way. And the engineers who understand agentic AI are not sitting around mourning. They are using it like crazy. There is a whole culture around vibe coding now, and yes, it feels like a mania. It has the energy of a San Francisco gold rush. Everybody thinks they can build an app. People are staying up all night making products, starting businesses, and trying ideas they would have talked themselves out of six months ago because the build looked too hard or too expensive. The barrier to creating a business has never been lower if you know how to use these tools. Things that used to take a long time to learn, coordinate, or pay somebody else to set up can now be attempted by one determined person. That does not mean every app becomes a good business. Most ideas still have to survive contact with customers, competition, and reality. But the cost of finding out has dropped, and when more people can take a real swing, we should expect more attempts, more strange ideas, and more useful things to come out of them. That is the economic mechanism people miss. AI does not only substitute for a task. It makes new capabilities affordable. It lets a small team try something that previously needed a bigger one. It lets a person move from thinking about a business to actually building the first version. Once the cost of action falls, the number of things worth attempting expands. Some old work gets compressed, absolutely. But compression in one place creates room for activity somewhere else, and we cannot predict all of that activity from the job titles we have today. My deeper reason is simple: humans do not run out of things to do. We always want more. We want a better house, a better business, a better tool, a better experience, a better life for our kids, and then something beyond that. No matter how capable AI gets, we are going to keep finding work because desire does not reach a final checkpoint. Give people more capability and they do not become furniture. They look around, imagine the next thing, and start reaching for it. I also reject the idea that people are not meant to work. Work is in our bones. It gives us meaning because it lets us do, give, provide, build, and become useful to somebody besides ourselves. People say family or hobbies can provide meaning, and of course they can. But what is a family if you cannot be the one who does, gives, and provides? They are just roommates. The point is not that every person needs a miserable job. The point is that contribution is part of how human beings understand themselves. So when AI gets better at the boring, shitty, hard work—the repetitive, annoying bullshit—that does not make human beings irrelevant. It gives us more time and more energy for the parts that require taste, judgment, courage, relationships, imagination, and care. I am not promising that every transition will feel good or that every company will handle it well. I am saying the destination is not a planet where nobody matters. The useful direction is less human life spent grinding through work we hate and more human effort pointed at creation. The job-apocalypse story treats the economy like a room with a fixed number of chairs. AI sits down, so a human has to stand up forever. That is not how I see it. AI can also build a bigger room, make new chairs cheaper, and give more people the tools to design rooms nobody had considered. We are not merely dividing today's work between humans and machines. We are expanding what can be done, what can be afforded, and what one person or one company can realistically attempt. That is why I am not worried about there being no jobs. I am worried about whether people learn fast enough, whether institutions adapt, and whether companies use the tools to create instead of only cutting. Those are real transition problems. But they are different from believing humanity reaches the end of useful activity. New capabilities create new expectations. New expectations create new products, services, responsibilities, and work. Some of the jobs will sound obvious in hindsight, and today we do not even have the language for them. Marc Andreessen has made the point that this doom story shows a severe lack of imagination. That is my paraphrase, not a quotation, and I think the point is exactly right. To say AI takes today's jobs and therefore we are all doomed assumes we never invent another product, another business, another need, another experience, or another form of contribution. It assumes the most inventive species we know suddenly becomes passive at the exact moment it receives the most capable tools it has ever had. I do not buy that for a second. My bet is not that nothing changes. Everything changes. Some roles shrink, some disappear, and a lot of people will have to learn a new way of working. My bet is that humans respond the way humans always respond: we want more, we make more, and we find new ways to matter. AI lowers the barrier, carries more of the grind, and gives ordinary people capabilities that used to belong to large teams. That is not the end of work. That is the beginning of a golden age of work, if we have the imagination to use it.

Sources