Artificial intelligence is killing digital nomadism

For most of the last decade, the digital nomad was one of the most seductive characters in the whole internet economy. A laptop, a decent Wi-Fi signal, a desk in some co-working space in Lisbon or Chiang Mai, and a steady trickle of remote gigs: copywriting, a bit of translation, some light graphic design, a little virtual assistance. That was enough to swap a fixed address for a passport full of stamps.

After more than twenty years working across marketing, technology and innovation, I’ve seen plenty of technologies arrive promising to transform work. AI feels different, because it is starting to change not just how work is done, but who gets paid to do it. So let me say plainly what a lot of the travel-blog world still won’t: that version of the nomad life is dying, and AI is becoming one of the main forces killing it.

The reason isn’t that people will stop working from Bali. It’s that the work which paid for the nomad life, the small standardised tasks you could describe in a sentence, is exactly what large language models now do fastest and for almost nothing. Freelance platforms are the canary in the coal mine for the whole knowledge economy.

There’s a second half to this, though, and it’s the more honest one. Nomadism isn’t disappearing. It’s being repriced, re-regulated and split hard into winners and losers. Let me go through what the evidence says, and then the part that matters if you’re trying to make a living this way.

The floor under the freelance economy is cracking

Start with the platforms that made all of this possible: Upwork, Fiverr and the rest.

The academic evidence is blunt about what has happened to them. In Management Science, Ozge Demirci and colleagues found that within eight months of ChatGPT’s launch, demand for automation-prone freelance work fell about 21 per cent relative to hands-on manual work, with writing down roughly 30 per cent. The Oxford Internet Institute, studying more than three million postings for the Journal of Economic Behaviour & Organisation, put the drop in substitutable skills like writing and translation somewhere between 20 and 50 per cent, while demand for complementary work such as chatbots and machine learning almost tripled. A paper in Organization Science measured a gentler but stubborn hit, and, tellingly, found the decline still deepening rather than fading.

The detail I keep coming back to, from the Management Science work, is what happened to the jobs that survived. They got more complex, carried bigger budgets and drew more bidders. The easy work vanished, and the hard work got harder and more crowded at the same time.

The most recent and most granular read comes from outside academia: an analysis of 3.25 million Upwork posts found writing jobs down about 54 per cent between January 2025 and June 2026, with editing and proofreading barely touched. That gap tells the story. Clients now show up with a machine-written draft and pay a human to fix it, not produce it.

You can see the same shift from the other side, in the work that has sprung up to replace the old. The Guardian reported that listings on Freelancer.com asking someone to fix AI output, tagged “correct AI” or “AI hallucination”, jumped 87 per cent in the year to June 2026, with similar rises on Upwork and Fiverr. One of the writers in that piece, an Australian editor named Kym Dunbar, now spends around 60 per cent of her time tidying up after machines, while her income from original writing has fallen to roughly a third of what she made during the pandemic. The human hasn’t disappeared. She’s simply moved downstream of the machine, and the money moved with her.

The shift reaches beyond the platforms, too. Axios reported that the American information sector has lost around 370,000 jobs since its 2022 peak, though the end of the streaming boom and a post-pandemic hiring bubble share the blame with AI.

This hits traditional digital nomads particularly hard, because the classic nomad service list, writing, translation, simple design, virtual assistance, basic admin, lines up almost perfectly with the tasks AI eats first. The romantic income model was built on exactly the work that is now being automated.

I want to be fair about one thing. Freelancing as a whole isn’t collapsing. Net demand has held up, and studies of entire national workforces tend to find milder effects than the platform studies do. The platforms are the leading edge, not the whole picture. They just happen to be where nomads live.

It isn’t the robots taking your job

There’s a popular image of autonomous AI agents logging into Upwork and quietly out-bidding humans. It isn’t happening, at least not yet. When researchers built a benchmark to test the best agents on real freelance projects, none of them managed more than 3 percent of the work, earning about 1,800 dollars out of a possible 144,000. Real projects mean bouncing between a brief, Figma, Photoshop, a folder of files and a delivery link, and that is where the agents come apart.

So the threat isn’t a robot doing your job. It’s your client doing the easy first 80 per cent themselves with ChatGPT and never posting the gig at all. The task doesn’t disappear. The paid freelance demand does, because the client now handles much of it before a job is ever posted.

The platforms can feel it happening. Amazon has wound Mechanical Turk down, and Fiverr, having bolted its own AI layer onto the product, cut roughly a third of its staff to run leaner and “AI-first”. The marketplaces that made this kind of work findable are now busy automating it.

AI is redrawing the shape of teams

If this were only about individual gigs drying up, freelancers could retrain and move on. The bigger change, and the one that interests me most as someone who builds and runs marketing teams, is structural. AI is redrawing how teams are put together, especially the remote ones.

The emerging shape is a handful of senior people sitting on top of a lot of tools and agents. The middle and the bottom of the pyramid- the people who used to write the first draft, do the initial research and handle the routine build- are being squeezed out. PwC’s 2026 AI Jobs Barometer found that junior roles exposed to AI are now seven times more likely to ask for skills we used to think of as senior, like leadership and strategic thinking. Entry-level postings in AI-heavy sectors have gone flat, while the “seniorised” entry jobs that remain have grown by more than a third since 2019. The ladder isn’t just getting shorter. Someone is sawing off the bottom rungs.

The hard numbers on young workers are sobering. A Stanford study tracking real payroll data, aptly called “Canaries in the Coal Mine”, found that early-career workers aged 22 to 25 in the most AI-exposed jobs have fallen behind everyone else on employment, a gap that started at around 13 percent and had widened to roughly 19 percent by the middle of 2026, while older and more experienced people in the very same roles held steady. The authors are careful to call it a correlation rather than proof, and they note other factors in the earliest months, but the shape is exactly what you would expect if AI were quietly eating the entry level.

A longer-term problem is hiding inside that finding. Senior people are worth what they are worth because they spent years being junior: producing first drafts, building simple things, researching, getting things wrong and learning from it. Those are precisely the tasks we are now handing to the machine. For any single company, automating them is rational. For a profession as a whole, it raises an awkward question. If the rungs where people used to become experts are gone, where does the next generation of experts come from? Freelance platforms were one of those training grounds, the place you could build a portfolio and a reputation from nothing, and that door is narrowing too.

Microsoft’s research calls this year’s shift unusually sharp, and makes a point I think most people miss: AI still helps a single person far more reliably than it helps a group. That is a big reason the winning unit is turning into the augmented individual rather than the small distributed team. Work is moving from doing to choosing, from making the thing to picking between the options a machine generated, and anyone who stops using their judgement in that handoff slowly loses the skill.

For a team spread across time zones, the maths is grim. The junior remote seat, the one a nomad usually starts in, is the first to go.

The great polarisation

Line all of this up and you don’t get “AI kills freelancing”. You get a market splitting cleanly in two. PwC calls it a two-track economy, and the numbers back it from every angle.

On the losing track is the nomad selling standardised, easy-to-specify work. Upwork’s Future Workforce Index shows earnings for basic AI execution tasks down 28 per cent in a year. Generative and creative production work is booming in volume, up about 90 per cent in contract starts, yet paying less per job. More work, less money per piece, which is what commoditisation always looks like. Microsoft’s economists see the same hollowing-out at the level of the whole economy.

On the winning track, something more hopeful is happening. The same Upwork data shows AI-augmented professional services up 72 per cent, with earnings up 22, and freelancers who fold AI into their work earn roughly a third more per hour than those who don’t. PwC finds the roles AI “professionalises”, the ones it reshapes to demand more human expertise, growing twice as fast as the ones it makes easier, and paying better. Upwork has a name for the person who thrives here, and I suspect it will stick: the AI orchestrator.

An orchestrator is not someone who’s good at prompts. It’s someone who brings real domain knowledge, judgement and accountability, picks the right tools, points them at a business problem and owns the result. They sell outcomes, not output. With a decent stack, one of them now ships what used to take a small agency. A one-person company, in other words, wearing an AI exoskeleton.

Now the part that should worry anyone counting on experience to save them. Where AI can substitute the task, being senior is not a shield but a target. Both the Organization Science study and Wharton’s research find AI flattening the gap between the best and the rest: an average operator can now produce something close to top-tier output, so clients stop paying a premium for reputation, and the highest-rated freelancers in the exposed categories lose the most. Experience protects you only when it means judgement on messy, non-standard problems, plus the ability to orchestrate. Being very good at a task a machine does for free is worth about what the machine charges.

If I had to draw the two profiles side by side, it would look like this:

The same technology is a demotion for one group and a promotion for the other, and the distance between them is widening fast.

The other pressures closing in

AI is the loudest force here, but it isn’t the only one. A few other things are pressing on the nomad model at the same time, and they feed each other.

The first is that competition went global, and got its own AI. Research from IMD, published in Harvard Business Review, argues that AI’s levelling effect actually strengthens lower-cost remote workers abroad: as it closes the quality gap, someone in Manila or Lagos with an AI stack can deliver first-world work at a fraction of the price. So even the augmented Western nomad now competes with an augmented everyone else.

The second is that the attention economy quietly funding a lot of “nomad income” is coming apart. Much of it was never freelancing at all, but ad- and affiliate-funded blogs, and Google’s AI Overviews have hollowed them out by answering readers before they ever click. The travel blogger Amanda Williams wrote openly about losing roughly 40 per cent of her traffic and a third of her ad income in a single year, with Google now serving her own words back to readers who never visit. Quartz quoted a travel-site writer laid off in the same wave, passing on the line she kept hearing from clients: “we used to hire writers and editors, but now we just use AI.”

The third is the return to the office, smaller than the headlines suggest but aimed almost perfectly at nomads. Remote work hasn’t vanished; nationally it has settled at around a fifth of paid days. But the mandates cluster at the biggest, most stable employers, precisely the ones whose salaried remote jobs quietly funded a lot of lives abroad. And “remote” was rarely “work from anywhere” in the first place: once tax, employment law and data rules come into play, few employers actually let staff log in from another country, and a hybrid rota closes the question completely. You cannot do two office days a week from Bali.

The fourth is regulation, which is ending the free ride. More than sixty digital-nomad visa schemes now exist, and each one signals that the lifestyle has been recognised. But recognition brings income thresholds, tax-residency rules and paperwork; governments are treating nomads less like an exotic loophole and more like ordinary residents. More legal recognition means less grey area, and the old trick of working from Lisbon on a tourist visa and telling yourself it’s fine is basically over.

Not an ending, a change of shape

So is digital nomadism dead? No, and the precise answer is more useful than the panic.

What’s dying is one particular version of it, the freelancer paying for a life on the move with interchangeable, easily replaced digital tasks. That model always ran on a gap in the market, clients paying humans for things software couldn’t do yet, and the gap is closing from every side at once. AI takes the task. The client does the draft. A cheaper competitor abroad matches the output. The ad-funded content model erodes. Big employers pull remote roles back home. Governments formalise the whole arrangement.

What grows in its place is smaller, harder and, honestly, more interesting. Upwork’s research suggests skilled freelancing is actually rising, from just over a quarter of skilled knowledge workers to nearly two in five in a single year, and more than half of full-time employees surveyed say they’d consider it. People aren’t fleeing independent work; they’re rearranging themselves around a new idea of what’s valuable. An academic review of AI and digital nomadism puts it neatly: for the mobile worker, AI is both a genuine productivity superpower and a fresh source of risk.

If you’re building or defending a location-independent career right now, the move is uncomfortable but clear. Stop trying to beat AI at what it’s good at, and climb from producing to deciding. Sell results rather than hours. Be the one directing the tools, not the one they replace. Build something you actually own, an audience, a brand, a product, so you don’t live at the mercy of one platform’s feed or one search engine’s mood. And get the boring things right, the tax, the visa, the contracts, because the grey zone that used to forgive sloppiness is gone.

One last thing stands out across much of the evidence. The parts of this life that can’t be automated- the lived experience, the trust, the taste, the simple fact of being somewhere- are exactly the parts that now hold the value. The agents meant to replace freelancers can barely finish 3 per cent of real work. What they can’t do is where your value now sits.

The beach and the Wi-Fi are still there. The easy money that paid for them isn’t. The nomads who make it from here will be the ones who worked out early that the scarce thing was never the output. It was knowing what to make, why it matters, and being willing to put your name on it.