Will AI Take My Job?

The more unique the problems you solve and the solutions you create, the harder AI will find it to replace you. The more generic your job is, the easier that replacement becomes. This holds even for skilled work: recent research shows generative AI reaches non-routine analytical tasks like writing and analysis, so what protects a role isn't skill on its own, it's how unusual the problems and their context are.^{1, 2, 3}

Take someone whose job is writing a report on how the top 100 companies in the world performed this quarter. That person's in trouble, frankly, because that kind of problem is common and fairly generic, and the information needed (how the top companies are doing) is so easily available online.

That's the thing to understand about AI: it relies heavily on how much information is publicly available, and how many people have already built solutions to similar problems. The more public information there is, and the more existing solutions there are (even with slightly different parameters), the easier it becomes for AI to produce something comparable. This is well documented: studies consistently find AI models perform dramatically better on problems that appear often in their training data, with accuracy gaps of over 70 percentage points between common and rare problems, and reliability dropping sharply on rare or novel ones.^{4, 5, 6}

Now flip it: think about a wedding cake baker. Every couple that walks in wants something different: a specific shape, a specific flavour, a story behind it, a colour scheme that matches their day. There's no generic version of that job, because every single order is its own one-off problem. That kind of uniqueness is exactly what makes AI's job much harder.

In other words: how easily can AI gather the information it needs, and how easily can it deliver the solution? Going back to the company-report example: the data is available online, and the output (a report) is also purely digital. Whenever both the input and the output live entirely in the digital world, AI thrives. Microsoft Research analysed 200,000 real workplace AI conversations across 785 occupations and found exactly this: the highest AI applicability sits with information work (writing, analysis, communication), and the lowest with jobs requiring physical presence, manual dexterity, or hands-on care.⁷

Take the wedding cake baker again: the conversation about the cake happens offline, in person, and once it's had, the cake still has to be physically made and handed over. AI can't help much with either end of that, mostly because it doesn't (yet) have the ability to actually bake a cake.

Now, all of that can sound pretty bleak if your job leans generic and digital. So here's the other half of the story, and it's a genuinely hopeful one.

The loudest headlines focus on jobs disappearing, but that's only half of what's actually happening. The same research bodies studying job losses are also tracking job creation, and the numbers are big: the World Economic Forum projects a net increase of 78 million jobs globally by 2030, even after accounting for the roles AI displaces.⁸ Separate projections put the number of entirely new roles AI could create at around 170 million globally by the same year, many in fields that barely existed a few years ago.⁹

And this isn't just a hopeful forecast. It's already showing up in how companies are actually hiring. A 2026 survey of over a thousand US employers found that among businesses where AI has already changed headcount, nearly a quarter are hiring more people because of it, against just 16% hiring fewer.¹⁰ Separately, a majority of senior business leaders now say they expect AI to reshape and grow their workforce, not shrink it: 60% expect their teams to grow, and 60% expect AI to reinvent human roles rather than replace them outright.¹¹

Here's the part that matters most for anyone worried about their own job: the emerging pattern isn't "AI does your job instead of you." It's AI taking the boring 80% off your plate so you're left doing the 20% that actually needed a human. Across real 2026 deployments, AI agents are mostly picking up the repetitive, low-value tasks (data entry, scheduling, first-draft research, routine follow-ups), freeing people up for the creativity, judgment, relationship-building, and strategic thinking that made the job worth having in the first place.¹² Even Anthropic's own usage data shows this playing out day to day: when people actually use AI at work right now, it leans more toward augmenting what they do than fully automating it.¹³

So the honest version of "will AI take my job" isn't doom. It's more like: the boring parts of your job are the ones at risk, not you. And the jobs coming out the other side of this shift tend to look better, not worse: less admin, less repetition, more of the parts of work that actually needed a person. If your role skews generic today, that's less a death sentence and more a nudge to lean into the judgment, creativity, and human connection that AI still can't touch, because that's where the new, more fulfilling version of your job is heading.