Choosing a career in 2026 when nobody knows what jobs will look like in 2040

I’ve had this sentence repeated to me dozens of times while working with families: “65% of children entering primary school today will end up in a job that doesn’t exist yet.” It’s attributed to the World Economic Forum, but its 2016 The Future of Jobs report merely quoted it as “a popular estimate”, in a footnote. The figure can be traced back to presentations from the 2000s, with no source study: the BBC (More or Less, 2017) debunked it, and its lead author stopped using it. This number has become a cliché of career guidance, and it is famous, unverifiable and yet instructive.

Unverifiable because nobody has ever found the original study. Instructive because the phenomenon it caricatures is, itself, well documented: according to the same World Economic Forum’s Future of Jobs 2025 report, 39% of workers’ core skills will be transformed by 2030. Some jobs are born, others die, that’s a historical constant. The real problem is that, presented this way, the observation paralyses rather than enlightens: if we don’t know which jobs will exist, why bother choosing a path at all?

This article tries to move past that paralysis. Not by predicting the future (I’m no fortune teller), but by identifying what doesn’t change in a world that changes a great deal, and what that means concretely for a high-school student making choices right now.

What we really know about the jobs of tomorrow

Jobs don’t “disappear”, they mutate

The classic example: the “secretary” role has all but vanished in its 1980s form (shorthand and typing, paper filing, switchboard). But the secretary’s skills (organising, coordinating, communicating, prioritising) are now spread across a dozen roles (office manager, executive assistant, project coordinator, personal assistant). The job mutated, it didn’t disappear.

AI will accelerate this phenomenon: the repetitive, document-based tasks of a job will be absorbed by tools. But the human core of the job (judgment, relationships, creativity, responsibility) remains. And that core reappears in a new role, often better paid and more interesting than the old one.

Three big trends we can observe with certainty

1. Value is shifting towards interpretation, not production. Producing a standard text, a piece of code, a run-of-the-mill design: AI does it. Interpreting a need, judging a situation, making a decision under uncertainty: AI doesn’t. The jobs of tomorrow will be jobs of interpretation.

2. Being cross-disciplinary is worth more than narrow specialisation. An engineer who understands business, a lawyer who understands tech, a doctor who understands data: these hybrid profiles are already the most sought after, and that will only intensify.

3. The ability to learn is worth more than what you’ve learned. An engineering degree in 2026 is proof that you can learn difficult things. It’s that ability that has value, not the specific content of the syllabus (which will be partly obsolete in 10 years).

The 4 capabilities that won’t lose their value

Rather than “which career to choose”, I encourage the families I work with to ask: which capabilities does my child need to develop to be resilient in a changing world?

1. Learning how to learn

This is the meta-skill, the one that makes all the others possible. A student who can learn on their own (read technical documentation, follow a MOOC, experiment, ask for help in the right place) will be able to change careers 3 times in their life without drama.

What this means for guidance: favour programmes that teach you how to learn (the prépa, or intensive university-preparation classes; the grandes écoles; demanding research universities) over programmes that teach “a trade” narrowly. A BTS (a two-year vocational higher-education diploma) isn’t worse, but it should be topped up with further study if you want this ability over the long term.

2. Communicating with humans (not just writing)

Negotiating, persuading, listening, rephrasing, presenting to an audience, managing conflict: these skills are anti-AI by nature. AI writes well, but it doesn’t negotiate, it doesn’t reassure a patient, it doesn’t motivate a team.

What this means: leadership experiences in high school (class representative, running a club, team sport, drama) aren’t CV padding, they’re training grounds for the most irreplaceable skill of the 21st century.

3. Solving fuzzy problems

A “fuzzy problem” is one that’s poorly defined, ambiguous, with incomplete data and stakeholders whose interests conflict. The opposite of the well-posed problems in maths exercises. Real professional life is made of fuzzy problems.

What this means: programmes that confront students with open-ended projects (internships, business cases, research, hackathons) develop this capability. Programmes that only offer multiple-choice tests and single-answer exercises do not.

4. Working with AI (not against it)

This isn’t about “learning to code” (even though that’s useful). It’s about understanding what AI can do, what it can’t do, and how to use it in your own field. A lawyer who uses AI to speed up case-law research genuinely gains from it: studies measure gains of roughly 10 to 30% on legal research and drafting. In medicine, the results are mixed: AI alone sometimes outperforms the doctor-plus-AI pairing on diagnostic reasoning (a 2024 JAMA trial), but the benefit is clear in imaging.

What this means: whatever field is chosen, a basic digital literacy is necessary. Not necessarily a computer-science degree, but a familiarity with the tools, a curiosity about tech, and no fear when faced with a new tool.

What I tell the families who reach out to me

When a parent tells me “my child doesn’t know which career to choose and it’s making me anxious”, my answer comes in two parts:

First part: “That’s normal. At 17, nobody knows. And it’s good not to know, because committing too early to an ultra-specific job means putting yourself at risk if that job mutates in 10 years.”

Second part: “The question isn’t which career, it’s which path best develops the capabilities your child will need no matter what: learning how to learn, communicating, solving fuzzy problems, working with tech.”

And then we look concretely at which programmes develop these 4 capabilities, depending on the child’s profile (science? Humanities? Both? Structure or independence?). The career comes afterwards, and it will come, because a student well trained in the 4 capabilities always finds their way, even if it didn’t exist when they started.

A final word on parental anxiety

The anxiety “I don’t know which career to choose for my child” is often the parents’ anxiety, not the child’s. The teenager, for their part, often has an intuition (sometimes vague, sometimes very clear) about what interests them. The parents’ role isn’t to choose the career in their place, nor to impose a “safe” path. It’s to give them the conditions to develop the 4 capabilities, and to trust them to build what comes next.

Higher education isn’t a financial investment that has to “pay off”. It’s the cognitive foundation of a 40-year working life. And the best foundations aren’t the ones that predict the future, they’re the ones that let you adapt to any future.

Key takeaways

• “65% of the jobs of 2040 don’t exist” is a famous but unverifiable figure. What is documented (WEF, Future of Jobs 2025): 39% of core skills will be transformed by 2030. Jobs mutate, they don’t vanish into thin air.
• The right question isn’t “which career” but “which capabilities to develop”: learning how to learn, communicating, solving fuzzy problems, working with AI.
• The programmes that develop these 4 capabilities are the ones that confront students with complexity (projects, research, internships, cross-disciplinary work), not the ones that lock them into a rigid syllabus.
• The anxiety of the “right career choice” is often a parental one. The parents’ role is to provide the conditions, not to choose the career.
• A student well trained in the 4 capabilities will always find their way, even if it didn’t exist when they started.

Going further

• Will AI really replace the jobs my child is preparing for today?
• World Economic Forum: The Future of Jobs
• OECD: Employment Outlook

Article written by Constantin Mardoukhaev, co-founder of Axiom Academic. Constantin guides French-speaking families through their orientation choices, including when the only certainty is that the future can’t be predicted.