Across classrooms and at dinner tables, children are often asked: “What do you want to be when you grow up?”
But in 2026, there’s a bigger conversation lurking thanks to the growth of AI — one that parents never really had to contend with in the past: “What job will exist by the time they grow up?”
The search for AI-proof jobs seems to change every week as these tools evolve, making it difficult to figure out what will eventually be outsourced away from humans.
Ebrahim Bagheri, an interdisciplinary researcher at the University of Toronto’s faculty of information whose work focuses on artificial intelligence, responsible AI and information retrieval, says AI technology has been a topic of conversation in academia since the 1960s. About a decade ago, radiologists were cited as an occupation that could soon be replaced by AI. Software development, however, appeared to be among the safest and most promising career choices.
Today, Bagheri notes, the picture looks very different.
“Radiologists are not even at risk of being displaced at this point,” he says, whereas some junior programming work is now under pressure thanks to rapidly improving AI coding tools.
Young students shouldn’t outsource their thinking, even while the tools are there
Maura Grossman, a research professor at the University of Waterloo, a lawyer, and a clinical psychologist whose work focuses on the trustworthy and responsible use of AI, began researching machine learning long before ChatGPT became the hot topic of popular conversation.
Her advice to students seeking AI-proof futures is simple: “Young people have to learn about what these tools can do and what they can’t do.”
Although these tools can provide obvious benefits, she cautions students against outsourcing their thinking, writing and reasoning to AI. She compares it to physical exercise: stop using a muscle and eventually it atrophies.
“You’re going to graduate from college, and you won’t know how to think,” she says.
For parents choosing schools, extracurriculars and eventually post-secondary paths, that may mean resisting the temptation to view every traditional skill as obsolete simply because a machine can now produce a version of it.
Writing still matters. Reading still matters. With the rise of generative AI among the general public, so comes the mass crisis of critical thinking, media literacy, and the ability to read. Students must learn to form an argument, change their minds when presented with evidence, communicate with others, solve difficult problems, and form their own opinions without the immediate intervention of a chatbot.
Take the unconventional path
Jake Hirsh-Allen, director of partnerships and advocacy at public policy and leadership think tank the Dais at Toronto Metropolitan University, spent almost a decade at LinkedIn working across higher education, learning, recruiting and labour market initiatives. He recommends students concentrate on human skills.
“Focus on how they work with other people,” he says, pointing to communication, critical thinking and clear reasoning.
He recommends taking a longer view: figuring out what genuinely interests a student and building the foundational abilities that can transfer across professions.
And for parents, he has an even simpler message: “Don’t get caught up in the hype cycle.”
The fear that AI is already eliminating an enormous number of entry-level jobs is, in his view, overstated in the short term. He cautions against steering children toward a handful of supposedly prestigious or secure professions simply because they seem financially safe today.
“You don’t need to be a doctor or a lawyer to do well,” he says.
Instead, parents can give children room to explore unconventional careers, from the arts to skilled trades, and support them while they discover what holds their attention.
That may sound counterintuitive at a moment when AI can generate a painting, compose a song or draft a short story. Yet Hirsh-Allen believes creativity remains one of the distinctly human abilities worth developing. AI can produce something that appears creative, he argues, but that does not necessarily make its creativity human.
“There is a way to create an AI-proof job,” he says. “And the way to do that is to focus on the things that are most human.”
And his favourite example may actually counteract what you’ve previously been told about AI. “Creativity is actually, I think, one of my favourite examples, because I think we can, at least till now, very clearly distinguish the kinds of creativity that are human from what Gen AI can create.”
It’s an argument that makes the humanities newly interesting in the age of AI. The skills developed through studying literature, history, philosophy and the arts provide a distinctly human lens. It allows students to interpret ideas, construct arguments, communicate clearly, think critically and creatively, and understand the human experience. Precisely the kinds of distinctly human abilities Hirsh-Allen believes students should be strengthening.
“If you look at people’s salaries over time, those who went into humanities have consistently done better than the sciences,” he says. “It’s just right out of school that they don’t do as well.”
Hands-on skills are key for an AI-proof future
For the moment, another obvious AI-proof answer is work involving the physical world. Bagheri points to one reassuring factor: tasks humans consider taxing can be relatively easy for computers, while skills humans perform almost without thinking can be extraordinarily difficult for machines.
Although AI can write an essay in seconds, it has a much harder time fixing a leaking pipe, giving skilled trades and other hands-on professions an advantage. Electricians, plumbers and some medical professionals, for example, perform physical work in an environment that cannot be replicated by software, making them AI-proof.
“Do you think you would like a robot to cut your hair?” Grossman asks. For now, probably not.
Bagheri also points to jobs requiring human accountability, positions where a person or institution must ultimately be responsible for a consequential decision, as well as jobs built around connection. Therapy, hospice care and coaching are examples.
Even when an AI system produces accurate information, he says, research has shown that people often still prefer interacting with another person in areas where trust and emotional connection matter.
Hirsh-Allen echoed much of the same in his value of the “care economy.”
If he could redesign where society directed its resources, he says, he would put them toward “people taking care of people.”
Those jobs may not always be the highest-paid today, but he argues that they are among the most socially valuable and the hardest to pass off to a machine.
That is why Bagheri is much more interested in what he calls “meta skills” than any individual occupation.
Among those skills are persuasion, communication, emotional intelligence, teamwork and the ability to learn continuously.
“The next few years will be a few years where you have to rapidly evolve and learn new skills to remain in demand,” he says.