AI at work: Economist says stop guessing which jobs survive
Daniel Susskind argues no one can predict AI's impact on jobs. He urges literacy, numeracy and critical thinking, plus a two-track education: master AI, and flourish without it.

Tom Whitfield
· 4 min read
Nobody, including the experts, can predict which jobs will survive artificial intelligence, according to economist Daniel Susskind — so employers, educators and workers should stop trying. Susskind, an economist at King's College London and a senior research associate at Oxford's Institute for Ethics in AI, makes that case in his new book, "What Should My Children Do: A Human's Guide to the Age of AI." His conclusion: stop guessing which roles will endure and start building skills and character that survive anything.
The book opens with an admission of uncertainty. At the dawn of the internet era, nobody foresaw that "search engine optimizer" or "social media manager" would become jobs. Most strategies for hedging against the future turn out to be fools' errands. Coding, long treated as the safe bet, is no longer a certainty, Susskind writes.
Which skills does Susskind say still matter?
He advises focusing on three basic skills, even if AI performs them better: literacy, numeracy — basic fluency with numbers and data — and critical thinking. Smartphones and social media distracted learners from these skills; AI removes the need to learn them at all. That, he says, worries him most.
Why learn math if AI can solve every problem? Why read a book if AI can answer any question about it? Susskind's answer is blunt. Losing literacy, numeracy and critical thinking would be "a disaster, leaving people unprepared for what lies ahead."
What does his two-track education model propose?
Students should learn to use AI so they can "stand on the shoulders" of the technology and perform tasks that would otherwise be "unthinkable." They should also learn to "flourish without the technology," so basic skills do not die from disuse. Susskind argues that one third of a student's time at college and university should go to learning about AI and how to use it.
"In making this change, there is a huge scope for creativity," he writes. In English classes, students could interpret a basic poem without AI, then a harder one with AI. Teachers could prompt AI to produce deliberately misleading readings of a poem and ask students to critique each one.
Susskind goes further on capability. The latest AI systems, he writes, explain ideas, question students and guide them through problems more effectively "than so many of the teachers I have watched in action over the years (and yes, I include myself in that group)." AI, in his framing, now offers everyone what Oxford's tutorial system once offered a privileged few — but better. That claim invites scrutiny from any L&D leader weighing tool adoption against instructor quality.
What should white-collar workers do?
Susskind concedes AI is warping and ingesting white-collar work faster than blue-collar work. His counterintuitive advice for early-career white-collar audiences is to stay the course. The future, he argues, will need more "domain experts" — experienced humans who know which problems need solving — paired with AI that excels at solving them.
There are caveats. "It is crucial to move into a field because the underlying problems grip you, not the traditional job," he writes. Do not choose law for the image or the lifestyle; become a lawyer because you want to help people through the legal system. The problem lasts while the "job" changes in unpredictable ways.
He also warns of a status reshuffle. The roles most at risk include historically high-valued ones, where humans are merely "a means to achieve some end rather than an end in themselves." People do not want a lawyer, an accountant or a marketing expert; they want to be represented in disputes, get taxes done, and sell products. His counsel: seek work where people truly want a human involved, for aesthetic, empathetic or moral reasons — artists and writers, athletes and musicians, nurses and carers, judges and politicians.
What happens if AI outperforms humans at everything?
If AI advances toward outperforming humans at every economically useful task, Susskind suggests education should shift from building workplace competence toward cultivating character. That means bringing moral questions into classrooms "in a structured way": What is the right way to act? Which laws and institutions are just? What do fairness and community mean? And which problems are worth solving at all?
The book lands amid a live crisis in one discipline. In early October, OpenAI released 722 AI-generated math papers claiming solutions or progress on hundreds of open problems, some among the field's most famous. Mathematician Steven Strogatz, co-author of "Big Math," shed tears in an interview with The New York Times, saying he and his students wanted to know what the point was if they could no longer discover anything on their own.
The book's reviewer, Dan Slater, tested the thesis with his 11-year-old son. He prompted Claude to create an illustrated, multi-chapter story, "The Great Piggy Bank Race," to teach linear equations. By the next day the boy solved them on his own. He still preferred a human teacher, but called the technology amazing — and freaky.
For HR leaders, the working implication is concrete: hiring for durable domain problems and character, not job titles, may age better than any skills forecast — assuming, as Susskind admits, that anyone can reliably tell the two apart.
Source: feeds.businessinsider.com
More from Tom Whitfield
Show full bio
Market editor covering money, funding and market shifts around Recruiting and talent acquisition, learning and development at HR Briefing.
287 articles