Doctors, lawyers, teachers: the “safe” jobs aren’t safe anymore. That’s how Daniel Susskind opened the keynote at Exponential Day 2026 (Barcelona, 8 July 2026).
- The AI fallacy: thinking a machine has to think like us to do our work. Deep Blue didn’t beat Kasparov by reasoning like a grandmaster, it crunched hundreds of millions of positions a second (330 million, in his figure). There are many ways of being competent that don’t look like ours.
- Jobs are bundles of tasks. Few vanish whole, but in about 60% of roles at least 30% of the tasks can be automated (a McKinsey figure). What changes is which tasks keep economic value.
- The answer is education: compete where humans still have the edge (judgement, empathy), or learn to work with the machines, and keep learning all your life.
- On AGI: risks acknowledged, but he’s optimistic about spreading expertise (health, law, education) to people who can’t pay for it today, while care, teaching and art keep their human value.
Not new if you’ve read A World Without Work, but a tidy version of it in under an hour.