The Wrong Question

Ask any chatbot which jobs are safe from AI and you get the same list in ten seconds, free, identical for everybody who asks. Trades, nursing, care work, emergency services, complex sales. It is not a bad list. It is just somebody else's judgement about a category, and it cannot tell you the one thing you wanted to know. There is a worse problem with it. In 2013 the most famous study in this field estimated that 47 percent of US employment was at risk. That number went everywhere. A decade later the US Bureau of Labor Statistics checked the occupations it had named most confidently and found they had grown. Heavy truck drivers, the single most predicted casualty of the 2010s, went from 1.8 million jobs to 2.0 million. This book explains exactly how that happened, because the same mistake is being made again right now with different numbers. Then it stops predicting and gives you something you can use: four questions to run against your own week. Does this task happen on a screen or in the physical world. How many times a year does the same thing happen. Does anybody have to personally sign for it. And when a tool takes part of your job away, does what remains get harder or easier. That last question is the one nobody asks, and it is the one that separates two people with the same job title and opposite futures. If automation removes the easy part of your work, your bar rises and your pay follows it. If it removes the part you were good at, your job survives while your leverage quietly does not. Every figure here is attributed and checked against the original source. Where the popular version of a claim does not survive that checking, the checking is shown rather than hidden. Where the evidence is genuinely unresolved, including the argument about whether AI is destroying graduate jobs, the book says so and explains who disagrees with whom. You will not get a prediction. You will get a test you can run again next year, when at least one of the four answers will have moved.

• Why the occupation is the wrong unit of analysis, and why the same question produced 47 percent one way and 9 percent the other
• Where the famous 47 percent figure actually came from, and why it was never as solid as it sounded
• What the current headline number really measures, in the authors' own words, and what they refuse to claim
• What has genuinely been measured so far, including the study that found work reorganising while pay did not move
• Why three serious institutions disagree about AI and graduate hiring, and what that means for you
• The four questions to run against your own week, with a way to score each one
• Why repetition rather than difficulty decides what gets automated first, and the arithmetic that proves it
• Which work the law still requires a named human to sign for, and why that protection is a delay rather than a defence
• How to tell whether a tool is raising your value or quietly lowering it while your job title stays the same
• Why automation-resistant work is not the same as well-paid work, and the number that makes that unavoidable

• Mid-career professionals wondering whether their role survives the next five years
• Anyone whose job title is not on any danger list and who suspects that is not the same as being safe
• People weighing an expensive decision about retraining
• Managers trying to work out which parts of their team's work are actually exposed
• Anyone who has read a confident prediction about AI and wanted to know where the number came from

1. 01Introduction: The list you already have
2. 02Part One: Why nobody can tell you
3. 03Part Two: The four questions
4. 04Part Three: Three weeks, run through the questions
5. 05Part Four: The parts that do not fit
6. 06Closing: Six things worth keeping

لا توجد تقييمات بعد.