, the British-Canadian computer scientist widely known as the “Godfather of AI,” has made a rare public statement highlighting which jobs are most vulnerable, and which ones might endure.Hinton, who played a key role in developing the AI systems that now power tools from chatbots to healthcare technology, stepped away from his position at Google in 2023. Since then, he has spoken more openly about the potential dangers of the very technology he helped shape. In a recent episode of *The Diary of a CEO* podcast with Steven Bartlett, Hinton shared his views on the future of work in an AI-driven world. In an unexpected insight, he named plumbing as one of the few career paths likely to remain relatively secure.

While talking about the risks of AI, Hinton made a simple point: machines are good at thinking, not at fixing pipes. “It’s going to be a long time before [AI is] as good at physical manipulation as us,” he said. “A good bet would be to be a plumber.”

Plumbing is difficult to automate because it demands manual expertise, on-the-spot judgment, and the ability to solve problems in unpredictable environments. A plumber might have to squeeze under a sink, deal with irregular water pressure, or fix an issue without clear instructions—challenges that current AI systems aren't equipped to handle.

According to Hinton, jobs that depend on hands-on skills are far less likely to be replaced by AI in the near future. While machines excel at processing data and generating content, they still can’t climb ladders or tighten valves. That makes skilled trade professions—like plumbing, electrical work, and carpentry—more resilient in an increasingly automated world.

While plumbers may have some protection for now, Geoffrey Hinton believes many office-based jobs are far more vulnerable. He specifically pointed to roles like legal assistants and paralegals as among the first to be reshaped by AI.

These positions often involve tasks such as reviewing contracts, analyzing documents, and summarizing information—functions that large language models are already capable of performing. With AI tools now able to draft legal content and even forecast case outcomes, the demand for human support roles in law firms could decline.

Hinton wasn’t sounding an alarm for the sake of it—he was simply reflecting what the data suggests. “In a society which shared out things fairly, everybody should be better off,” he said. “But if you can replace lots of people by AIs, then the people who get replaced will be worse off.”

In short, while AI may drive productivity gains, those benefits may not be shared equally across society.