A leading artificial intelligence researcher has issued a dire warning about the future of work, arguing that most human jobs could disappear within the next five years and that society could pass the technological point of no return by 2045. Speaking at length about artificial general intelligence, automation and the limits of human relevance, Dr Roman Yampolsky said the coming changes will be unlike anything seen in previous industrial shifts. Dr Yampolsky, a Latvian computer scientist and professor at the University of Louisville who has published more than 100 academic papers on AI security and risk, made the comments while appearing on The Diary of a CEO, hosted by Steven Bartlett.
‘There is no task that cannot be automated.’
During the conversation, Dr Yampolsky argued that the advent of artificial general intelligence, systems capable of outperforming humans at most cognitive tasks, could occur as early as 2027, with dramatic consequences for employment by the end of the decade. “In five years, all manual labour could also be automated,” he said. “So we’re looking at a world where we have levels of unemployment that we’ve never seen before. It’s not the 10 per cent that’s scary; that’s unemployment, but the 99 per cent.”
Dr Roman V. Yampolsky is a leading computer scientist, author, and researcher specialising in artificial intelligence (AI) safety and security.
Unlike previous technological revolutions, there will be no new category of human work waiting on the other side, he said. “Every task can be automated,” he said. “This had never happened before. All our earlier inventions were like a tool to do something.” Even creative and media work will not be spared. Dr Yampolsky suggested that content creation, including podcasting, could be done more efficiently by machines, telling Bartlett that his own profession may eventually become obsolete as AI systems are faster, more accurate, and more data-driven. “You’re only left with jobs where, for whatever reason, you want someone else to do the job for you,” he said. “There are jobs where you want a human being; maybe you’re rich and for whatever reason, you want a human accountant.” To clarify the issue, he said, “Warren Buffett will not switch to AI. He will use his human accountants.”
Dr. Yampolsky identified five types of work that can survive.
pressed on whether Any While humanitarian roles may persist, Dr Yampolsky outlined a narrow set of exceptions, though he stressed they would support only a small portion of today’s workforce. One category included what he described as “fetishes” for man-made objects. “You may get some small share of the market for people who still prefer man-made crafts,” he said, comparing it to the premium people pay for handmade products compared to mass-produced alternatives. But, he said, this would be “a small subset” and not enough to sustain employment on a large scale. Another area was work based on lived human experience. He argued that mentors and similar roles can retain value because humans uniquely understand what it feels like to be human. What can you offer in a world where superintelligence surpasses humans in every way? He said. “You know better than anyone what it’s like to be you.”
Most jobs may disappear due to AI, but experts say some jobs like AI inspection and consulting, may survive. Image: Pexels
Two more roles will be present because, despite this, instead of A.I., one would involve oversight and regulation. While Dr Yampolsky said it may be impossible to completely control AI in the long run, he argued that human supervision could slow the pace of change. “At this point we’re trying to get more time,” he said, suggesting that regulation could extend the five-year transition to 50 years. The second will be intermediaries, people who understand AI systems well and can explain and deploy them to organisations and individuals who do not.
‘Every day, as a percentage of total knowledge, I become stupider.’
Looking ahead, Dr Yampolsky warned that humanity could cross what is known as the technological singularity around 2045, at which point AI-powered progress accelerates beyond human understanding or control. “This is the definition of eccentricity,” he said. “The point beyond which we cannot perceive, understand, predict, or foresee what is happening in the intellect or the world.” He illustrated this idea using consumer technology. “If I have an iPhone, I can expect a new iPhone to come out next year,” he said. “Now imagine that this process of research and development on this phone is automated. It happens every six months, every three months, every month, week, day, hour, minute, and second.” “You can’t keep up with 30 iterations of the iPhone in one day.” In fact, he suggested, researchers may already be lagging behind. “Apparently, we may already be there,” he said, acknowledging that even experts struggle to keep track of the latest developments. “Every day, as a percentage of total knowledge, I am becoming stupider. I may still know more, as I keep reading. But as a percentage of overall knowledge, we are all becoming stupider.”
For Dr Yampolsky, the concern is not just technological progress, but what happens when human labour, judgement, and relevance are no longer economically necessary, a change he believes will come much faster than most societies are prepared for.
