Speaking at The Economist Innovation Summit in London last month, Morag Watson, vice president and chief digital innovation officer at BP said that “AI will revolutionize the future employment market, creating jobs we won’t recognize and can’t even conceive of today.” This new breed of job, she continued, “risks creating its own form of skills shortage, with employees requiring retraining and upskilling.”

It’s the sort of view we have become accustomed to, ever since Oxford University researchers Carl Benedikt Frey and Michael Osborne published their paper in 2013, on the most likely jobs to be lost to automation. There have been countless studies since, not least a recent OECD study, which diluted Frey and Osborne’s predictions that 47 percent of US jobs and 35 percent of UK jobs were “high risk”. The OECD put the US figure at 10 percent and the UK at 12 percent.

Clearly there is concern. This was exacerbated in the US recently when the Department of Labor released data showing that for the first time ever, US job openings exceeded US job seekers. It suggests something is very wrong with the skillset of the unemployed, despite US Secretary of Labor Alexander Acosta saying that “this is a great time be a job seeker in the United States.”