AI: Which jobs are most at risk from the technology?

As the man widely seen as the godfather of artificial intelligence (AI) warns about growing dangers from how it is developing, businesses are scrambling to see how they can use the technology to their advantage.

Geoffrey Hinton, 75, who used to work for Google has warned that AI chatbots could soon be more intelligent than humans.

Many businesses bosses are telling me that the hot topic presented at board meetings is how to deploy ChatGPT style technology across their businesses as quickly as possible.

A few weeks ago, I watched as the boss of one of Britain's biggest consumer-facing companies looked at his computer, entered the transcript of a customer complaint call, and asked ChatGPT to summarise it and respond to it based on set of rules he made up on the spot.

In about a minute it came up with a very credible answer, with no need for any coding.

The end result was, I'm told, about 85% accurate. That is a bit less than human call centre staff, but it cost a fraction of a percentage point of the cost of deploying staff.

The good news for all, the pure enhancement to productivity, would occur if all the staff were now focused on the 15%, that could not handled by AI. But the scope to go further, and cut back on staff, is clearly there.

AI Large Language Models are, however, getting more powerful. Not yet quite as capable as an intelligent adult, but not far off.

Advances are occurring faster than expected, and could be reaching the point where they become exponential.

The pace of change and adoption means there is scope for an economic and jobs shock to the economy as soon as this year.

The moment it becomes cleverer than the cleverest person, in pretty short order, we could get to "runaway capability" - more advanced than the entirety of humanity, on the way to what has been described by another former Google AI insider Ray Kurzweil as the "singularity". Are we at the start of that exponential moment right about now?

AI has the possibility of taking a bunch of sectors of the economy, which have been immune to productivity improvements up until now, because they were time and knowledge intensive sectors, and transforming them.

Technology has given us lots of improvements in the quality of life. All of our smartphones now have all the content we could want, always instantly available on streaming services.

One top policymaker told me that "a lot of that innovation has made our leisure time more enjoyable. It's not made our working time, more productive. It may have eradicated boredom as a human experience. But has it made you more productive at work?"

The real shock has been that these technologies are usable in a commercial context, not just for "low-cognitive, repetitive" - i.e. robotic - tasks, long thought susceptible to automation.

The surprise has been how deployable these technology is to highly creative, high-value work, which had been assumed to be relatively protected from competition.

The Open AI/ ChatGPT founder Sam Altman has himself expressed his surprise at the use so far. Specifically, the "blank page" or "first draft" stage at the start of the creative process of writing copy, creating an image, or music, or coding a programme can be achieved in seconds rather than weeks of briefing and refining.

Again this is what is possible with AI's not-yet-as-intelligent-as-an-adult human. So the good news is that rapid deployment of this technology, faster than the rest of the world, could solve the UK's longstanding productivity crisis.

The bad news is that it could occur so rapidly as to overtake workers' ability to adapt in time, creating social and economic crises. Could we face in call centres and creative studios in the 2020s, the equivalent of what happened in the coal mines in the 1980s?

Some of the people most reluctant about the size of government in Silicon Valley have started to suggest that states might need to provide a basic income. The response of techno enthusiasts is the mantra: "You wont be replaced by an AI, but you might be replaced by someone who knows how to use AI".

But they used to say that's why everybody should learn how to code. That might not be such sage career advice any more.