DeepSeek, a Chinese startup, has released two large language models that rival the power of the dominant tools but, it says, were developed at a fraction of the cost.  As artificial intelligence (AI) becomes cheaper, requiring fewer and smaller data centres that use less energy its availability will increase.  Much has been made of AI’s potential to improve productivity by taking over routine tasks.  Does this mean there will be fewer jobs?  Perhaps we have been here before.

Edward Hollingsworth, a 19th century hosiery manufacturer, equipped his Nottinghamshire factory with mechanised stocking frames.  In November 1811, seeing a threat to their traditional skills and fearing for their livelihoods, a group of textile workers broke into the factory and destroyed several machines, beginning the Luddite movement.  However, although the need for skilled weavers diminished until it almost disappeared, the new mechanised mills brought down the price of textiles, which increased demand until more people were employed to operate automatic looms than had been employed as weavers. Associated industries, such a fabric dying, also expanded.

Automated teller machines (ATMs) began appearing in the USA in the 1960s and 70s with wide anticipation that the number of human bank tellers would fall.  The number of tellers in a branch did fall and those who remained no longer dealt with routine deposits and withdrawals but spent their time on more complex services such as loan consultations, financial planning and customer service.  A branch was now cheaper to run, needing fewer people who could now spend their time gaining customers, so banks opened more branches to compete in new locations and the total number of human tellers increased.

Today, we see companies taking advantage of AI for simple customer service interactions powered by chatbots while escalating more complex questions to a human.  Does this mean that companies are employing fewer customer service staff, or does it mean that the staff it employs are free to respond to more complex enquiries, while the customers that could have got their answer from web site or who ask the simple common questions receive a faster response?

At the Queen Elizabeth University Hospital in Glasgow, a pioneering project is using AI to improve the efficiency of CT scan processing.  The technology examines head CT scans and accurately pinpoints critical areas of concern, creating prioritised reporting that allows prompt diagnosis and treatment of head injuries.  Not only does this improve the patient experience, it also alleviates the pressure on the A&E department.  Will this reduce the number of jobs for specialist radiologists?  That seems unlikely, given your correspondent’s recent experience of being told it would take six weeks for a radiologist to review the images of his scan.

So is AI coming for our jobs?  Mechanisation took the weavers’ jobs but created opportunities for machine operators.  ATMs allowed bank tellers to do something other than count money.  Will AI replace last mile delivery drivers and long-distance lorry drivers?  Will financial analysts be replaced by automated trading and AI financial analysis systems?  Or will AI simply allow us to get more done?