Artificial intelligence: the worker’s assistant or adversary?

In 1811, amid the chaos of the Napoleonic wars, reports of strange disturbances in the Nottinghamshire countryside reached London. Textile workers were smashing cotton looms, attacking employers, and protesting wage reductions. Led by the mysterious General Ludd, incidents spread from the Midlands into the North of England, eventually leading to clashes with government soldiers and the execution of 17 protestors in January 1813.

The Luddites have since entered the historical imagination as fighting a vainglorious defence against the march of progress and embodying the inchoate fear of being left behind by technological change.

As AI expands its capacity in new domains, and use cases for the software accrue, it has resurrected similar questions raised by the Luddites about the relationship between technology and work.

AI: better, faster, stronger

Generative AI is still in its early stages, but there’s growing evidence of its productivity-enhancing effects across a range of sectors. A series of research papers have found meaningful improvements in performance involving AI in consulting, finance processing, writing and even communication with medical patients. Preliminary evidence in contract law found that current AI programs could complete contract review at 0.03% of the cost of human lawyers, and in some cases more accurately.

AI could invert the orthodox relationship between education and technology, where technological change is ‘skill biased’, benefitting skilled workers at the expense of unskilled. IMF research found large portions of professional or managerial work would be affected by AI, and holders of college degrees more affected than those with less formal education. An analysis of AI in customer service found productivity gains concentrated among less-skilled workers.

For proponents of AI, integrating the technology reduces costs, and improves productivity in sectors previously operated by humans, like law or customer support. Jobs will likely be lost as a result. With so many sectors affected by AI, might affected workers feel the same foreboding that gripped the Luddites in centuries past?

Labour, dislocated?

Indeed they might. Among more agile technology firms, the AI rollout is already under way. Swedish fintech firm Klarna announced in February that it was using an OpenAI-powered assistant, doing the equivalent work of 700 full-time staff. Law firm Allen & Overy is utilising a legal AI tool Harvey for negotiation and research purposes. Morgan Stanley Wealth Management is implementing a similar internal tool for its clients.

Beyond the company level, the pace of technological adoption has been accelerating. It took over 70 years to get 90% of US households to use a landline telephone, but a similar level of adoption in computers and cell phones took under 15 years. A report from consultancy Accenture found that 40% of all working hours across industries will be affected by AI. A similar study from the Institute for Public Policy Research (IPPR) found that as much as 8 million jobs are at risk in the UK along from the AI rollout. If this happens faster than we expect, might Luddites be proved right over 200 years later?

Change is the only constant

If so, it would be a deviation from the norm. Fears around the rollout of technology have been seen frequently, accompanying railways, electricity and computing, all now central to modern life.

Straight-line assumptions about how technology impacts work tend to overlook how much jobs themselves change as technology evolves. As an example, the work of accountants has changed substantially since 1950, as technology and automation have changed the job to involve other tasks. It seems reasonable to expect something similar with AI.

Competitive dynamics between firms may also play out in recognisable ways, meaning the AI rollout doesn’t happen overnight. Established firms, with large customer bases and brand recognition, may face off against more adroit AI-supported entrants, offering cheaper services, much like online banking today.

As we cannot be certain how reliable this past maps onto an AI future, we remain watchful for sharp dislocations in labour markets as the technology is rolled out, with accompanying stress in rates and cyclical equities. While the battle between technologists and Luddites is an old one, it remains relevant for investors today.

This is the second in a series of blog posts covering recent innovations in generative AI; the last instalment will discuss the impact on long-term growth and asset prices.