Column-Tight jobs market? AI meets worker shortage :Mike Dolan

By Mike Dolan

LONDON (Reuters) – The two investment obsessions of the year so far – artificial intelligence and super-tight labour markets – meet head on.

If the hype about the former is to be believed, concern about the inflationary impact of the latter should be well wide of the mark. If only they were so perfectly aligned.

Timing is everything of course. The speed with which ChatGPT-style AI tools zap swathes of white-collar desk jobs could be more glacial than any Big Tech rah-rah suggests – and at least slower than the 12-18 months of the Federal Reserve’s current policy horizon.

But two reasonable questions are being asked around investment houses.

Does the wave of layoffs in the digital and banking worlds this year relate directly to the presumed quantum leap in so-called generative AI – just as pandemic-related overstaffing and more recent job hoarding is being pared back?

And if it does, should policymakers relax more about what could be temporary worker shortages in the service sector, where most of the wage and inflation concerns seems to centre?

Far from relaxing, should office or home-based workers now fret that we’re in for anything but a tight jobs market over the coming years?

More questions than answers perhaps – but enough to have investment strategists thinking laterally and joining dots.

Morgan Stanley’s thematic research team said this week it was inundated with enquiries about generative AI during its recent client visits. And while investment fads come and go, they said, this one is “worth considering seriously” given the speed of takeup and its diffusion across very many industries.

Aside from stock price and valuation frenzies, the team said a new wave of AI fed the debate about white-collar industry disruption in a “creative destruction moment” – with possible side benefits from reskilling workers to better wage diffusion.

Citing numbers indicating employment in business, knowledge, customer and developer outsourcing in excess of 100 million across Asia alone, Morgan Stanley said the impact was already being felt even if the jury was still out on “the degree to which it is deflationary or productivity enhancing.”

If this generative AI takes the tech transformation to non-routine office work that it largely skirted over the last decade, it will affect tens of millions more jobs than currently assumed.

The two sides of the theoretical debate at least are whether that then leads to mass unemployment and demand problems – requiring a reconsideration of things like universal basic income to support economies – or whether productivity gains lift wages and see workers simply choosing to work ever fewer hours over time as bots take their place.

London-based Fathom Consulting on Thursday concluded that a “fourth industrial revolution powered by artificial intelligence could greatly affect the demand for and supply of labour” and the United States and China were bound to vie for leadership.

“The speed and impact of this change will be profoundly disruptive for global politics and for the structure of the labour market,” economists Erik Britton and Andrew Harris wrote, adding the United States needed to keep investing in tech that both supports and replaces labour in order to retain its edge.

Achilles heel

But just what is the scale of the likely disruption?

A frequently cited study by business consultant McKinsey from 2017 showed 60% of occupations worldwide have at least 30% of work activities that could be automated – even though automation may well create more jobs in tandem.

That tallies with numbers from the Organisation for Economic Cooperation and Development, which reckon 10-15% of jobs will be lost due to tech changes over the next 20 years – but about as many may be created in other industries.

While varying hugely among the 46 countries it examined, the McKinsey study said up to 30% of activities could be displaced by 2030 – with advanced and ageing economies more likely to move faster given higher wages and incentives.

More recent polling from McKinsey last year showed companies saying at least a quarter of their tasks could be automated over the next five years but less than a fifth of respondents reckoned their firms were yet in a position to do that.

And that observation underlines the timing of all this in terms of years. How soon do tech revolutions change the world – and at least aggregate demand or supply for workers?

As the flub by Alphabet’s chatbot Bard illustrated in spectacular fashion this week, the big problem for the latest wave of emerging AI is still one of accuracy.

“While ChatGPT’s output is credible, accuracy is its Achilles’ Heel,” Morgan Stanley’s team wrote. “Manual validation should act as a breakwater to this employment threat for now.”

If creases take years to iron out, perhaps it’s not so useful to see the craze providing a timely offset to tight labour markets and wage inflation.

There’s even a chance the trepidation may exaggerate the prevailing conundrum and cause as many problems as the reality.

In a discussion paper published by the Centre for Economic Policy Research last month, economists Marta Golin and Christopher Rauh said their work found a “strong relationship” between worry about automation and intentions to join a union, retrain or switch occupations, preference for taxation and government handouts, populist attitudes and voting intentions.

Much like the pandemic, fear of automation could have as big an economic impact as its actual spread.

The opinions expressed here are those of the author, a columnist for Reuters.

(Reporting by Mike Dolan, Twitter: @reutersMikeDd; editing by Jonathan Oatis)