BRIAN KANTOR: Future shape of the advanced AI labour market has become a known unknown

The US economy is in great shape judging by the flow of quarterly earnings reports from its leading companies. S&P 500 index earnings have grown about 8% over the past year and analysts are expecting further growth in earnings over the next 12 months — as much as an additional 14% by this time next year.

Sales are growing and profit margins remain at highly elevated ratios. Fears of recession have proved greatly exaggerated and the prospect of further reductions in short-term interest rates is still widely expected, though several big employers have revealed a reluctance to hire more workers, despite their impressive top and bottom lines.

A wait-and-see approach to employing more workers and managers seems to be a widely adopted strategy by US businesses, and understandably so. Waiting to see more clearly what the future of adopting AI means for their workforces seems an eminently sensible approach for all businesses everywhere, including SA.

So the US Federal Reserve has a dilemma and not only because the ongoing government shutdown has denied its policymakers the usual flow of updates on the state of the labour market. There may well be something far more fundamental at work that makes the conventional Fed playbook less useful.

The Fed has a dual mandate as it is instructed to control inflation and to maximise employment with its interest rate and asset purchase settings. And while the prices of goods and services are still presumably subject to the usual somewhat predictable supply and demand forces, the labour market may not be, given the hard-to-predict pace of AI adoption.

The relationship between the state of the economy and the state of hiring and firing may be changing fundamentally, at least for now. As may the link between wage rates, costs of production and prices charged. More (or less) demand for goods and services may not translate into coincident demand for workers.

Lowering interest rates to encourage demand for goods and services may therefore be further good news for the revenue and profit lines of US business and lead to higher prices — without doing much to raise employment levels. If so, the case made now in the US for lowering interest rates is weaker.

That access to AI and AI-empowered robotics will make workers of all kinds ― those behind computers or assembly lines or down mine shafts ― more productive and safer is indisputable, as has been true of technological advances in the past made mostly with the aid of improved mechanics rather than computing power.

Workers clearly produce more when complemented by more and better tools. But for additional profit margins, including covering the cost of the extra capital used, to translate into higher rewards for workers requires a relative shortage of workers. And competition for them. It is this competition for relatively scarce labour that has raised the rewards for work enough for the average worker to consume more leisure, that is, to work fewer hours.

The adoption of AI might also lead to an excess supply of potential workers, enough to drive down the rewards of the average, insufficiently skilled worker while the same forces help promote the exceptional incomes of a highly skilled and productive few. For example, the $1m a year AI engineers now being paid ever more handsomely to move from one AI firm to another.

Among the beneficiaries of more profitable, perhaps less labour-intensive, technology would be the owners of the more profitable firms ― the employers of AI and the computing power. Include among the owners the many members of retirement and pension funds with valuable shares in the AI success stories ― who may then be called upon to shield those unable to find decent jobs by accepting higher taxes ― to reduce inequality, as will be argued.

However, one can be optimistic about the effect of increased productivity on economic outcomes. Intense competition to apply the technology and scale up production for additional profit will mean more produced, lower prices and improved service. Lower prices and greater convenience greatly encourage demand, not only for robots but also for humans to supply a growing market.

Think of Uber as an example. The convenience and price of e-hailing has grown the market for taxi services and the number of drivers. If cutting the lawn or your hair took a third of the time and cost a sixth of the price we would surely trim more frequently and spend more time in the gym.

The demand for AI advanced hair or lawn dressers who can share your woes might well improve rather than decline. Hairdressing for boys does incidentally seem to be a growth story.

• Kantor is head of the research institute at Investec Wealth & Investment. He writes in his personal capacity.