How should bosses talk about AI?

Job insecurity is not great for anyone. It has the most unpleasant effects on employees. A literature review by Magnus Sverke of Stockholm University and his co-authors—published in 2019, before the start of the generative-AI age—found that it is generally associated with worse health outcomes and decreased job satisfaction. But firms suffer, too. In theory, job insecurity might benefit organisations if it spurred greater employee effort. In practice, it seems to lead to impaired performance.

The degree of trust that employees have in the management of their firms can alleviate the effects of job insecurity, however. A paper by Wen Wang of the University of Leicester and her co-authors analysed data from a national survey of British employees and employers, covering over 16,000 respondents at 1,100 organisations. The researchers found that objective measures of job insecurity, such as cuts in working hours, were generally associated with lower levels of employee commitment, but that this effect was weaker if workers viewed bosses as being honest, reliable and fair.

On this score, Mr Winters got one thing badly wrong but a lot of other things right. His choice of words was his big mistake. Referring to people as human capital is always pretty weird; no one gives a farewell speech saying “what I’m really going to miss is the human capital”. Appearing to divide people into lower-value and higher-value categories was never going to work out well, whatever Mr Winters’s intention.

The clumsy phrasing also missed an important truth. Some jobs are more vulnerable than others to AI: roles that comprise repetitive and easily verifiable tasks, for example, or that do not require interpersonal skills. But exposed jobs cannot be neatly categorised as low-value, as the fate of software engineers shows. Programmers are in the firing line because AI is good at coding and because capital investment in the technology has to be paid for somehow—Meta connected its latest lay-offs to the need to offset other spending.

Look at the fuller version of Mr Winters’s remarks, and in other respects, he deserves credit. In particular, he talked about the pre-emptive efforts that the bank made to find new jobs for employees who were going to be affected by a back-office automation project but wanted to stay at Standard Chartered. Preparing employees for changes in the labour market is the best way to retain the trust and expertise of existing staff. That might mean giving people the skills to do a less automatable kind of job; DBS, a Singapore-based bank, has positioned employees to move from being customer-service agents, say, to becoming salespeople. It also, inevitably, means lots of training on AI tools.

The long-term impact of AI on jobs is unknowable. Firms and employees have different incentives—they are not all in this together. And the burden of managing widespread disruption will fall largely on governments. But bosses can still mitigate job insecurity by identifying the skills they will continue to want from humans, and then helping employees to acquire them. For people to embrace AI, they have to be braced for it, too.

Step inside the world of work with our Bartleby newsletter. Each week our white-collar oracle muses on the agonies of office life.