By Carys Roberts

At any time of flux there are opportunities to entrench or overcome the problems of the status quo. One such problem is the gendered world of work that results in a gender pay gap of 17.9 per cent, unequal access to jobs and hours, and women doing 60 per cent more unpaid work than men. These problems are magnified by race, class and age.

Automation presents a moment of flux. As automation reshapes the labour market, the change it brings will be gendered. But in what way will depend on the collective decisions of society and of government.

Despite the stories that abound in the media, mass unemployment is an unlikely result of new technologies. Whether automation takes place will depend not just on technical feasibility, but on the societal acceptability of particular tasks being done by machines – from cab-driving to ‘carebots’, as well as whether it makes economic sense for a particular firm to automate. And where automation does take place, the effect on employment will depend on how prices, and, in turn, demand change as a result. If it becomes cheaper to deliver take-away meals, we may find people want more of them. Or, if costs go down, that might free up income for spending on other things.

What is true, though, is that where automation occurs, it is likely to radically reshape what we do at work, and who benefits from the wealth generated. These changes may well affect men and women differently, because men and women tend to have different jobs in the UK labour market.

The literature to date gives different estimates of how susceptibility to automation varies by gender in the UK and globally. That’s because the likely impact of automation depends on a complex interplay of factors, and any attempt to model that implies a set of assumptions relating to those factors. The Office for National Statistics estimate that women hold 70 per cent of jobs that are at high risk of automation. At the Institute for Public Policy Research, we will be publishing our own estimates of how automation could affect men and women’s jobs next month.

But it’s not just the extent of automation that is likely to have a gendered impact. The type of impact could also differ for men and women.

A range of factors, of which automation is one, will mean that particular sectors will expand in the coming years. Our ageing population will increase demand for care. The need to rapidly decarbonise will bring with it new jobs in sectors like renewable energy. But who is able to get the jobs that pay well and have good conditions will mediate the impact of automation on gender. Currently, while the tech sector is expanding, just 22 per cent of people working in artificial intelligence are women. If automation is to enable a more gender equal future, there is a task for policy to desegregate the labour market, including ensuring that new roles being created can be reached by all.

As the structure of the labour market changes and new jobs are created, so too the skills that are in high demand will change. Automation is likely to increase demand for and the premium on technical, emotional and higher cognitive skills. This could be a positive story for women, given that emotional and social skills are typically seen as ‘feminine’. But whether this means women can access the ‘lovely’ jobs will depend on whether demand for those skills leads to remuneration for their use: in our current labour market, there is plenty of need for these skills but the jobs – such as care work – that rely on them, are poorly paid. To address this, a recognition of the full value of feminised work will be needed.

A likely outcome of automation, absent of intervention, is that even though tasks could be automated, in occupations where it remains cheaper and easier to employ low-paid labour, that automation won’t take place. That leaves part of the population in low-paid work while the wages of others pull away due to increased productivity. Currently, the UK has too few robots, not too many: lifting the pay of workers at the bottom of the pay scale will need to go hand in hand with a strategy to increase productivity in those occupations, including through greater use of automation. This could go in either direction. As well as productivity enabling higher pay, there is growing interest in and evidence for the idea that increasing pay, for instance through minimum wage increases, can increase productivity as firms respond with productivity-enhancing behaviours.

Automation could create a ‘paradox of plenty’: be far richer in aggregate, but with entrenched reward and inequality. To avoid this, policymakers will need to think about how financial wealth – which is set to gain from automation – is distributed, including through pensions. They should also think about how we use the rewards of higher productivity. The TUC have argued that this should include working time. If accompanied by changing norms over who delivers care, this could have a powerful effect in rebalancing who does unpaid work.

Finally, as decisions are automated, and if decision-making algorithms are based on biased data, automation by artificial intelligence could exacerbate gender inequalities.

The future is not technologically determined, and to state that automation will definitely disadvantage one group or another is to accept fatalism by the hand of the tech companies. We have collective agency in deciding the future we want automation to bring. We should use that agency to narrow, rather than widen, gender inequalities.

Carys Roberts is head of the centre for economic justice at IPPR and co-author of Managing automation: Employment, inequality and ethics in the digital age.

This blog is part of our Workers and Technology: the key issues series. Read more about the series here.