40% of time spent on mundane chores could be automated within 10 years - AI experts

Grocery shopping is the most ‘automatable’ household task Childcare is the least automatable domestic task Experts in UK and Japan divide on how much time will be saved.

Four in ten hours currently devoted to unpaid housework and domestic care responsibilities could be automated within the decade, according to research, from the University of Oxford and Japan’s Ochanomizu University, in the journal PLOS ONE.

Among household tasks, time spent on grocery shopping was seen as most automatable. On average, experts predicted time currently spent on this task would fall by nearly 60%. Meanwhile, respondents believed time spent on physical childcare would only be reduced by 21% as a result of automation.

Dr Lulu Shi, a postdoctoral researcher with the Oxford Internet Institute, says, ‘Our research suggests, on average, around 39% of our time spent on domestic work can be automated in the next ten years.

‘The degree of automation varies substantially across different types of work, however: Only 28% of care work, including activities such as teaching your child, accompanying your child, or taking care of an older family member, is predicted to be automated. Yet 44% of housework, including cooking, cleaning, and shopping, are expected to be automatable.’

The findings are based on responses from AI experts in the UK and Japan, when asked what difference automation was going to make to housework and other unpaid work. The researchers found the estimates were influenced by the personal background of the experts.

Ekaterina Hertog, associate professor in AI and Society, Oxford Internet Institute and Ethics in AI Institute, explains, ‘We found male and female experts had different expectations about automation of domestic work, potentially reflecting the differences in their lived experiences with technology as well as their involvement in housework and care work.’

The research found male UK experts tended to be more optimistic about domestic automation compared with their female counterparts. This is in line with previous studies, which show men tend to be more optimistic about technology than women. But this was reversed for Japanese male and female experts – and the authors speculate the Japanese gender disparity in household tasks could play a role in these results.

According to the study, the general level of optimism in respect of domestic automation also varied by country. On average, UK-based experts thought automation could reduce domestic work time by 42%, compared with a 36% reduction expected by Japanese respondents. The authors suggest this may be because technology is associated more with labour replacement in the UK. In Japan, meanwhile, new smart technologies are expected to work alongside humans rather than replace them.

Previous studies show working age people in the UK spend nearly 50% of all their work and study time on unpaid domestic work such as cooking, cleaning, and care. The new findings suggest a large potential increase in leisure time as domestic tasks get automated.

The effects are likely to affect women more than men, however. In the UK, working age men spend around half as much time on domestic unpaid work as working age women. In Japan, the difference in time spent on domestic tasks is even more striking, with Japanese men spending just 18% of the time spent by women on domestic tasks. Technologies that save time currently spent on domestic work could, therefore, result in greater gender equality at home, according to the researchers.

The study involved 29 male and female AI experts from the UK and 36 experts from Japan. They were asked to estimate the degree to which 17 housework and care tasks might be automated over the next decade. The study’s diverse sample is not statistically representative but, as the authors note, the experts’ backgrounds offer the potential for contextualising their predictions.

Few studies have examined the automation of unpaid domestic work or predictions about automation and how they differ - depending on the AI experts consulted.