Automation: labour replacement or displacement?

This thread on twitter the other day reminded me of the all too frequent hype and assumptions around the automation of tasks, and the explicit or implied labour-saving impacts. Rather than mean the cafeteria could save on cashiers’ salaries (and let’s not debate the ethics or economic sense of that right now), an AI-powered self-checkout scanner created more work - and not just for cashiers but also for customers. Instead of replacing labour, it displaced it, and created new forms.

Cashiers had to spend longer telling it whether it had identified the right food or not; microworkers had to input data to train the system to recognise the various different foods and meal options; and customers were asked to arrange the food on their plates in particular ways to aid the AI’s recognition of their (attempted) purchases. 

Many of us will recognise the customer experience of self-scanning at a supermarket, where the scanner won’t scan an item, we can’t find the broccoli listed in the fruit and veg list, or an unexpected item appears in the bagging area. While the jury is out on whether self-scanning has reduced staff headcounts in supermarkets, it has not reduced the overall amount of work that’s needed to scan and pack our shopping, it’s just displaced it onto consumers. 

Touch-screen ordering stations at fast food restaurants like McDonalds also displace labour onto customers, and probably reduce the number of cashiers, but they also create new jobs - someone has had to programme the software, and someone will need to maintain the screens and devices. If these ordering stations bring in more customers, or speed up their ordering process, does this lead to more jobs in the kitchen, or staff to clear tables? I’d love to know what the net loss or gain of staff in a McDonalds with touch-screens is. 

Much has been written about the poor quality, low paid and often psychologically challenging work created by AI; moderators on social media who review images and posts and decide if they breach a platform’s policies. Micro-workers who label and classify images and other data to train AIs. These workers are hidden from view behind the image of AI as fully automated, free from human intervention or oversight. Again, the promise of automation as a labour-saving invention is shown to be a mirage. 

Of course, some technological advances really have saved us time and energy: the washing machine, the personal computer. But this doesn’t mean that all technology will do so, and we should be very wary of assuming it will.