There is an unexpected item in the bagging area: the hidden story of automation.
Last year on a rainy day in April, I walked into a busy surgical clinic during one of my hospital placements at my medical school. I was shadowing a consultant surgeon, a senior doctor with decades of experience. His first patient was a child with chronic hearing loss. He carefully took a history, examined the child’s ears, and discussed the treatment options for her condition with her parents. The family then left, and the surgeon turned around to his desk to document the consultation. However, to my surprise instead of bringing out his Dictaphone he instead took out his smartphone. He explained that the hospital management had introduced a new app for the doctors to narrate their notes. Instead of the traditional method of recording voice notes to later be typed onto the computer system by a secretary, the hospital had automated the process by using the doctors’ phones to directly transcribe their spoken notes onto the hospital system.
Replacement
We often hear that we are on the precipice of an AI revolution. Pundits are sounding the alarm, predicting that AI may soon get better than many of us at our own jobs. Vast numbers of jobs will disappear, and the unlucky will be forced to adapt and find new work.
Last year we watched from our TV screens as SAG-AFTRA’s (the unions for Hollywood’s Writers and Actors) strike action became dominated by the issue of automation. Writers fought for guarantees that their pay would not be slashed, and their credits stolen, whilst actors feared they would be left destitute by movie casts of digital avatars. In a widely circulated Goldman Sachs report in 2023, analysts claimed that 2/3rds of workers were exposed to some degree of automation and that, “generative AI could substitute up to one-fourth of current work”, exposing the equivalent of 300 million full-time workers to redundancy in the near future. Many jobs (like our surgeon’s secretary) may be under threat.
As he began tapping on his phone his face screwed into a look of frustration. I asked what was wrong and he complained that the Wi-Fi wasn’t working; again. The next thing I knew the surgeon was perching in the corner of the room, awkwardly positioning his phone towards the ceiling in a futile attempt to connect to 4G. As the clock on the wall ticked away, I watched his frustration slowly fizzle into despair as he failed to connect to the internet.
Integration and Displacement
Yet whilst thoughts of replacement by AI and other forms of automation dominate our headlines; our surgeon paints a different image. Most of us are heading into a more mundane and yet in some ways more concerning future. One where we must work with AI rather than be replaced by it.
In practice, lots of automation needs help from us humans to work properly. This is the invisible hidden, human infrastructure of automation and the informal and often unacknowledged work we do to get our machines to work.
The Labor of Integrating New Technologies by Alexandra Mateescu and Madeleine Clare Elish, presents the threat of AI as an issue of integration rather than replacement. They argue that we have miss-framed our debate on automation by focusing on its potential capacity in an ideal world. Instead, if our experience is anything to go by, we should be thinking about its integration into our work and recognise that automation often leaves behind workers who are disempowered; instructed by a higher authority to facilitate the technology originally intended to replace them.
A ubiquitous example is the self-checkout machine. Next time you’ve finished shopping and arrive at the self-checkout aisle look around at what you see. In the forest of self-checkout machines, you will find an underpaid cashier swerving from machine to machine, fixing a problem with the scales or helping an elderly shopper scan their nectar card. As the interviews by Elish and Mateescu illustrate, these cashiers, once a familiar face to shoppers, have become an invisible extension of the machines once meant to replace them. Self-checkout machines were developed in the late 1980s and saw mass adoption in the early years of the 21st century, initially heralded as a significantly more cost-effective solution; the imminent obsolescence of the cashier seemed near. Yet the collapse in the number of store cashiers never came. Rather this workforce has adapted to become more part-time, short term and under-trained.
Lilly Irani, an ex-user experience designer from Google, also makes the case for displacement and integration over replacement. She argues that the miraculous appearance of machine learning in search engines and social media is underpinned by a human workforce; as it “is necessary to configure, calibrate, and adjust automation technologies to adapt to a changing world, whether those changes are a differently shaped product or a bird that flies into the factory”. She uses the label cultural data workers or “data janitor” to describe these outsourced labourers, who perform tasks such as data labelling and content moderation.
Often big tech companies outsource the worst of this hidden work to the developing world. A prime example is documented by the Financial Times and Time Magazine, which highlight the unsuccessful efforts of a group of (now fired) content moderators working in Nairobi, to form a union. Working to scrub Facebook of its most graphic content, their case demonstrates that the jobs created by automation often entail poor pay, few rights and draconian working conditions.
Technological Defeatism
The narrative around automation often makes us feel defeated. We are often told that the path of progress is immovable, that resistance is futile and that it would be impossible or even immoral for us to argue otherwise. Yet, these hidden stories of automation tell us a different tale: not a rapid revolution where AI strolls in to take our jobs, but instead a messy process of adaptation, trial and error; a process where we may have more agency than we’d enjoy to admit. As the writer Evgeny Morozov puts it: “We’d be far better off examining individual technologies on their own terms, liberated from the macroscopic fetishes of Silicon Valley”.
Eventually, defeated and embarrassed the surgeon marched out of the room in silence. I looked out the window and saw him standing alone in the car park as he narrated his medical notes. A while later, he walked back into the clinic and announced that ‘he wanted them to give him his bloody cassette recorder back!’
