Guards! Guards!
I was in Bombay recently, at the same time as Independence Day, Dahi Handi, and an extraordinary monsoon. The city was an emulsion of people and taxis, punctuated by Poisson-distributed horns. The city feels full up; there are just so many people here.
A human pyramid of Govindas at Dahi Handi.
As I moved around the city, I began to notice the guards. You might see a guard sitting on a chair outside a shop, perhaps asleep. At an expensive apartment building, hotel, or mall, a security guard will open the gate and check your identity; but alongside the security guard, there are invariably other guards, with a vacant expression, standing alongside. When you drive over the toll bridge from Bandra to Worli, there is someone in each lane, checking the cars that go by. Alongside them, in the same booth and in the same uniform, are one or two guards, checking nothing.
Guards are employees that don’t do productive work. Guards don’t have bullshit jobs - it’s not that the work they do is socially or economically useless. Rather, there just doesn’t seem to be any work for them to do. Guards stand by and watch while other employees do the work. By definition, a guard’s productivity rounds to zero. If an employee adds value, they are not a guard. It’s trivially true that guards exist - of course some people, in a city of 22 million, fit this description! The important claim that I’m making is that there are more guards in India than in the West.
Why do companies hire guards? Perhaps the other workers are too incompetent and untrustworthy to be left alone, and the guards perform a management function - they create redundancy and social pressure to ensure that the job actually gets done. But it doesn’t seem to work that way: rather, one person does the work, and the others - the guards - are helpless or apathetic. Perhaps they’re an insurance policy, and they swing into action during an eventuality like a robbery. Perhaps companies are stockpiling labour in expectation of a future shortage. Is there a frictional reason for the underemployment of the guards? Perhaps there used to be twice as many lanes on the Bandra-Worli bridge, necessitating twice as many people to check the tolls, but management just hasn't noticed yet. This, too, seems unlikely.
My working hypothesis is that guards are employed out of a sense of social responsibility on the part of their employers. There is such a social stigma around unemployment in India that businesses spare people that humiliation as an act of charity. Companies provide guards with a minimal salary, a job to go to, and a place in the social fabric.
Guards are a symptom of India’s problems: the country has low labour costs (although the minimum wage in Bombay is still roughly £13,000 per annum!), high unemployment, and a low skills base. Educated Indians lampoon official unemployment figures. Barely half of university graduates are deemed employable.
But they’re a surprising symptom! If you believe my hypothesis (and if you don’t, please let me know), managers at companies are ignoring their fiduciary duty to shareholders in order to lift people out of unemployment; perhaps India is practicing a form of stakeholder capitalism.
I don’t think we see the same phenomenon in the UK. Labour is too expensive, companies are too efficient, and they lack the sense of community-minded social responsibility that would lead them to hire guards. The companies in the UK that do have guards are often retaining long-tenured elderly workers past the point where their work ceases to be useful. In the UK, young workers who display guard-like traits are either not hired in the first place or fired as soon as they turn out to be unproductive.
But the existence of young guards in Bombay implies one of two things: either guards are hired as guards in the first place, or there is an expectation that some hires won’t work out, and that they will become guards rather than being fired; you hire three security guards, and one will end up doing all the work. The former would be evidence for my “social responsibility” theory of guards; the latter supports the “redundancy” theory of guards - that you need to hire two people to do one person’s job - is part of the hiring process. Sam Wanamaker said that “half the money I spend on advertising is wasted; I just don’t know which half” - maybe the same goes for guards. Either way, guards seem to be a peculiarly Indian cultural phenomenon.
AI is going to cause white collar job losses. Automating away people’s jobs is not an easy task - but it’s possible, and today it’s the hardest it will ever be. It’s happening in domains like law and recruitment, and more broadly it seems to be affecting young people across the US.
Much of the discourse around this fact is concerned with job creation; will new jobs spring up, as they did after previous waves of technological development? But the guards of Bombay provoke a different question. Once AI renders workers unproductive, will companies make them redundant, or keep them around as guards?
Two things stand out to me from the papers above. First, the recruiting study in South-East Asia shows that AI can do jobs (like recruitment) that one might imagine require “human touch”; human touch, like handmade craftsmanship, will become a luxury. While it stretches my definition of “guard”, perhaps guards will be the ones to provide that luxury - the illusion of humanity, the same thing you pay for from a barista in a nice coffee shop. Second, it seems that early-career roles are being hardest hit by job losses from AI. That suggests to me that guards won’t be hired into the job - they’ll be retired into it, rendered useless by AI but kept around. Bad news for young people!
A few years ago, I had lunch with a young insurance broker, who described his job to me. He was a smart guy; far smarter, perhaps, than he should have been, given his job. His job was to process Excel spreadsheets (claims bordereaux). He told me that he’d made a huge amount of progress automating his job with VBA; he works with a lot of quite older people, and they ask him for help. They were stunned by his ability to automate away things they've been doing for years. In fact, he knew that there was one guy who was close to retirement, whose entire job could probably have been automated away. So even without using LLMs, my friend was able to make people’s skills redundant; and the culture of the City of London responded to that redundancy by making its older employees into guards rather than letting them go.
The way that companies respond to AI-generated redundancy will be driven by culture. If I’m right that the guards of Bombay are a cultural phenomenon, a response to a particularly Indian stigma around unemployment, then we won’t see the same pattern in the West. That said, there are plenty of reasons to doubt that conclusion. Perhaps enclaves like the City - enclaves with excess profits and strong cultures - will buck that trend for economic and cultural reasons, rewarding members of the in-group with a job for life. A cynic would say that the Civil Service already has plenty of guards. Perhaps UBI combined with the abolition of minimum-wage laws will reduce the cost to hire to the point that Western companies, faced with an oversupply of labour, will hire more guards than ever before. We shall see!