Last year, the BBC launched a website where people could see how likely it was that a robot would take their job in the future. It was based on some research from Oxford University and Deloitte. All you had to do was select your job title, and it would give you your odds as a percentage.
Journalists like myself are lucky. We have just an eight percent chance of being replaced. Medical practitioners were even safer, with odds of just two percent. That’s because it’s hard to teach a robot creativity and imagination. It’s even harder to teach how to interact with humans.
But others aren’t as lucky. Take retail cashiers, for example. They have a 90% chance of being replaced by machines. You can already see this happening in most supermarkets, as automated self-checkout machines increasingly take up more floorspace.
Taxi drivers have reason to be worried too. They’ve got a 57% chance of being replaced by robots. Hardly surprising, given the incredible progress being made with self-driving cars.
It may come as a surprise that a number of desk-based office jobs are under threat too. In fact, of the top ten most at-risk jobs, virtually all of them are office-based.
If you’re a finance manager, you’re almost certain to be replaced with an AI program. They’re at a 97% risk. Receptionists and Personal Assistants aren’t much better off, as they’ve got 96% and a 68% chance, respectively.
As it turns out, there’s a number of historical precedents for office jobs being automated.
A Long, Machinated Past
Jobs have traditionally been categorized as either white collar or blue collar.
White collar jobs tend to require higher academic certifications and professional skills, are based in offices, and tend to be centered around administration and management. Blue-collar jobs tend to involve manual labor. Actually, physically doing something.
More recently, another “collar” has been defined. Pink collar jobs are those that are centered around customer interaction and sales. These are jobs that are based in restaurants and shops.
There’s a kind-of collective assumption that blue-collar jobs are the first to go when technology advances. Meanwhile, jobs based around soft, interpersonal skills (those would be your pink-collar jobs) and those that require high levels of expertise and specialization are secure. And why wouldn’t we think so? Our collective cultural and historical canon are filled with examples of how semi-skilled and unskilled workers lost their jobs, thanks to new technologies.
One of the best examples of this was found in 19th century England, during the Industrial Revolution. Advances in machinery and technology, coupled with steam and water power, had allowed the factories of the North to become the most productive in the world.
Stocking frames, spinning frames, and power looms had allowed factories to reduce labor costs, and to produce large quantities of textiles that were previously unthinkable.
This resulted in the small towns and cities that dot Greater Manchester and Northern Yorkshire to become some of the richest in the British Empire. But this technological advance left a great many people behind.
Previously, the textile industry was dominated by skilled craftsmen, who were made redundant by the shift to machination. They simply couldn’t compete. They cost more to employ, and they worked slower. The machines, on the other hand, could be operated by an unskilled worker with just the minimal amount of training.
This shift to machination resulted in an extended period of civil turmoil, as the now-unemployed craftsmen attacked the factories operating the new equipment. Machines were smashed, factories and warehouses were torched, and industrialists were murdered.
It has been claimed that the first person to destroy these machines was called Ned Ludd. It is from Ludd where they derived their name – Luddites. This term persists today, and is a derisory word used to describe those opposed to technology, not just that used in textile fabrication.
There are, of course, more contemporary examples of how automation has devastated the employment prospects of blue-collar workers. You need only look at the automobile industry.
The first move to automation was made in 1961, when General Motors introduced their first robotic arm. Since then, robotics have became common-place in the factory floors of Detroit and Kentucky. Subsequently, Europe, Japan, and North America have hemorrhaged auto jobs.
Office Employment and the Machine
But you’d be mistaken if you thought it was just blue-collar employment that has fallen victim to automation and improved industrial processes. The same has happened to office-based employment. We just don’t talk about it all that much.
Long before the invention of the transistor and the vacuum tube, and before digital computers became financially viable for large companies, computers were actual people. The job of a human computer was to quite literally compute things. They would perform complex calculations by hand, following pre-defined algorithms and algebra.
These were commonplace in the offices of science, industry, military, and finance. But ultimately, they were replaced as it became affordable for businesses to buy their own electronic computers. Not only did they allow companies to save on labor costs, but they also worked faster than the human computers, and with greater precision.
Computers don’t make mistakes. They just do what you tell them to do.
There was a deeply sad side to these “Human Computers”. The early 20th century was a fundamentally patriarchal society, and for many highly educated women, it represented the highest position they could get in the sciences. One of the most touching accounts of this can be found in “When Computers Were Human” by David Alan Grier, whose own grandmother worked as a computer.
Another long-forgotten casualty of the digital age was the typing pools, often called “secretarial pools”.
These were secretaries who weren’t assigned to any one employee in particular, but were shared among an entire company. Their job was to type, store, and manage documents, minutes, and correspondence.
Again, this was a highly female dominated field, largely as a consequence of the time in which they existed. Secretarial jobs were seen as being best suited to women.
The electronic computer was the death-knell for the typing pool. Companies no longer needed twenty employees to type the same letter, when mail merge in Microsoft Word worked just as well. They no longer needed people to manually type copies of documents, when they could just be saved to a hard drive.
Automation and the Office: What’s Happening Now
I think we’re close to seeing the end of many office-based jobs.
Companies, who are under increasing financial pressure, are looking to innovations in machine learning and artificial intelligence, as well as the automation of repetitive tasks, in order to drastically cut costs. It’s been happening for a while.
One personal account of this dates back to 1997, and was published on the Tales from Tech Support community on Reddit. This largely consists of stories from front-line IT support workers, who are complaining about the challenging customers and clients they’ve had to support.
The story, which was titled “My first day on the job, and I accidentally got the secretary fired”, was published by user Zarokima. It describes the first day on his job, after he completed his masters degree in Computer Science.
“While the boss is showing me around, he gets an important phone call leaving me outside with his 3 secretaries — he was always very busy, and would be lost without their assistance. We strike up some conversation about our jobs, and one complains about how she has to keep track of some stuff on the server to make reports that the boss wanted daily, and it’s just the most boring, tedious crap.”
Eager to impress the boss, and to help the secretary with her day-to-day tasks, he wrote a small script on his lunch hour that automatically created the reports for her. It transpired that the creation of these reports was the sole duty of the secretary, and she was immediately made redundant.
The author of the story ended up receiving a promotion, and a subsequent rise in his wages.
While we can’t guarantee that this story is accurate, or just something that was contrived for Reddit Karma, I don’t doubt that many jobs could be replaced with automated scripts. Jobs where tasks are repetitive and outcomes are predictable are especially vulnerable to this.
This is essentially what happened to the automotive industry.
The Threat From Consumer Grade Automation
But we’ve also seen jobs which require creativity and human interaction become machinated. Translators, for example, have a mere 33% chance of being automated. There have been advances in translation algorithms, but they still can’t match a human translator when it comes to accuracy, and understanding the nuance of a language.
Unbabel works by first translating the passage of text through an algorithm, much like Google Translate. A human fluent in that language would then go through the text, and check that it reads correctly, and would resolve any errors. While not fully automated, it’s certainly not far off.
Personal assistants (PAs, who if you’ll remember, had a 68% chance of being automated) are another profession that’s presently being replaced by complicated, AI algorithms.
PAs have a lot of responsibilities. They manage correspondence and maintain calendars. They prioritize schedules and make travel and hospitality bookings. But many of these tasks can now be performed (and performed well) with cheap, consumer-grade artificial intelligence.
Right now, Facebook is road-testing their flagship personal assistant AI, called Facebook M. Although it’s only available to a small cadre of US-based beta testers, it’s looking really promising.
That’s because it uses a medium of interaction that almost all of us are familiar with – Facebook Messenger. It’s also great at understanding what you mean, thanks to its use of Natural Language Processing (NLP) algorithms. So, if you type something “Find me somewhere to take a client to lunch”, or “send my mom some flowers”, the odds are good it will understand you.
Facebook M is so potent because it ties into a number of third-party services through its APIs (Application Programming Interfaces). This allows it to make travel bookings, order products, make reservations, and even provide recommendations about places to eat and visit.
Although Apple’s Siri and Microsoft’s Cortana have both made overtures at being truly comprehensive personal assistants, very few have come close to Facebook M. I can imagine Facebook M one day being a business-class product, that will ultimately make some question the value in having a human-based personal assistant on the payroll.
How Businesses Are Using Bespoke and Enterprise-Grade Automation
Some businesses have tasks which are so complicated, they cannot depend on off-the-shelf automation and AI products. They have to either create their own, or outsource the problem to another company. Britain’s New Statesman magazine profiled a number of these companies last year.
One of the companies highlighted was the mobile network O2.
Owned by the Spanish phone giant Telefonica, the O2 brand is present throughout much of Europe, with operations in Ireland, Slovakia, the Czech Republic, and the UK.
In the face of an ever-crowded mobile marketplace, and increasing threats from budget carriers, O2 launched a campaign of cost-cutting in 2012. By using an automation program bought from Blue Prism, they were able to reduce their dependence on offshoring and slash the number of customer service jobs.
This automation program allowed them to process simple customer service tasks with a minimal of human interaction. These tasks include the replacement of SIM cards, the porting of phone numbers, unlocking phones after the conclusion of a contract, and migrating customers from prepaid plans to contracts.
One of the UK’s largest banks, Barclays, has also been using an AI built by the Blue Prism Limited. This is the same company that produced the software used by O2 to streamline their customer service processes.
This AI was used to help the bank cope with the thousands of requests for insurance reimbursements, in the wake of the Payment Protection Insurance (PPI) scandal.
In both of these cases, AI and automation was used to replace jobs that were either occupied by humans, or could have been.
Not Just Automation: How the Sharing Economy Threatens Office Work
Could a change in business model also have an equally damaging effect on office-based jobs as automation?
Over the past ten years, we have seen the world of work become increasingly casual. Jobs which were once steady, and came with benefits like healthcare and paid time off, are being transformed into digital services that you can summon with a smartphone. This is a natural consequence of the mainstreaming of the sharing economy, and our increasing appetite for cheap, on-demand labor.
This race-to-the-bottom started off with traditionally blue-collar work, but increasingly, office-based jobs have been falling under the spell of the sharing economy.
The quintessential sharing economy product is Uber, which seemingly has just as many detractors as fans. Uber works by pairing casual drivers with willing passengers, essentially creating a new class of taxi driver overnight. On one hand, they had to pay their own expenses. However, they were not beholden to same rules as traditional taxi drivers, and they could work the hours that suited them.
Through the tap of an app, you could summon a craftsman or cleaner. You can even summon someone to assemble your flat-pack Ikea furniture. But don’t confuse them with mainstream employment. You’re being connected with an independent contractor, and you have little-to-no ongoing obligations to them.
In recent years, we’ve seen a few of these companies transition from unskilled and semi-skilled jobs, as they start to offer on-demand, skilled labour.
TaskRabbit in particular have pivoted, and now they allow companies to employ personal assistants, data entry workers, and even web-developers and designers on the same informal, short-term basis as they would a cleaner. Although these would be considered office-based jobs, these “TaskRabbits” work remotely, and can be based anywhere.
Another service, called UpCounsel, allows people to hire lawyers on a similar basis. The range of legal specialties on offer is absolutely dizzying, and it promises to be as much as 60% cheaper than engaging directly with a law firm.
The End of Employment as We Know It?
The office of the future will look a lot different to how it does today.
It will be leaner and depend more on automated processes and artificial intelligence than on human assets. It will probably be more elastic too. When demand increases, offices will be able to simply increase the number of machines they have running in order to cope. When the task can’t be automated, they’ll be able to summon some short-term workers from a service like Task Rabbit.
But what does the bigger picture look like?
For years now, people have been predicting that computer technologies and automation will have a destructive effect on employment. Some have predicted that they’ll cost more jobs than they’ll create, and ultimately there won’t be enough to go around. There’s even a term for it – “technological unemployment”.
Perhaps the cruelest irony is that while the first jobs to be lost to computerization were predominantly held by women, the jobs that are currently most vulnerable to automation are being held by men.
It’s important to remember that these jobs will be lost at every end of the employment spectrum – from the blue-collar jobs, to those in administration, and beyond.
It’s what we do next that counts.
There’s always a hope that a future breakthrough will make up for the jobs that have been lost to AI and automation. This seems plausible. After all, although human computers and typing pools were made redundant by the electronic computer, it resulted in the creation of thousands of jobs in IT departments.
These new systems needed people to maintain them, and people to write software for them. They needed people to train others in how to use them.
But if that breakthrough fails to happen, our society will look radically different.
We might all find ourselves working part-time, in order to ensure that everyone has an opportunity to have gainful employment.
Our leaders might introduce an unconditional, minimum basic income (often called “mincome” after the Canadian program of the same name), where those unable to find employment will be free to pursue their own interests, hobbies, and spend time raising their families.
Experiments with basic income in Canada, the United States, and the Netherlands have been incredibly promising.
Our Automated Future Awaits
If we can learn anything from the Luddites, it’s that technological advances are a kind-of Pandora’s Box that cannot be undone after the fact. Automation and AI will undoubtedly have a radically transformative effect on our labor force, and it’s too late to stop it.
Whether that will necessarily be a bad thing remains to be seen.
Are you worried about automation? Do you think it could ultimately be beneficial? Are you in a job that’s at risk of being automated? How are you planning to cope? I want to hear about it. Leave me a comment below, and we’ll chat.