3 August 2015

5 jobs we will never let the robots take


The robots are taking our jobs!

Yes, yes, it’s a refrain we’ve been hearing since the Industrial Revolution, but this time it’s serious. Driverless cars, giant automated production facilities that make those cars, robotic vacuum cleaners that can navigate the floor of your home, real-time language translation and even algorithms that can write news articles – is there anything that the computers can’t do, or can’t be trained to do in the future?

Not really, no. But that shouldn’t worry us, writes Geoff Colvin, author of How Humans are Underrated. The point is not to try to find jobs that robots will always find impossible, searching for the most complicated and seemingly random occupations we can think up. Instead, we need to forget ability, and work out which jobs we, as the human race, will always insist are carried out by another human.

“A large category of those activities comprises roles for which we demand that a specific person or persons be accountable. A useful example is making decisions in courts of law, which we will require that human judges render for quite a long time to come. It’s an example in which the human vs. computer question is not hypothetical. Parole decisions are made by judges in some countries, such as Israel, where researchers investigated how those decisions are influenced by the critical human issue of lunch. Over the course of a day, the judges approve about 35% of prisoners’ applications for parole. But the approval rate declines steadily in the two hours before lunch, almost to zero just before the lunch break. Immediately after lunch, it spikes to 65% and then again declines steadily. If you’re a prisoner, the number of years you spend behind bars could be affected significantly by whether your parole application happens to be the last one on the judge’s stack before lunch or the first one after. Data-driven algorithms have proved superior to human judges and juries in predicting recidivism, and it’s virtually certain that computer analysis could judge parole applications more effectively, and certainly less capriciously, than human judges do. Yet how would you rate the chances of that job getting reassigned from judges to machines? The issue isn’t computer abilities; it’s the social necessity that individuals be accountable for important decisions. Similarly, it seems a safe bet that those in other accountability roles—CEOs, generals, government leaders at every level—will remain in those roles for the same reason.”

With this new criteria in mind, here are five jobs that I am fairly confident will never be stolen by the robots, and why.


We may think it inconceivable for doctors of any kind to be computerised, but, as Colvin points out, it’s already happening: “researchers at the University of California at Berkeley are training a robot to identify and cut away cancerous tissue—not like today’s surgical robots, which are actually tools used by human surgeons, but entirely on its own.” And if surgery, the most delicate and sensitive of medical specialties, can be performed by robots, the rest are soon to follow. All except therapy, that is. It isn’t that a robot wouldn’t be capable of memorising the entirety of academic psychiatry and using it to diagnose patients – all the evidence implies that it would. But would the patient trust the judgement? Moreover, would individuals be prepared to pay thousands of dollars per session for a computer to tell them what’s wrong with their marriage? However much we try to medicalise it, therapy relies on human connection, and even if the Pepper robot can sense, respond to and emulate our emotions, in any kind of high-stakes setting, we are not going to let it.

Priest, Rabbi, Mullah, or other Religious Leader

Theoretically, this one should not be beyond the scope of artificial intelligence. Leading a congregation firstly requires a thorough knowledge of religious scripture. With most mainstream religions, that’s thousands of pages of text, often in an unfamiliar language. It would be far easier for a google-like algorithm to store every spiritual text ever written, recalling relevant sections when required, than for a human to attempt the same. And if robots can write news articles, why not sermons? Why not eulogies and speeches for weddings? Because we would never listen to them, that’s why. Even when artificial intelligence becomes so adapt at the Turing Test that the congregation would never know the difference, no one wants to be offered religious guidance by a computer programme which, by definition, cannot have an understanding of the illogic of faith, however convincingly they can fake it. Pastors, rabbis, monks, you’re all safe.


Really, athletes should probably have been automated years ago. Would it really be that much harder to programme a tennis player than a car in busy traffic? For individual sports like running, gymnastics and archery, it’d be even easier. And of course, a robot could far surpass human ability. Which is exactly why these ‘jobs’ (if you can call them that) will never be automated. Sport is lucrative because of the entertainment aspect, and what’s entertaining about watching a robot with super-human strength and accuracy take down an ordinary human? And while Robot Wars may have been a success, it’s a very different spectator experience to watching Arsenal beat Manchester United or Andy Murray claim a Grand Slam title. The personal edge, where we empathise and identify with the athlete’s struggle, just wouldn’t exist. So no robot athletes for now.

Sports Coach

If our athletes are going to be human, so are our coaches, however irrational this may be. As with Colvin’s example of judges, a robot coach might actually prove more objective and effective than a human one: free from bias or emotion, with an enhanced ability to pinpoint errors and potential. But getting the best out of a team isn’t just about analytics – it’s about psychology, bringing players together, ensuring personalities gel as much as possible, and dealing with the emotional fallout when things go wrong. Even if a robot was theoretical capable of these things, no team would award it the respect necessary. Simply put, a computer coach would have no authority, and would therefore be useless. (Incidentally, a computer referee would face exactly the same problem, even if its decisions were flawless every time, proving there are some things we value more than accuracy.)


This is probably the most significant one of all. Even if schools eventually encounter the technology revolution they’ve been avoiding for so long, and our children are taught by computer programmes far more competent than a teacher could ever be, there will still be humans picking them up at 4pm. This isn’t just a safety issue, based on the paranoid fear that one day the robots will turn against us. (If that were the case, why are we considering giving computers control of cars and weapons?) It doesn’t matter that a robot parent would never lose its temper over a late bedtime or forget to pack lunch for the school trip, or that it could deliver short lessons every topic from ants to astrophysics in child-friendly language. The most important thing a child needs to learn is how to interact with other humans, which is something we learn from those around us. A robot might be able to trick us into believing it’s a human during a short conversation, but would we ever risk finding out whether children raised by robots are as convincing? Never.

So if you want to make sure your job is safe in the next innovation overhaul, consider working out, becoming religious, or having a child. Or you could try beating the computers at their own game, and take a job as a ‘Robot Boyfriend’. It turns out play-acting an imaginary online partner is something only humans can do.

Rachel Cunliffe is Deputy Editor of CapX.