Foreigners aren’t taking our jobs. Machines are!

It was shocking. London Underground was shutting down their ticket offices. Now you would have to buy your ticket from a machine instead of a human being.

It was cold hard business decision that was going save £270 million over five years. But no one thought what would happen if you needed a refund. And, if the ticket machine was broken and swallowed your credit card or money late at night, then tough luck! The saddest part though is that it resulted in the indiscriminate firing of 750 hard working Tube employees. At the time, austerity measures in Britain were really starting to bite, so it just seemed wrong.

I thought to myself, well if the rise of the machines has begun, no one had noticed. We were too busy bashing foreigners and preparing for our economically illogical vote to Brexit. We were pointing our fingers in the wrong direction, and our jobs are at stake.

We could all become obsolete if we’re not careful. We are special today because we bring valuable skills to the workplace, but tomorrow we might not be. No sector will be left unscathed because we are all at risk. If you don’t believe me, look at the McKinsey infographic below or type in your job into willrobotstakemyjob.com to see if your job is at risk from a robot.

Enormous increases in data, significantly improved algorithms and substantially more-powerful computer hardware are achieving what was once thought impossible. These factors are synergistic to one another, amplifying the results they achieve exponentially, so a “singularity” between AI and human intelligence may happen sooner than we think.

Prominent industry figures such as Elon Musk and Nick Bostrom have already warned us about what this could mean. It has spawned a multitude of white papers and lavish conference events. For the first time it’s being taken seriously.

After all, robots can work 24 hours day and seven days a week. They don’t get tired or sick, require pensions, a salary, or a benefits package. And, you can get rid of them whenever you want.

But that’s not what worries me. Though the efficiency gains are evident, there are some serious unintended consequences that haven’t been thought through.

It was a spring afternoon in 2014, and Brisha Borden and her friend spotted a blue Huffy bicycle and silver Razor scooter in the park. They had been left unattended by two six-year-old boys. So, the two eighteen-year-olds decided to go for a ride.

One of the boys’ mothers wasn’t too please and confronted them. A passer-by then called the police and they were arrested.

A year earlier, 41-year-old Vernon Prater was also arrested for shoplifting $86.35 worth of tools from a nearby hardware store.

When both Brisha Borden and Vernon Prater were booked into a holding cell, a computer programme spat out a score based on their profile. It predicted the likelihood of each committing another crime.

Vernon Prater, who is white was rated low risk for reoffending. Yet he had previously carried out two armed robberies and one attempted armed robbery. He had also spent a collective five years in jail, for these crimes.

Meanwhile, Brisha Borden who is black, was rated a high risk for reoffending. Her only previous brushes with the law were four minor juvenile misdemeanours.

Fast-forward two years later from their respective crimes and Prater is serving an eight-year prison sentence for another crime – he broke into a warehouse and stole thousands of dollars of electronic equipment. Meanwhile, teenager Brisha Borden has never reoffended.

Unfortunately, this is not an isolated incident. AI systems have completely missed the mark on racial sensitivity on several occasions. For instance, the New York Times revealed that Google Photos had been tagging black people as gorillas. Meanwhile, Gizmodo has demonstrated how Nikon cameras have assumed that Asian people were blinking when being snapped.

Now add this dysfunctional AI into the workplace and very real human issues such as inequality and discrimination can be exacerbated. Researchers at Princeton University have found that machine learning AI can embed aged-old cultural biases into the pattern of wording.

Male names in Google had a stronger association to certain professions with higher salaries. It’s no surprise, therefore, that Google was found targeting ads for high paying jobs primarily to men. A quick visit to Google Translate can also reveal these biases in action as the image below shows.

Sometimes these self-learnt algorithms are so complicated that it’s impossible to understand how these machines came to the decision they made. Even our most powerful supercomputers are unable to decipher their logic. At a time when robo-advisors and robo-portfolio managers are starting to pop up, I can imagine future asset bubbles and flash crashes occurring in the industry I work in.

But rather than assume the worst, let’s imagine we don’t lose our jobs. Automation is far more likely to replace part of a job than an entire job. As a result, by working side-by-side with humans, overall productivity in the workplace could increase.

Perhaps, if companies make more money with the same number of workers thanks to automation, they can theoretically pay us more. A more likely scenario, however, is that we will be forced to take relative wage cuts over time because a portion of our productivity has been replaced by robots.

There’s another issue – inequality. Economic output may increase, but we may not be the ones receiving the rewards. Instead, it’s the asset owners that profit i.e. those that own businesses, such as shareholders and wealthy individuals. It’s strange to think, but robotics in the workplace could actually fuel the disparity in wealth between the rich and poor.

There are other implications too. Advances in automation could also erode our autonomy and freedom as workers. Employers would have a lot more to bargain with thanks to the efficiency gains they get through automation, which could mean weaker standards in the labour contracts we sign and the disappearance of a proper living wage.

There could also be a cost to society that is not necessarily monetary, such as heightened political tensions spurred on by populist political parties that have gained popularity due to greater income inequality and rising poverty.

But then again, this increase in productivity thanks to automation, could lead to new policy measures that are now affordable, such as a universal basic income. This would put a floor on poverty by paying everyone a low, but liveable income.

Automation could also shift our global economy away from one of scarcity – the old economic problem – into one of abundance. Here, marginal costs would plummet to near zero, making goods and services almost free. The structure of this new economy would be very different from the one we live in today. This is because there would be no need to marshal resources in the way the free-market mechanism operates because automation will make them readily available.

In the meantime, some ingenious suggestions have been made. Bill Gates for instance believes that robot productivity could be taxed, similar in the way that humans are taxed on their incomes. Education and training programmes could also be set up to help workers shift into new jobs: this for instance would make sense for drivers replaced by autonomous vehicles.

However, there little evidence of government action to make our job markets more resilient to change. The cost of further education has never been higher in the developed world, despite living in an abundant information age. Equal opportunities have never truly been resolved even in the US: who your parents are still has a significant bearing on your future success.

So far politicians like Donald Trump have focused their attention on preventing companies from hiring people into manufacturing jobs overseas. What they should be doing is preparing the economy for the impact of automation. If they don’t, then it’s not foreigners that are a threat to our jobs – or even robots – it’s the politicians themselves.