A recent study from PWC has come to some scary conclusions about the use of AI in the future. It claimed that ’30% of UK jobs could potentially be at high risk of automation by the early 2030s, lower than the US (38%) or Germany (35%), but higher than Japan (21%)’. These headline grabbing statistics have certainly caught the attention of the press. The Daily Mail headline reads ‘Robots 'will take 10million British jobs in 15 years': Storage, retail and admin roles are most at risk, says report’, The Telegraph reads ’Robots will take a third of British jobs by 2030, report says’ and even the usually more balanced Guardian reports that ‘Millions of UK workers at risk of being replaced by robots, study says’.
These news stories all have a similar theme that hundreds of others have had over the past decade - essentially that we all need to be scared of AI because they are going to take our jobs. The huge developments that humankind is making in the creation of artificial intelligence that could have profound positive impacts on our society are generally being covered in the press through negativity. There are stories that clearly look at the potential it could have, but in a news cycle that is dependent on clicks and income, unfortunately fear and negativity are necessary to sell advertising.
It has meant that the conversation, outside of scientific and business circles, is around the potential negative elements of AI. This fear and negativity isn’t totally illogical and is, after all, based on historical truths. We can look back to industrial revolution, where millions of people lost their manual jobs and even spawned the derogatory label ‘Luddites’ after the anti-industrialization group that existed between 1811-1813. There is a historical precedent of technological advancements putting people out of work, we need only look at factory workers in the US today. Despite the blame of the loss of millions of these jobs being placed at the doorstep of globalization, the truth is that ‘trade accounted for 13% of the lost US factory jobs, but 88% of the jobs were taken by robots and other factors at home’ according to Fortune. This has resulted in 5 million manufacturing jobs being lost in the US since 2000, impacting around 1.5% of Americans in that time.
The PWC study estimated that 38% of jobs could be lost to AI, which represents 2433% more jobs lost than in manufacturing automation. It could represent more than 1 in 3 Americans losing their jobs to computers, which is a genuinely terrifying prospect.
However, the only way this would ever happen is if millions of people make very poor decisions, perhaps even the most clearly and demonstrably wrong decisions in the history of the world. The reason mass stupidity needs to take place in order for this to happen is simply that people need jobs as much as jobs need people, every economy has a breaking point and a 38% loss of jobs is well above that line. To put it in perspective, it would put the employment rate at roughly the same level as Afghanistan, a country that has been both ravaged by war and has millions of people living subsistence lives.
It would also bankrupt the government too, with a 38% reduction in taxes collected representing a $1.48 trillion decrease in taxes collected in a year and an increase of $18.8 billion in unemployment benefits every single week, based on an average benefit payment of $310 per person. These two figures alone would represent a loss of close to $2.5 trillion every year, which is 64% of the total tax collection estimate for 2018.
However, this alone doesn’t mean that AI can’t steal 38% of people’s jobs like many of the headlines would have the population believe, instead it means that it is very unlikely that it will. We can look at employment rates in the US across the past 70 years and clearly see that these similar disruptive technologies, like manufacturing automation, digitization, and the invention of the internet have had no impact on the general trend of those in employment. Even though AI represents a threat beyond any of these, it is clear that our economy has a residence to this kind of disruption. There are clear peaks and troughs, but no peak has exceeded 11% and this is likely to be the case with AI too.
There is a very simple reason that this has been the case and it is that only the most poorly run companies fail to realize that you need a market to sell to and if everybody decided to slash that market by 38%, you are not going to be able to be profitable. Instead, companies need to utilize AI to increase productivity per worker, essentially multiplying the value of every worker rather than cutting them completely. It isn’t to say that there won’t be losses, we only need to look at the manufacturing sector in recent years to see that companies run to increase profits rather than employ more people, but these will not be close to 38%. There may be retraining or other opportunities sought, which is fundamentally what the PWC report is saying, but ‘38% of you may have a different job by 2030’ is hardly likely to sell many newspapers.