From the Spinning Jenny to the desktop computer, humans have always invented new tools to make their work lighter, more efficient and more productive. While these inventions have generally made our lives easier, there has always been a corresponding concern about the impact they might have on jobs. The classic example is of 18th century Luddites who destroyed industrial machinery which they considered to be undercutting their salaries and which they feared would displace them. Many argue the answer to such fear is that technology has always led to new jobs – the Industrial Revolution created more work and in a greater variety than had ever previously existed – it’s an argument which, in theory, should apply to the future too.
After decades of development, Artificial Intelligence (AI) – machines that can autonomously learn and find solutions to tasks without human input – is now seen as a credible replacement to human workers in a wide range of professions. The first instances of this expected revolution are staggeringly close – Google’s famous self-driving cars are in testing, Microsoft’s Delve which works like a PA and gets to know users’ needs is now widely available, and less glamorously, the automated checkouts now common in supermarkets are relieving shop workers of customer service duties.
Also, there’s a major debate about what all this means for humans. Proponents of AI favour a more relaxed attitude; as robots learn how to carry out human tasks, new tasks that the machines are unable to do will be created to replace them. However, there’s also a great deal of concern about the possible negative impacts of AI. As little as a decade ago, films about robots usurping humanity seemed completely preposterous, yet a range of prominent scientists and industry experts have, over recent years, predicted dire consequences of unmanaged AI. In late 2014 Stephen Hawking warned the BBC: “The development of full Artificial Intelligence could spell the end of the human race.” The short-term threat was outlined in a 2013 paper, The future of employment: how susceptible are jobs to computerisation? by University of Oxford academics Carl Benedikt Frey and Michael A. Osborne, who estimated that up to 47 per cent of US jobs were at risk of being automated within the next 20 years.
So, is AI something to fear, and if so, how should it be managed?
Which jobs will go?
In their 2011 book Race Against the Machine, Eril Brynjolfsn and Andrew McAfee, both researchers at MIT, explored the reasons behind stagnating wages, growing inequality and unemployment in our societies. In contrast with more traditional economic arguments, they suggest that many workers are simply unable to keep up with advances in the technology, which is replacing them at work. Specialised robots have already taken millions of factory jobs around the world, and the outlook isn’t great in other fields either.
Office workers and professionals have tended to think of themselves as protected from this kind of technological advance, assuming that AI will only ever be able to carry out repetitive, mechanical tasks. Nonetheless, AI is creeping into fields such as medicine, law and financial services too. Doctors, for instance, are likely to face increasing competition from AI as it learns to analyse symptoms and prescribe treatments as effectively as GPs (if not more so). AI is simply much quicker than humans and is therefore able to scan an enormous range of up to date scientific literature. As a consequence, machine doctors may therefore be able to provide more effective treatments than their human counterparts. “Watson”, the AI machine developed by IBM, which famously won US trivia, show Jeopardy! In 2011 is currently being primed to carry out exactly this kind of task.
Are we right to worry?
If Google can build cars that drive us, IBM machines that improve our health and Microsoft systems that replace the need for boring administration tasks, is that really such a bad thing? Austin Tate of Artificial Intelligence Applications Institute (AIAI) of the University of Edinburgh talked me through some of the tools the team have produced over the last 30 years. Key achievements include Ghost Writer, a program which provides multilingual support for aero-engine maintenance manuals, Formation, a system which lays out Yellow Pages directories and fraud detection programs which carry out case-based reasoning. All of these tools appear to be benign, and it’s tempting to see the critics as scaremongering.
Perhaps the strongest argument against fearing for our future is that people simply misunderstand what AI actually is. Over recent years there’s been a growing group of AI sceptics, led by so called “neo-humanists” such as Jaron Lanier and Nicholas Carr, who complain that many people hold an almost magical belief about what computers can do. Yes, they argue, the machines we have built are incredibly powerful, yet describing them as intelligent is a total misnomer – they’re simply very sophisticated tools carrying out very complex tasks. They might do highly complex activities, yet they don’t actually know they’re doing them.
That’s all very well, but how should the truck driver feel when a Google-designed lorry replaces him or her? The solution is simple, the sceptics argue: we need to work alongside robots, let them do the hard, mechanical work, while we humans carry out tasks which require skills only we have – empathy, care, and creativity. The ideal scenario sees humans being improved and augmented by machines, our “higher” skills flourishing as we no longer need to carry out mundane and repetitive labour.
No easy answers
I spoke to Oxford scientist Dr Carl Frey who co-authored the 2013 study mentioned earlier. While his research suggested 47 per cent of US jobs were at risk, he emphasised, “this was what we considered technically possible – that doesn’t mean they will all be replaced, but the potential for them to be replaced exists”.
However, while jobs will be taken by machines, that doesn’t mean half of the US will find itself suddenly unemployed, according to Dr Frey: “What we’ve seen with the digital revolution of the last couple of decades is that a lot of new jobs have also been created, often in areas we never would have expected before.”
Even if AI does reach its full job-replacing potential, it’s very likely we’ll see a whole range of new jobs springing up in fields we never could have imagined before. Most obviously, it’s highly likely there’ll be many jobs for software engineers and data scientists as well as technicians to repair machines.
However, while he doesn’t predict an employment disaster, Dr Frey also warns against a naively optimistic view: “No-one knows whether there will, in fact, be enough new jobs which spring up to replace the old ones, it’s simply something we cannot predict… there is no law that says industries are actually bound to produce new jobs for people whose positions have been replaced.”
While it might not be clear what the future holds, Dr Frey was much more certain of what we need to do to prepare for a world where humans and AI live side by side. Of course, it’s going to be essential to prepare younger generations with skills that will serve them most effectively. There’s going to be a growing need for IT and technical knowledge, yet it will also be important to encourage creativity, lateral thinking and interpersonal skills – the very same abilities machines will probably never master.
Dr Frey also pointed to some less obvious yet practical policies which we will have to implement in years to come. In particular, he predicted a pressing need for improved infrastructure and cheaper housing: “What we’ve seen with the digital revolution so far is that, despite the promise that it would make geography redundant, location is still extremely important. Most new jobs are skill intensive and rely on close interpersonal working as people solve problems together. As a result we’ve seen the emergence of hubs in places like San Francisco or London, and these kinds of hubs will keep growing.
There’ll be a need for cheaper homes and better infrastructure to help people get to the places where work is found. This will make it easier to build hubs that will then create jobs.”
It will be some time yet before we can really gauge the impact of AI on jobs, and until then it’s very hard to predict just how significant or otherwise it will be. What we do know, however, is that it’s going to have often unpredictable effects on a wide range of jobs that have never previously been threatened by automation. It will therefore be crucial to begin implementing policies which will help people prepare best for the coming era.
Nonetheless, with well-informed ethics boards, intelligent policies and adequate training, there is cause for optimism: a world where dangerous and mundane jobs are performed by machines, and where AI supports and augments our creative capacities could be just around the corner.