AI will take our jobs – but it can happen in an orderly fashion

There’s no doubt that artificial intelligence will replace humans in many roles, argues Nikos Bozionelos, but we have to act now to make sure the process is smooth

AI will take our jobs – but it can happen in an orderly fashion

In the last months of the pre-Covid era, I attended the speech of an accomplished scholar in computer science who assured us without any hesitation that AI will not take our jobs. However reassuring, that statement is problematic. Firstly, it doesn’t specify the time frame (never? Or not during the lifetime of the generations represented in the audience?) and second, it doesn’t specify the breadth (‘will not take anyone’s job’ or ‘only a few jobs’?).

Leaving semantics aside, the way most people understand it – that there is no danger of mass unemployment because of AI – that statement is most likely incorrect. In the few years it has been around, AI has already taken jobs, and will continue doing so. AI will make people redundant because it will perform the functions they perform more efficiently, but will not create an equivalent number of jobs for these people to take; and the pace of that process will accelerate. The major difference between AI and earlier technologies (such as machines, programmed machines, computers, and information communication technology (ICT)) is that it will create very few jobs in comparison to the number it eliminates, and even the creation of such jobs will be diminishing because net job creation will be increasingly negative. 

A counterargument is that our jobs survived the industrial revolution and the information revolution. Indeed, the industrial revolution created new jobs, partly because it was combined with expansion in trade and financial advancements; and the computer and ICT revolution created needs for numerous new jobs, especially ‘cleaner’ ones. If that has arguably already happened twice, then we should expect it to happen again. 

There are two ways to view that argument. The first is philosophical: surviving serious illness twice in the past does not necessarily mean we will survive the third time we fall ill. The second is to view that argument factually: how AI (and robotics) is different from earlier technologies, and how this may (or not) differentiate its impact on jobs. AI means that a ‘machine’ is able to learn and adapt its responses and behaviour accordingly independent of human intervention. The very core of AI is that over time and with experience – and without any human aid – the ‘machine’ changes the way it makes decisions (or physically acts, in the case of a robot) and improves either the quality of the decisions or effectiveness and efficiency of the behaviour. This occurs continuously and increasingly quickly. AI is also capable of planning, and in the future will be capable of motives and intentions (this is already a reality for some experts). 

The difference, therefore, between earlier technologies is in learning and decision-making, and everything that comes with it, for example creativity and innovativeness. These earlier technologies were taking physically challenging and/or repetitive physical and cognitive tasks away from humans, who would then move into tasks that required more complex and more frequent learning, with decision-making and creativity remaining firmly within the human jurisdiction. AI is different because for the first time technology intrudes into the very core of human uniqueness: learning, decision-making, creativity, innovation and sensory-motor coordination, in the case of robots. The clear implication is that eventually, there will be no jobs for humans because the higher-order human capacities will be simulated and bettered by machines. Simply, there will be no higher-order functions for humans to shift into. There is already clear evidence for this, from the fact that computer science-related jobs are becoming less available (most jobs that were performed by software engineers in the early 2000s are now performed by algorithms, with no equivalent alternative job creation) to the rapid ‘thinning’ of the middle class (the people who were traditionally occupied with creation and utilisation of knowledge to make decisions. 

Though the specifics of the disappearance of jobs for the majority of humans is an open question (a realistic window would be between the end of the next two decades and the end of the century), the simple answer to the question “will AI take our jobs?” is yes. The consolation is that, provided there is the will, we may be able to go through that process in an orderly fashion – that is, elimination of jobs will occur gradually while we are preparing our social system for it. As we have found during years of research into sustainable careers and employability, two conditions are necessary to meet that significant challenge: first, key stakeholders and decision-makers (politicians, scientists, employers, and the lay people) are aware of this issue (currently, most of them are either not aware, or they think the benefits of AI outweigh the dangers); and second, they are willing to act in a coordinated fashion soon.

Nikos Bozionelos is professor of international HRM and organisational behaviour at Emlyon Business School