From technological achievement to technological progress, AI and robotics have advanced enormously in recent decades and are expected to exert a major influence on society and the labour market in the years ahead. That long run trend looks set to be accelerated by Covid, as firms respond to cost cutting pressures and exploit the fact that robots don’t need to social distance.
Whilst these technological achievements are impressive and exciting they don’t necessarily feed into social progress. We need to make sure that we aren’t blinded by amazing technological achievements but focus instead on how we can use wonderful new technologies to improve our lives.
That’s one of the main themes of my latest book (with Lynda Gratton) “The New Long Life – A Framework for Flourishing in a Changing World”. Human ingenuity underpins technological ingenuity but we also need social ingenuity if we are to ensure these technological achievements produce true social progress.
To discuss these issues I was delighted to host, for the London Business School’s Wheeler Institute, a conversation with Daron Acemoglu, Institute Professor at MIT and one of the leading economists writing about technology. Daron’s research interests are exceptionally wide ranging and in particular also focus on political economy. That’s important for issues to do with technology given the breadth and scale of its impact on society in the years to come.
As always, any conversation with Daron is wide ranging but there were three key themes he stressed i) how technological progress unfolds is not inevitable ii) technological progress doesn’t inevitably work for the benefit of society iii) we don’t have to sign up to the Silicon Valley vision of where technology should take us.
His concern is that the future is jobless if AI doesn’t achieve productivity growth. The ability of AI and robotics to perform routine tasks is increasing and this automation will lead to job losses. That isn’t necessarily a bad thing if its supplemented by a shift of workers into other tasks which will become more valuable in the wake of automation. For instance, in education AI could lead to lecturing, examining and testing all being done by machines (machine learning?). However, this could also create whole new roles for teachers focused on really supporting individual students, recognising areas they need help in and focusing on providing that support.
Not only would shifts such as these be good for employment they also should be good for productivity given the value they can unlock and that in turn is good for wages. It can also help make work more rewarding. During the Industrial Revolution commentators from Marx to Ruskin expressed concern at the alienation of labour and the loss of human potential in work. If robots and AI can take away the drudge work we can lessen this problem. As emphasised in The New Long Life, when machines become better at being machines we have an opportunity to become more human.
The problem is that currently there are too many ‘so-so’ technologies which are
producing jobless growth through a focus just on automation rather than creating new tasks and roles. We aren’t thinking enough about how technology can be used to influence the labour market for broader social good. For Acemoglu that’s a problem caused both by technologists and economists. Technologists, he argues, need to ‘correct their way of thinking about technology and the labor market’ but economists also need to think more deeply about how technology works. The canonical textbook model says that technology underpins long run prosperity as it leads to higher productivity, higher wages, more leisure and not higher unemployment. Acemoglu warns that this outcome isn’t inevitable. There are economic and political factors that produce self-correcting forces to the job losses that come from automation. But these self-correcting forces don’t always operate or work to the degree required to ensure that technological advancement produces higher productivity and not higher unemployment. We need to ensure that technological achievement to technological progress works for us.
Supporting better labor market outcomes will require three steps. The first is a societal realization of what is at stake, a public debate about what outcomes are needed and construction of a social narrative to achieve these better outcomes. He points to exactly that process which has started to shape political debate around the environment and led to major breakthroughs in terms of renewable energy.
The second step is the need for government to have its ‘fingerprints’ on the direction of change. The substantive nature of technological change means that it is characterised by numerous externalities (as with the environment and trade) and needs to be driven by a government vision and a consortium view as to what is needed for technological achievement to produce social progress,
This links in with a third step which is wrestling the vision of what technology is trying to achieve away from Silicon Valley. It isn’t the entrepreneurial energy that is the problem but, for Acemoglu, an excessive focus on automation. Part of the measure of success of technological achievement in Silicon Valley is the removal of a human role, the very opposite of what is needed to augment work and support employment. “The sort of vision is one of let’s get rid of the fallible troublesome humans’. That may be appropriate in many sectors as a criteria for automation but it shouldn’t be the main focus on how we judge technological progress.
The lure and vision of Silicon Valley is striking and also affects the way universities think about AI and robotics. Students taking computer science are looking for jobs in Silicon Valley and are often taught by Faculty who interact on a substantive basis with top firms in the region. Such close connections aren’t necessarily a bad thing but creates a uniformity of vision that runs the risk of a focus on technological rather than social progress.
Whilst there are many novel things to marvel at with AI and robotics the issue of how society should deal with new technologies is far from new. During the Industrial Revolution broad swathes of society came together to push for reforms and influence the way in which technologies were introduced and providing the social ingenuity that allied with technological ingenuity brought about social progress. That is the means whereby a social narrative is created and a broader vision around technology is formed.
However just as Acemoglu warns that the self-correcting economic forces that produce good labour market outcomes don’t automatically work so too he warns that we can’t rely on self-correcting political forces. During the progressive era in the US the robber barons used their financial strength to influence debate but were outweighed by mass public movements. Acemoglu’s concern is that the relative power of money and its influence and social movement may be different this time.
In The New Long Life I talk about the ‘Frankenstein Syndrome’ – how we fear that our own inventions will rise up against us and cause human misery. Human ingenuity has achieved two wonderful things – longer healthier lives and fantastic new technologies. Yet we fear both when we talk about an ‘ageing society’ and the ‘rise of the robots’. Just as demography isn’t destiny, Acemolgu reminds us that technology isn’t either. It does feel though that we need to urgently start asking what we want from these changes so we avoid dystopian outcomes.