FOR the moment at least, automation is occurring all too slowly.
The displacement of workers by machines ought to be bringing on a consistent rise in productivity and a faster growing economy. This is not happening in the US, Europe and Japan as well as in most parts of the world, including Malaysia – all being persistently disappointed with the outcomes. Experts warn that rapid advances in robotics and artificial intelligence (AI) could destroy millions of jobs and pose a “Terminator” style threat to human life as we know it.
However unnerving the prospect, present day limitations of AI appear all too evident. There is a lot of over-promising and a real risk of under-delivering in my view. At this time, I don’t see AI delivering any life-changing “bolt-out-of-the-blue” product; certainly not representing any imminent threat to humanity.
Robotics: Bill Gates recently created controversy by suggesting that governments should consider a tax on robots – either on their installation or on profits made from savings of labour displaced. Revenue could then be used to retrain workers, including on health and education. It’s not a good idea because: (1) robots are a capital investment; they work as a replacement of labour; (2) it deters new investment, without raising too much revenue; (3) robots make workers more productive rather than expendable; (4) why pick on robots?; in practice, it’s difficult to separate labour-saving activities from labour enhancing ones; (5) workers as a whole are better off with robots since the end result is availability of cheaper and often, better goods and services; and perhaps, most fundamentally (6) society should enjoy the extra output from robots, and find more suitable ways to tax or transfer to protect displaced workers. Prof Larry Summers’ thoughtful piece on this issue recently concluded that such a tax works to reject technological progress. As always, my Harvard friend is right.
Anxiety over jobs
Many are fearful that a fresh wave of the second machine age involving technological advances in robotics will render many workers redundant. The smarter machines and computers become, the greater the likelihood that the space remaining for uniquely human skills will shrink further. The central banks in England and Italy had estimated that up to 50% of jobs in UK and Europe are at risk of automation. But not everyone is panicking, however. Certainly, mass unemployment is not at risk. The basic fact is, I believe, that technology do eliminate jobs, not work.
Evidence points to winners and losers as middle-skilled jobs (eg routine secretarial, clerical and manufacturing tasks) have declined, but the ratio of people at work has not budged much. So far, there is no sign yet that technology is supercharging productivity by supplanting workers with robots. If anything, labour productivity (output generated per worker) has been slowing in developed as well as in emerging nations like China and Brazil. Japan, well known as the home of robotics, has productivity growing below the OECD average, and far lower than the US and Germany. Indeed, technological breakthroughs have yet to reach the workplace en masse. Driverless cars are not yet replacing jobs; but it is not difficult to envisage that they eventually would.
AI: Realistically, experts’ warnings of the widespread substitution of machines for human labour has yet to come through. After many false dawns, AI has since made extraordinary progress; thanks to “deep learning” (making use of deep neural networks, modelled on the brains’ architecture) AI today can do all kinds of things – they power Google’s search engine, Facebook’s automatic photo tagging, Apple’s voice-assistant Siri, Spotify’s music playlist, Tesla’s self-driving cars, etc. But, even the most inventive AI have a tough time getting the most sophisticated vacuum cleaners to identify dog-poop! Sure, AI is not rocket science – it’s already embedded in many products we use every day. Already, the AI market in the US is huge: US$645mil in 2016 and expected to rise rapidly to US$35bil-US$40bil by 2025. However, it will move to the next step to achieve artificial general intelligence (AGI) at which point, computers will outsmart humans and invent ever smarter machines. AGI is an algorithm that will not have to be taught a specific skill, like a new language. It’s acquired through trial and error – just like how a child learns.
For now, I am sure AI will lead to significant, but incremental changes across most areas of our lives – education, healthcare, transport, energy, entertainment and security. AI systems can help oncologists identify cancerous tumours; education robots already teach kids how to learn; predictive policing has massively eased congestion and transformed airport security; and they help to connect more and more devices to the internet, using Wi-Fi for speed. Also, AI applications have enabled much wealth to be created. Looking ahead, I see the relentless march of tech disruption challenging many existing businesses with further digitalisation and AI automation, including: (i) online travel booking platforms (under-threat: high-street travel agents); (ii) 3-D printers (threatening small manufacturers); (iii) driverless cars (with less collisions, thus threatening motor insurers); (iv) Robo-adviser websites (bringing financial advisers under threat); and (v) electric vehicles (because they involve low maintenance, car repair garages get threatened). But AI do leave dark spots: it impacts jobs, inequality, ethics, privacy and democratic transparency.
What’s pressing is an assessment of AI’s social impacts, especially on our way of life. The anxiety goes way back: panics about technological unemployment in the 1960s (following the first entry of computers and robots) and in the 1970s (when PCs first landed). Each time, workers’ jobs get threatened with automation.
To be fair, each time, it turned out that technology created more jobs than it destroyed. Replacements of bank-tellers with ATMs created new jobs in sales and customer service. While it’s true job losses in the short-run are more than offset in the long-term, experience over the past 30 years showed that the transition can be traumatic and do lead to massive discontent. Income inequality is already growing. Indeed, it has taken too long a time for real benefits to be translated into higher wages – even so, only for the higher skilled.
The real challenge: how to help displaced workers acquire new skills, and how to prepare future generations for a workplace stuffed full of AI. We all know AI does not simply change the skills needed from workers. It requires them to acquire new skills to complement it. This surely means education and training have to be better and flexible enough to teach new skills quickly and effectively. AI may itself help this process. In the end, what matters more are social & character skills. When jobs are perishable and people live longer, the situation calls for social traits that are beyond machines, giving humans an edge.
So, welfare safety-nets have to change. Denmark’s “flexicurity” systems can offer useful lessons – it lets firms hire and fire at will, while supporting displaced workers as they are being retrained and looking for new jobs; the welfare benefits follow the workers and not tied to employers. The message: welfare-support safety-nets need to be made more flexible, and retraining and continuing education schemes modernised. Populist support for Trump and Brexit should teach us that the livelihoods and social status of those disrupted by technology matter. It was true in the era of the industrial revolution; it remains true in today’s era of AI.
Picasso is known to have said that computers are useless: they can only give you answers. Fair enough in the context of early 20th century when computers were essentially souped-up calculating machines. Today, rapid technological advances in AI, robotics, nanotechnology and biotechnology have lifted society scrambling to understand their full ethical dimensions. Who is to provide the answers? How to ensure that AI is used for beneficent, rather than unethical, purposes? or put differently: there is a need to make sure that AI remains an extension of the human will. Self-driving cars and smart drugs, for example, raise ethical dilemmas. Who’s responsible for ensuring the use of AI is not abused? National governments rarely have the political bandwidth to consider such abstract challenges. Can’t leave it to universities, think-tanks or private “tech” institutions either. They often hide behind a narrow interpretation of the law. Consider the lessons of abuses at the time of the global financial crisis (2007-2008), when the pursuit of self-interest resulted in behaviour that not everything that is legal is ethical.
What then, are we to do
I consider every technique human beings have invented as technology. No doubt, the arrival of robots and AI will transform labour markets. Just a matter of time. We all need to bear in mind that technologies are mere tools. They offer opportunities and can inflict pain. What we make of them is, I am afraid, up to us. There is no other way. It’s a grave responsibility.
Tan Sri Lin See-Yan is the author of The Global Economy in Turbulent Times (Wiley, 2015). Feedback is most welcome; email: firstname.lastname@example.org.