Post by Amritanshu Patnaik
A highlight from earlier this year was AlphaGo, Google’s artificial intelligence (henceforth AI) program developed in collaboration with DeepMind to play the ancient Chinese board game of Go.
Go, as defined by an enthusiast, could be thought of as “super ultra mega chess”, a game so famously complex that the possible number of moves at each step equal all possible atoms in parallel universes i.e. about 10 to the power 170.Unlike chess, there are no dominant heuristics (like the rook, knight, or queen). Absence of such hierarchy makes Go both wonderfully simple and yet Byzantine. Dominance would depend on pattern recognition and subjective judgement. The usual desktop trained for tasks involving routine simulation wouldn’t know what to do in face of such confounding complexity. Humans, on the other hand, rely on observation, improvisation and practice. So does AlphaGo, based on what is called “deep neural networks”-analyzing several professional games and incorporating best practices. Lee Sedol, a professional South Korean Go champion was challenged by team AlphaGo for a best of five match. AI skeptics believed it would take at least 10 years from now for an AI platform to beat a world champ at Go, but in a surprise turn of events, Sedol lost in 4 games out of 5, marking a silent but significant step in the science of automation and artificial intelligence. This time, the audience comprised of more than computer geeks.
The AlphaGo victory demonstrates machines are now, more than ever capable of replacing human cognitive power in addition to physical power. AI is making giant strides in automating human labour. Examples are abound, from Tesla’s self-driving cars, Amazon’s headline grabbing drone delivery system, to the now ubiquitous Apple Siri. Computers have been taking over such tasks as clerical work and production jobs in manufacturing which tend to be of the blue-collar variety, but more complex jobs are not far beyond machine capability once deep neural networks are well-incorporated. Deep learning ability in machines means they learn from each pitfall and simulate the same over millions of scenarios. In this sense, they exhibit both human ingenuity and brute force mechanization.
VIEW FROM THE TRENCHES
Economists have long pondered over the possible impacts of automation on general employment.
“We are being afflicted with a new disease of which some readers may not yet have heard the name, but of which they will hear a great deal in the years to come – namely, technological unemployment. This means unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour”-
these surprisingly contemporary words are those of John Maynard Keynes, written back in 1930.
Modern day debates between economists on this issue seem to be coming from two camps- the optimists, who believe automation will not permanently alter overall employment growth, and the so called “alarmists”, who essentially believe that robots are coming for your jobs.
The latter camp comprises of economists like Erik Brynjolfsson and Andrew McAfee from the MIT Sloan School of Management. They point to the sluggish employment growth of the past 10-15 as a symptom of increasing automation. In a research paper, they show that 2000 onwards, while industrial productivity aided by technological innovation continues to rise, employment shows an abrupt downward trend. This decoupling between productivity and employment has been termed by them as the great paradox of our age- productivity and innovation remain at record levels, but median income and employment are falling. The underlying explanation is that rapid technological advancements are destroying jobs faster than creating them, contributing to stagnation of blue-collar incomes as well as overall inequality. Economists in the past have regarded productivity as the most important metric in measuring progress. That seems to be changing now. Although technological advancement is creating wealth and productivity, it can’t be thought of as a wave that lifts all boats, and that losers may outweigh the winners may not be a far fetched idea after all.
The optimists comprise of the likes of Harvard economist Lawrence Katz and his MIT counterpart David Autor. According to Autor, the fuzzy data (and circumstances) make it difficult to single out technological progress as the main source of job destruction- overall macroeconomic outlook has been generally bleak pointing to myriad factors. The delegation of middle-pay job to machines has been accompanied by a proliferation of higher-paying jobs requiring a creative skill-set. They agree that such polarization has significant implications on the inequality metric creating a gulf between the higher paying and lower paying jobs, but this is very different from saying that overall job creation has slowed down. Autor and Katz make a distinction between overall employment rate and the changing nature of jobs. A short recap of economic history would help in this regard. Pre-industrial revolution age jobs were replaced by manufacturing, development of steam power and textile mills, eventually leading to the factory system and other advances in England during the early 19th century. The process was called “creative destruction” by the 20th century economist Joseph Schumpeter, and led to a painful restructuring of the English working class. Creative destruction is perhaps best summed in the master’s own immortal words, “industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one“. The net result as seen today is resoundingly positive, aided by entrepreneurial vigour. By the same token, Schumpeterian perspective dictates transition in the tech-age may take some time but job numbers will eventually rebound, albeit in a slightly different fashion.
Detractors point to the questionable aptness of the industrial revolution parable- the pace of technological progress back then was nowhere near the breakneck speed it has assumed now. Autor addresses this concern in a recent paper which talks of an early thought experiment conducted by the 1960’s philosopher Michael Polanyi. He proposes that humans often do remarkably complex cognitive tasks like identifying the face of a person in a series of photographs as they age, without being able to explain how exactly they do it. Hence, if there exist such tasks that humans cannot explain, a computer cannot be programmed to replicate the same, at least not in the near future. A centre-piece in the alarmist view is the speed of technological innovation, but the above counter-example shows there may still be a long way to go.
FREE LUNCH, ANYONE?
What remains certain, to a point, is the hollowing out of middle-income earners that automation brings. Keynes’ prophecy of a machine aided world struggling with involuntary unemployment presents a grim picture of the future. The prospect has been subject to a variety of social investigations. One possible solution, though not new, seems to be universal basic income- a form of social security that guarantees citizens a periodic, unconditional sum of money. It’s an idea being toyed with by major European nations (Netherlands, Norway, and a recent failed referendum in Switzerland) and was temporarily implemented in the United States during Richard Nixon’s time. The left has been a perennial supporter of basic income owing to its re-distributive impact on income inequality, viewing it as the citizen’s return on capital owned by society. The right favours it as an efficient means to cut through the red-tape and corruption (not to mention, ideology) prevalent in welfare states. But the pressing need for an economic system that can incorporate such technological change as we see today contextualizes basic income above the binaries of left and right.
If we extrapolate the case to a highly automated society, say, somewhere in the not-so-distant future, that allows universal basic income, there would be some ‘able-bodied’ citizens who won’t be employed at all and yet receive some basic compensation. As disconcerting as the case may sound, there is a cultural barrier which envisages all individuals as working for their hard earned bread. The commitment to a work ethic and its moral high ground may be one reason for the reluctance toward collective contemplation of such a future. As stated earlier, technology is ‘spiky’, it doesn’t enrich everyone at the same time. In fact, great technological breakthroughs often translate to smaller resources creating bigger output. To elucidate, Google’s market value of $377 billion is supported by just 53,000 global employees, in contrast to General Motors’ market value of $60 billion on the backs of 216,000 employees. Such trends demand a radical rethink of our simplistic concepts of work and leisure.
A curious venture in this direction comes from the very people who bet their money on the bots. Y Combinator, one of the high priests of Silicon Valley seed funding scene came up with a pilot Basic Income project. Under the program, some 100 California families will receive $1500 per month for about a year. The goal of this experiment is to observe how recipients’ lives change when they don’t have to worry about the abject poverty that unemployment brings. Data and research methodology will be provided to researchers after the duration of this trial. In that sense, the whole setup is open ended. Aside from the utopian zest(!) and buzz that Silicon Valley ideas often gather, Y Combinator’s program is extremely relevant in engaging with the real as well as psychological ramifications of employment, or the lack thereof. Indeed, it is not surprising that companies heavily invested in automation foresee its implications on the very consumer they seek to enthral.
But this venture is curious not only in its uncanny and unsure benevolence, but also in the ideological common ground that it breaks. Let’s keep in mind that Silicon Valley is the true golden boy of capitalism, if ever there was one. Now for a moment consider Marx’s ideal of the communist society from The German Ideology.
“In communist society, where nobody has one exclusive sphere of activity but each can become accomplished in any branch he wishes, society regulates the general production and thus makes it possible for me to do one thing today and another tomorrow, to hunt in the morning, fish in the afternoon, rear cattle in the evening, criticise after dinner, just as I have a mind, without ever becoming hunter, fisherman, herdsman or critic.”
While not ditto, the basic income project does seem like a poor cousin of the above idea. An open ended setup does not guarantee positive outcomes, but one can’t overlook the pleasant irony that this situation presents. The Project thus succeeds in overcoming the glass ceiling of left-right rhetoric.
A more proliferated discourse on the implications of automation and the potential of basic income, or something in similar spirit, would be a desirable baby step in our race against the bots.
(Amritanshu is MA(P) student of Economics at DSE and is Junior Editor, Eostre)