Watson, the supercomputer computer created by IBM that vanquished humankind on Jeopardy! a few years ago, is now going after the descendants of the nerds who created it.
International Business Machines Corp. announced last month that it had launched Watson Discovery Advisor, a cloud-based service that is capable of analyzing thousands of research articles in a matter of days or weeks. It would take an army of human researchers to perform such a feat.
Armies are expensive. They also complain a lot. (Watson, you're hired!) IBM said Johnson & Johnson is using Watson Discovery to plow through clinical trials in order to decrease the time it takes to evaluate the relative effectiveness of various drugs. Sanofi is testing Watson's ability to produce toxicology reports with an eye to cutting down on the number of drugs trials that are scrapped because a treatment proves harmful. IBM suggests future clients could include investment advisers, law firms, security agencies and chefs. (Yes, chefs.)
White-collar jobs were supposed to be safe from the machines. That no longer feels true. Academics Erik Brynjolfsson and Andrew McAfee thrust themselves into the public discourse with their argument that the current technological age could increase economic growth and depress jobs and and wages. Earlier this year, the Atlantic magazine lamented the "end of labour." Only 59 per cent of the American population had a job in July, compared with rates closer to 63 per cent in the years ahead of the recession. The decline surely is partly explained by the spread of automation.
This question has moved beyond the academic. The U.S. Federal Reserve is trying to understand the extent to which the smaller work-to-population ratio is a product of the Great Recession and/or fundamental changes in the structure of the labour market. The former scenario argues for demand-stoking stimulus measures. The latter suggests the U.S. lacks the capacity to grow as fast as it used to without stirring up rapid inflation.
Central bankers considered the extent to which they need to worry about the march of the machines 10 days ago at the Kansas City Fed's annual economic conference in Jackson Hole, Wyo. The message they heard from the labour market expert invited to address the subject: relax. "People overstate the degree of substitution between machinery and people and fail to recognize the complementarities," David Autor, an economics professor at the Massachusetts Institute of Technology, told me in an interview after his closed-door presentation at Jackson Hole.
Prof. Autor is somewhat mocking of the fretful tone of much of the more recent analysis of automation. The paper he wrote for the Jackson Hole conference notes that technological advances have been upending labour markets since the Industrial Revolution. Change is disruptive. Yet, "short-term employment losses sparked by rising productivity were eventually more than offset by subsequent employment gains – in some cases in the innovating sectors, in many cases elsewhere," Prof. Autor wrote.
In 1900, 41 per cent of the U.S. population worked in agriculture. Today the percentage is 2 per cent. A century ago, it would have been difficult to foresee that industries such as health care and finance would employ more people than farming. "Arguably, we stand at a similar moment today," Prof. Autor wrote.
To be sure, history isn't future. Serious economists, including Robert Gordon at Northwestern University and Tyler Cowen at George Mason University, argue that the wealth-generating leaps in productivity of the past have run their course. The slow-growth future predicted by Prof. Gordon and Prof. Cowen resonates because recent economic history is so dreary. They argue the newest innovations pale against the technological breakthroughs of the past: Gutenberg and his printing press changed the world; Zuckerberg and his Facebook mostly turned a handful of people into billionaires and multi-millionaires. Technology used to spread the wealth. Now it appears only to concentrate it.
Prof. Autor is skeptical that automation suddenly is shrinking the labour market rather than leading to its expansion. He argues there is a natural barrier to what machines can achieve because human intuition can't be programmed. Prof. Autor uses the example of an automobile assembly plant. A robot can expertly install a windshield in the controlled environment of the factory. Aftermarket replacement, however, still is the preserve of humans because it is too difficult, if not impossible, to program a robot to respond to the variety of jobs a garage owner could face on a given day. Prof. Autor sees a resurgence of "middle-skill" jobs that complement the machines, which will continue to lack basic skills such as literacy and common sense.
What does this mean for the Fed and other central banks? According to Prof. Autor, they should keep on doing what they are doing. Labour isn't over. We're just lacking a proper amount of demand for it.