Offshoring is dead, long live Watson

Many of us followed IBM's most recent stunt. Watson, a "question answering" supercomputer faced two of the most successful contestants of ‘Jeopardy!’, the popular TV game show in which contestants are presented with several riddled sentences that are actually answers to questions with which they should respond. Human contestants were defeated by Watson. The event was closely followed by several media, including The New York Times, which published a great article.

The fact that a computer can beat a human in sorting through stored information, and activating a buzzer once the right answer has been found shouldn't be noteworthy nowadays. What makes this particular situation remarkable is that Watson has the "intelligence" to sort through riddled statements – or as the NYT calls them “convoluted and often opaque” – like the following: “Yes, I should’ve worn long pants to ski, but my purple legs aren’t symptoms of this cold-induced tissue damage” in a category called “What me worry?”, and accurately match it with its right answer - or question in pure Jeopardy! style: “What is frost-bite?”.


Some time ago, Gartner Analyst Frances Karamouzis shared with us some revealing information about the actual reasons of job loss in car manufacturing. These were the results of a study carried out by Gen3 Partners, and it showed that by 1979 General Motors employed 91,000 people to produce a million vehicles. By 2004 the actual number of GM employees was 24,000 to produce the same million vehicles. A whopping 67,000 jobs lost in only 25 years.

Digging deeper, the report detailed what happened with those 67,000 jobs. Nineteen thousand had been outsourced to on-shore providers like Johnsons Controls, Arvin Meritor or Lier Siegler; 8,000 jobs had been outsourced offshore. These add to 27,000 jobs. What happened with the remaining 40,000 lost jobs? – They were displaced by productivity gains, resulting from better processes, but largely from automation.

The future of Watson, according to IBM, is adding voice recognition capabilities, while also working with different universities and companies, to make the computer perform all sorts of memory-based and script-following jobs, such as physician’s assistants, technical support and buying decision assistants.

Every person that has ever had to say her account number to a telephone banking system, or has tried to dictate his frequent flyer number to the airline’s automated customer service, knows that these voice-recognition software programs are still far form perfect. But it will get there sooner rather than later.

In its 2003 report Who Wins in Offshoring, McKinsey stated that “reduced costs are by far the greatest source of value creation for the US economy”. The GM case shows that the biggest chunk of cost reduction comes from automation, not offshoring.

Offshoring is here to stay, and will always be a part of the cost-savings equation, yet I am certain that Watson will be performing most of those low-value-added / easily offshoreable jobs.