Labor Arbitrage, Automation and Customer Service
Monday, June 21, 2010 at 3:21AM The feature article in the NY Times Magazine yesterday told the story of IBM's AI team creating a credible Jeopardy contestant. Clearly the IBM team has made some progress since Deep Blue beat Garry Kasparov in May of 1997. The computer may not win, but IBM will win a great deal of attention during the event next fall. Probably both great technology and great marketing.
While reading the article, some roads converged in my technology imagination, mostly in the areas of labor arbitrage, automation, and customer service. The effects of these changes are going to be felt slowly over some time -- but they will be significant.
Labor Arbitrage
We are a decade into the Internet enabled off-shoring movement fueled mostly by low cost labor. Technology innovations only happen when the innovation is ten times better. Offshore labor does not have to be 1/10th the onshore cost, but it needs to be at about a third in order to work. If we are paying $9 per hour onshore for something that can be done for $3 per hour offshore -- the inefficiency of distance and the added cost of travel and/or transport can be overcome. If onshore and offshore labor rates converge, off-shoring will become less compelling. This convergence can happen by offshore labor rates rising as competition for workers and living standards are raised in offshore markets, or as onshore labor rates fall. Wait, how can onshore labor rates fall? Through automation.
Automation
Everywhere we look we see automation. Cars are still being built in this country because robots do most of the work. We see the combination of automation and self service every time we go to the bank machine or the grocery store. Google signs up customers without any salespeople -- which is automation displacing labor in yet another way. The IBM Watson project may seem too theoretical to start displacing humans, but as the NY Times piece points out, the first application may be in the call center. Giving the computer the job of answering customers complex questions. Just like on the manufacturing line, the bank machine, or the grocery store, the computer does not have to answer all of the questions, just a good percentage. When the human's job becomes handling the extreme exception and managing the machine -- the skills required and the associated pay are each increased significantly. At the end of this road lies a customer service capability for companies who have never operated in that mode.
Customer Service
Search for "Google Lack Customer Service" and you can read for days about how Google just does not do it. This is a cause for relief by some of Google's more customer centric competitors. When IBM delivers to Google an engineering driven answer to this deficiency it will be as big as any significant change in an ecosystem. Kill all of the wolves and the elk population goes through the roof. Google is not the only engineering driven company that will benefit. HTC and many of the other sophisticated OEMs, will be able to accelerate their evolution from manufacturer for others to full competitor. The ecosystem will never be the same.
Earlier this year I heard a presentation by Jaron Lanier where he gave the low cost labor countries like India, China, and the Philippines 20 years to get up the education ladder far enough to be safe from the flood caused by automation. Could be 20, I would guess 10.
China, Google, HTC, IBM, India, Philippines in Economics, Innovation
Reader Comments (2)
A computer playing Jeopardy would be awesome if it could compete against humans with super fast reflexes to press the button if they think, or do, know the answer.
"the skills required and the associated pay are each increased significantly" -- But only for the small minority that would be required to maintain that equipment, much of the workforce would end up displaced or pushed in other directions (but what would those be if everything were automated?)
That is just the thing Andrew, the computer comes up with its answer, tests its confidence, and makes a choice about pushing a button. Quite a step forward in AI.