Jonathan Zittrain explored the light and dark sides of innovation with equal aplomb at the Belly Up in Aspen on Monday night. His unique position as a professor of both law and computer science at Harvard was much in evidence as he took us through the intricacies of crowdsourcing and the labyrinth of human motivation. The talk was a tour de force that covered many positive developments, but also uncovered some unsettling possibilities.
At the top of the innovation pyramid are big ticket challenges, such as the X Prizes, which offer large sums of money for the solution to really hard problems, such as sending a robot to the moon. At the bottom are microtasks for which people are paid pennies for exercising simple human skills, such as labeling images or reading handwriting. These are all seemingly harmless with an employment upside for many workers but, as with the Internet itself, crowdsourcing technology enables bad actors as well as good.
The downside ranges from simple scams to the abolition of privacy. Simple scams include paying Mechanical Turks to ‘like’ products they’ve never owned and give them favorable reviews. The more serious stuff involves encouraging citizens to monitor surveillance cameras, record events of interest to intelligence agencies or identify political demonstrators. The scope creep from neighborliness to patriotism to voyeurism to spying for a fee takes us progressively along a very uncomfortable spectrum of human behaviors and outcomes.
(In a fascinating aside, Zittrain noted that it should be possible to identify a single Iranian demonstrator from a photograph by crowdsourcing the comparison of the photograph with the ID cards of 72 million Iranians four photos at a time. He estimated that the total cost would be about $14,000 using Mechanical Turks, and that American schoolchildren could be given this task as a game, which they might even play for free.)
In spite of the dark side, this was a very amusing talk in which we were all invited to marvel at the ways in which folks will spend their time on the Internet. Whether tending virtual farms or playing games for worthless points, it’s clear that many people have hours to kill each day that can be harnessed for good or ill. There were several after-talk queries about how to encourage the upside and regulate the downside, some of which the speaker wisely side-stepped. One suspects that easy answers are nowhere to be found, but at least we can now ask the right questions.
Peter Jackson is chief scientist and vice president of Thomson Reuters, where he’s been since 1995. He has built a group of 40 research staff with expertise in the areas of document search, text and data mining, and machine learning. Jackson is also responsible for university collaboration with respect to joint research projects. His most recent book, Natural Language Processing for Online Applications, came out in a second edition in 2007. From 1992 to 1995, Jackson taught post-graduate classes in artificial intelligence and parallel computing at Clarkson University in New York and was a visiting professor at Singapore Polytechnic. In 1988, he moved to the US and became a principal scientist at McDonnell Douglas Research Laboratories. Before coming to the US, he taught in the Department of Artificial Intelligence at Edinburgh University from 1983 to 1988 and wrote the textbook Introduction to Expert Systems.