Jul 17 2015
Artificial intelligence (AI) has given rise to machines that can match humans at chess, rival them in the stock market, and aid them in the clinic. But researchers' ultimate goal is to create a well-rounded, human-like computer program -- an aspiration that raises many questions and poses a wide range of challenges.
In this special issue of Science, researchers discuss recent and remarkable progress in AI, along with its potential implications for humankind. A Review article by David Parkes and Michael Wellman, which will be available online for free after the embargo lifts, describes the pursuit of machina economicus -- a perfectly rational (and therefore mythical) economic machine, completely free from the biases that influence human decisions. The authors suggest that new rules and incentive systems are needed to guide AIs, since more and more machines are becoming players in the global financial market. In a separate Review article, Michael Jordan and Tom Mitchell explain how "Big Data" and advanced computation have driven recent progress in machine learning -- a growing, technical field that lies at the intersection of computer science and statistics. They suggest that, by designing computers that improve automatically with experience, machine learning represents one of the most transformative technologies of the 21st century -- and that society should start considering ways to maximize its benefits now. A Review article by Julia Hirschberg and Christopher Manning highlights challenges to and advances in language processing technologies, such as speech recognition and machine translation, which have made talking to one's cell phone and translating phrases on the Web commonplace activities, especially for young people. Samuel Gershman and colleagues emphasize similarities among AI, cognitive science, and neuroscience in a final Review article, suggesting that a combination of the three fields could lead to an increased efficiency, known as computational rationality.
The special issue of Science also includes a Policy Forum by Eric Horvitz and Deirdre Mulligan that discusses some of the serious privacy concerns that arise with AI research -- as well as opportunities to advance equality and the public good. Two related news stories by John Bohannon further illustrate the impacts of AI on modern society: One examines the increasing popularity (and utility) of computerized psychotherapists -- programs that interact with patients and diagnose their emotional problems -- while the other provides a Q&A with Stuart Russell, an AI pioneer who now questions where the technology is leading us.
This news release is available in Japanese.