IBM Watson (Jeopardy at Carnegie Mellon)
GIST OF IT: IBM‘s Computer, Watson, now have enough language skills to beat humans in a game of which the questions involve “subtle meanings, irony, riddles,” and other linguistic complications.
WHY SCYNET CARES: Read carefully. Watson was built for a mere $ 3 000 000.00. And it was tremendously successful in “being human” in the Jeopardy Show. This is happening now, here, in real-time – not on some futuristic horror movie. And think about it. How will this artificial humanity be utilized and for whom will it work? It will be in the service of various corporations, designed to outsmart you in situations where you have to make choices. And the corporations have a different aim than you do. Think about it. Imagine a machine interviewing your son, and by using vast intelligence and all the records from your son’s life, it can succeed in getting your son, for example, to join the Marines; even though you know he would never have wanted that for himself? How do you protect yourself from superior intelligence aimed at getting you to do things that might not be good for you?
C-3PO is a protocol droid designed to serve humans, and boasts that he is fluent “in over six million forms of communication.”
This year’s FOSE conference had an unexpected theme: artificial intelligence.
The subject came up in Apple co-founder Steve Wozniak’s keynote speech on the first day of the conference. It took center stage, though, on day 2 of the conference, when IBM Research Vice President of Software David McQueeney presented the Beyond Jeopardy! – The Implications of IBM Watson keynote.McQueeney heads the division responsible for Watson, the $3 million natural language computer IBM made famous on the television quiz show Jeopardy!, where it bested two humans in a two-day trivia contest. At first, McQueeney seemed to be an odd choice of speaker for a gathering of government acquisitions professionals and the people who want to sell them goods and services – until he started to draw connections for the crowd.
IBM, McQueeney said, had left little doubt that computers, with sufficient processing power, could best humans at quantitative tasks – even those that involve some forethought and creativity. The company’s previous artificial intelligence experiment, known as Deep Blue, had defeated world champion Gary Kasparov in a series of chess matches in 1997. But a challenge remained: language. Computers, in the past, hadn’t been able to understand language the way humans use it. McQueeney explained, “When humans communicate with computers, they have to use a discrete, exact programming language.” But when humans communicate with one another, they are more creative and fluid. “Unstructured sentences are very hard for machines to process,” he said. Human speech involves shades of meaning and intent, even puns and word games, that flummox traditional computers. “It took us much more scientific effort and computational work to tackle human language than to win chess games,” he said. Jeopardy, McQueeney said, was the ultimate challenge because the game’s questions involve “subtle meanings, irony, riddles,” and other linguistic complications. To succeed, the system couldn’t simply search through structured information (something that, McQueeney said, Watson did for just 15 percent of the questions asked in the contest). Instead, it had to parse meaning from sentences, much like a human does. “We consider it a long-standing challenge in artificial intelligence to emulate a slice of human behavior,” said McQueeney. Ultimately, the effort succeeded – Watson won the contest – but breakdowns in the company’s solutions were obvious. The computer struggled, McQueeney noted, with the shortest questions.
So, IBM pulled off quite a parlor trick with, by McQueeney’s estimate, $3 million worth of off-the-shelf parts used to build Watson. But what does the technology mean for government? McQueeney rattled off a series of possible applications for the system’s natural language abilities. For instance, he noted, Watson’s technology could be used to “support doctors’ differential diagnosis, using data in the form of cited medical papers and giving quick, real-time responses to questions.” The technology could be used, he said, to assist technical support and help desk services, improve knowledge management at large companies, and improve information sharing among national security workers. “Computers can now support human interaction in new ways,” he said.
IBM expects to see the first application of the technology in the healthcare field, McQueeney said. The company is currently working with several medical schools to develop knowledge systems that will assist doctors with rapid medical decisions. The company has announced hopes to develop a commercial offering by the end of 2012.
FOSE 2011: IBM’s Artificial Intelligence Coming – Learn More at GovWin.
- Artificial intelligence 7/7/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence 7/22/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence 7/21/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence Diagnoses Abuse : Discovery News (scynet.wordpress.com)
- Artificial Intelligence 7/16/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence (scynet.wordpress.com)
- Artificial intelligence 7/6/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence 7/9/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence 7/12/2011 (empressoftheglobaluniverse.wordpress.com)
- Video: IBM Watson – The Science Behind an Answer (insidehpc.com)
- Artificial Intelligence 7/27/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial Intelligence 7/26/2011 (empressoftheglobaluniverse.wordpress.com)
- Artificial intelligence 7/14/2011 (empressoftheglobaluniverse.wordpress.com)