AN IBM SUPERCOMPUTER has whipped two Jeopardy game show champions in a face-off.
Biggish Blue's data analytics and natural language processing computer called Watson has beaten Jeopardy champions Ken Jennings and Brad Rutter in trial demonstration. The actual man versus machine competition will have three rounds that will air on television in February.
When we say whipped, it wasn't an unequivocal thrashing. The two blokes are known as the best Jeopardy players in the US and managed to get many answers right, but not enough to win. The computer racked up $4,400 to win and Ken fared well with $3,400, but Brad came third with just $1,200.
Most unnerving was hearing Watson choose the "chicks dig me" category and doing a good impression of Hal from 2001. The audience laughed in the background but we felt a shiver go down our spine as we realised that the Singularity, if not end of the world, is nigh. Would the last human being please turn out the lights before we hand over power to our new computational overlords?
"We are at a very special moment in time," Dr John Kelly, IBM SVP and the director of IBM Research told Wired.
"We are at a moment where computers and computer technology now have approached humans. We have created a computer system that has the ability to understand natural human language, which is a very difficult thing for computers to do."
The Watson supercomputer is named after IBM's founder, Thomas J. Watson, not Sherlock Holmes' roommate Dr Watson, and it has taken 25 IBM boffins four years to program. So far Watson has had 200 million pages of information entered into its database and it isn't networked so it's not quickly referencing Wikipedia online when it's answering Jeopardy questions.
Watson consists of ten racks of IBM Power 750 Linux servers with 15TB of memory. It has 2,880 CPU cores to process a question, search for the answer and produce it in less than three seconds. µ
Uses 20 percent less power than traditional systems
It's becoming more prevalent in car research and development
Sign up for INQbot – a weekly roundup of the best from the INQ