Watson, IBM's latest intelligent machine, crushes human opposition in a practice round of Jeopardy. Is this the end of human dominance of the planet?
We reported at the start of December that IBM had a master plan to bring AI into the limelight by getting a shot at winning Jeopardy with Watson - a computer. The details of the probable approach being used was described in the earlier news item:
What we can now add to the mix is the fact that Watson has been subjected to a studio trial run and the results are impressive to say the least. The human opponents, Jeopardy champions Ken Jennings and Brad Rutter were simply beaten by the machine's speed and lack of emotion. Watson won convincingly in the first practice round, earning $4,400, while Ken Jennings trailed with $3,400. Brad Rutter, who has earned a cumulative $3.3 million on the show, came in last.
Its voice sounds like HAL and there is a certain creepy sort of chill about the whole thing - watch the video to see if you agree.
Watson tends to get things wrong when the question involves humor or other subtle human qualities - it is far from perfect - as yet. Despite lapses he did, however, win convincingly.
The most important points are that this isn't just game AI as demonstrated by Deep Blue the winner in the famous chess match with Kasparov (1997) but language understanding. This has more potential applications than playing chess and so it is unlikely that IBM will give details of how Watson works.
Watson will appear on the real show on the 14th, 15th and 16th of February.
We are all more or less committed to using high-level languages, but there is always a background worry that they might not be fast enough for some tasks. An interesting set of benchmarks shows how to [ ... ]