Science and technology
Babbage
Artificial intelligence
The Difference Engine: The answering machine
Feb 18th 2011, 10:07 by N.V. | LOS ANGELES
IT WAS not quite a foregone conclusion, but all the smart money was on the machine. Since the first rehearsal over a year ago, it had become apparent that Watson—a supercomputer built by IBM to decode tricky questions posed in English and answer them correctly within seconds—would trounce the smartest of human challengers. And so it did earlier this week, following a three-day contest against the two most successful human champions of all time on “Jeopardy!”, a popular quiz game aired on American television. By the end of the contest, Watson had accumulated over $77,000 in winnings, compared with $24,000 and $21,600 for the two human champions. IBM donated the $1m in special prize money to charity, while the two human contestants gave half their runner-up awards away.
IBM has a long tradition of setting “grand challenges” for itself—as a way of driving internal research and innovation as well as demonstrating its technical smarts to the outside world. A previous challenge was the chess match staged in 1997 between IBM’s Deep Blue supercomputer and the then world champion, Garry Kasparov. As shocking as it seemed at the time, a computer capable of beating the best chess-player in the world proved only that the machine had enough computational horsepower to perform the rapid logical analysis needed to cope with the combinatorial explosion of moves and counter-moves. In no way did it demonstrate that Deep Blue was doing something even vaguely intelligent.
Even so, defeating a grandmaster at chess was child’s play compared with challenging a quiz show famous for offering clues laden with ambiguity, irony, wit and double meaning as well as riddles and puns—things that humans find tricky enough to fathom, let alone answer. Getting a mere number-cruncher to do so had long been thought impossible. The ability to parse the nested structure of language to extract context and meaning, and then use such concepts to create other linguistic structures, is what human intelligence is supposed to be all about.
Four years in the making, Watson is the brainchild of David Ferrucci, head of the DeepQA project at IBM’s research centre in Yorktown Heights, New York. Dr Ferrucci and his team have been using search, semantics and natural-language processing technologies to improve the way computers handle questions and answers in plain English. That is easier said than done. In parsing a question, a computer has to decide what is the verb, the subject, the object, the preposition as well as the object of the preposition. It must disambiguate words with multiple meanings, by taking into account any context it can recognise. When people talk among themselves, they bring so much contextual awareness to the conversation that answers become obvious. “The computer struggles with that,” says Dr Ferrucci.
Another problem for the computer is copying the facility the human brain has to use experience-based short-cuts (heuristics) to perform tasks. Computers have to do this using lengthy step-by-step procedures (algorithms). According to Dr Ferrucci, it would take two hours for one of the fastest processors to answer a simple natural-language question. To stand any chance of winning, contestants on “Jeopardy!” have to hit the buzzer with a correct answer within three seconds. For that reason, Watson was endowed with no fewer than 2,880 Power750 chips spread over 90 servers. Flat out, the machine can perform 80 trillion calculations a second. For comparison’s sake, a modern PC can manage around 100 billion calculations a second.
For the contest, Watson had to rely entirely on its own resources. That meant no searching the internet for answers or asking humans for help. Instead, it used more than 100 different algorithms to parse the natural-language questions and interrogate the 15 trillion bytes of trivia stored in its memory banks—equivalent to 200m pages of text. In most cases, Watson could dredge up answers quicker than either of its two human rivals. When it was not sure of the answer, the computer simply shut up rather than risk losing the bet. That way, it avoided impulsive behaviour that cost its opponents points.
Your correspondent finds it rather encouraging that a machine has beaten the best in the business. After all, getting a computer to converse with humans in their own language has been an elusive goal of artificial intelligence for decades. Making it happen says more about human achievement than anything spooky about machine dominance. And should a machine manage the feat without the human participants in the conversation realising they are not talking to another person, then the machine would pass the famous test for artificial intelligence devised in 1950 by Alan Turing, a British mathematician famous for cracking the Enigma and Lorenz ciphers during the second world war.
It is only a matter of time before a computer passes the Turing Test. It will not be Watson, but one of its successors doubtless will. Ray Kurzweil, a serial innovator, engineer and prognosticator, believes it will happen by 2029. He notes that it was only five years after the massive and hugely expensive Deep Blue beat Mr Kasparov in 1997 that Deep Fritz was able to achieve the same level of performance by combining the power of just eight personal computers. In part, that was because of the inexorable effects of Moore’s Law halving the price/performance of computing every 18 months. It was also due to the vast improvements in pattern-recognition software used to make the crucial tree-pruning decisions that determine successful moves and countermoves in chess.
Now that the price/performance of computers has accelerated to a halving every 12 months, Mr Kurzweil expects a single server to do the job of Watson’s 90 servers within seven years—and by a PC within a decade. If cloud computing fulfils its promise, then bursts of Watson-like performance could be available to the public at nominal cost even sooner. Mr Kurzweil believes that once computers master human levels of pattern recognition and language understanding, they will leave mankind way behind. By then, they will have combined the human skills of language and pattern recognition with their own unique ability to master vast corpora of knowledge.
Will that mean game over for humans—with robots keeping people around merely as pets? “Absolutely not,” says Oren Etzioni, director of the Turing Centre at the University of Washington in Seattle. But it does mean, he notes, that computers will be able to achieve vastly more than they can today. For a start, super-smart machines capable of answering questions in English (or any other natural language) will change search engines out of all recognition. No longer will Google and Bing bombard users with hundreds or even thousands of dumb links to dubious sources. Instead, people will get the unique and meaningful answers they are seeking.
Long term, Watson’s progeny could help people sift through the thousands of possibilities they confront in their public and private lives, and come up with handfuls of appropriate recommendations—whether in medical diagnosis and treatment, legal precedents, investment opportunities, design configurations or whatever. Your correspondent looks forward to the day when such a superior intellect can do his thinking for him.
About Babbage
In this blog, our correspondents report on the intersections between science, technology, culture and policy.
Follow Babbage on Twitter »
Advertisement
Economist blogs
- Americas view
- Asia view
- Babbage
- Bagehot's notebook
- Banyan
- Baobab
- Blighty
- Buttonwood's notebook
- Charlemagne's notebook
- Clausewitz
- Daily chart
- Democracy in America
- Eastern approaches
- Free exchange
- Global Leadership
- Gulliver
- Johnson
- Leviathan
- Lexington's notebook
- Multimedia
- Newsbook
- Prospero
- Schumpeter's notebook
- The World in 2011: Cassandra
Most commented
- Charlemagne: No time for doubters
- Banyan: Taiwan's commonsense consensus
- China's pre-emptive crackdown: The people doth not protest
- Turkey and Europe: Mr Erdoğan goes to Germany
- China's foreign policy: Setting sail for Libya
- Language and opinion: Framing climate change
- Precautions in Xinjiang: Catching a whiff of jasmine in Kashgar
- Qaddafi and his ilk: Blood and oil
- Turkey's election: A Muslim democracy in action
- Venezuela's economy: Oil leak
Over the past five days
Most recommended
- Qaddafi and his ilk: Blood and oil
- China's pre-emptive crackdown: The people doth not protest
- KAL's cartoon
- Banyan: Taiwan's commonsense consensus
- A special report on feeding the world: The 9 billion-people question
- The Arab uprisings: Endgame in Tripoli
- Energy prices: Tax away vulnerability
- China's foreign policy: Setting sail for Libya
- Labour law in America: Showdown in Madison
- Venezuela's economy: Oil leak
Over the past seven days
Advertisement
Subscribe to The Economist's free e-mail newsletters and alerts.
Subscribe to The Economist's latest article postings on Twitter
See a selection of The Economist's articles, events, topical videos and debates on Facebook.
Advertisement
Readers' comments
The Economist welcomes your views. Please stay on topic and be respectful of other readers. Review our comments policy.
Sort:
And even better, we can fire a bunch more of people and replace them with machines :)
Yea for development!
So the Turing Test will be passed in 2029?
At current rates of innovation and development expect AI to be perfected by 2039 and ubiquitous in lives of humankind by maybe 2049?
Fantastic!
I'll either be crapping in my Depends with dementia or dead.
Either way I'm relieved that I'll be missing it.
So they dumbed down a computer! Big f'ing deal!
But seriously, search engines have nothing to fear from Watson. You forget the questions in Jeopardy are so precise there is only one answer. In fact, the contestant is told the answer and he has to determine the question. Google bets its existence on you not even knowing the question to ask. And how can you when you're trying to determine what someone on the other side of the world might have said about something six hours ago or six years ago.
In other ways, Watson is even more powerful. Jeopardy is a timed contest. In real life, given sufficient time and resources, Watson would be better than any human in any subject. And if it was given a hint by humans whether is was getting warmer, it would eventually determine the right answer with enough time and assistance. In that way, it's the best librarian ever but like a librarian, it doesn't have to be a physicist to find the correct physics answer; it just has to eliminate all the wrong ones.
It's power resides in its ability to determine equivalent meaning from two different sentences or phrases because in the end, language is finite.
I think that these guys actually new the answer most of the time but they could not beat the computer on buzzing in. That is where the battle is won or loss. I was sad to see the huamans loose or should I saw terminated :-(.
should I say*
A man stares into his search avatar in the near future and asks:
"What is the best form of government?"
The avatar, without perceptible pause, answers:
"A computational dictatorship"
Game over.
The thing didn't know that Toronto was in Canada. It flubbed the question... I wouldn't trust it for much.
Now, can we please use this to get some better online translators? It is faster for me to hand-translate something in Japanese every time than to pop it into either GoogleTranslate or Babelfish. The vodka is good, but the meat is rotten indeed.
Ah, but can Watson tell us how to reverse entropy?
I'm really scared of this. This is going to make translators and call centers in India obsolete. It's not bad per se. I just worry that soon reading, speaking and writing in a foreign language will no longer be a commercially valuable skill.
> "What is the best form of government?"
The avatar, without perceptible pause, answers:
"A computational dictatorship"
Unfortunately, that computer is right, at least the computer-dictator is not swayed by stashing money in a Swiss bank or by pretty girls. You have only to look at Congress to despair of human democracy.
For a while, it has been predicted that artificial intelligence will cross over humans around 2040, and see Time magazine of February 21.
It didn't know Toronto is in Canada because it was programmed by Americans!
As bad as it was, Watson is certainly better versed in world geography than Sara Palin or Janet Napolitano.
If this natural language query software is ready for prime time, why not put it to work on the most obvious useful application -- an internet search engine? Playing Jeopardy is great publicity, but search engines are where several much-ballyhooed natural language query parsers have foundered and sunk. Remember "Jeeves"? The other reason to put Watson to work as a search engine is that the software would then have the opportunity to learn a great deal about how people really ask questions. Watson will probably be surprised by how many people ask "How can I get laid tonight?", and will have quite a learning curve trying to figure out how to answer that question in the most successful way.
It reminds me of the races b/w horses and locomotives in 19th century. They might have been neck and neck at the time, but it's clear in the long run which one is going to win.
Toronto isn't a part of the good Old US of A? Pishaw!
As many a Science Fiction author has already noted and reviewed this type of event: computers, even super-smart, beat your pants off at Trivial Pursuit computers, will always have their limitations. An excellent object for discussion would be Robert A. Heinlein's magnificent portrayal of the supercomputer as a tag-a-long government subversive in "The Moon Is a Harsh Mistress"
For those who don't spend their days wading through copious amounts of futuristic fiction; the revolting prisoners of the moon use the brain of the super computer as a tool to help them foment rebellion and declare independence from the Earth. The Moon’s computer did all of the planning and statistical analysis of the rebellion for them. The only thing the meatbags had to do in their rebellion was to carry the guns.
One can only guess at the uses that a fanatical underclass could use with this type of informational tool in their possession. If only there were ways to prevent information from getting to the proletariat, the ruling class would be safe for future generations.
It is instructive to note that indeed a computer can perform calculations much faster an a human. However the human brain which is about 2 pounds can do many things that that much heavier machine cannot do. Also a human only uses no more than 10 percent of total brain capacity.
Rabbi Dr katherine Hans Von Rotes Schild Zilter, Physics, theologian
"One can only guess at the uses that a fanatical underclass could use with this type of informational tool in their possession. If only there were ways to prevent information from getting to the proletariat, the ruling class would be safe for future generations."
I assume you're being sarcastic there, but the revolutionaries will always be the low-tech "meatbags", and they will always have the advantage in a protracted insurrection or guerrilla war. Just to pick an easy example, who is going to go up in the mountains in the dead of winter to replace the power tower that fell over because some meatbag with a wrench unbolted it, thus cutting the electricity to the city where the computer lives? An intelligent robot no doubt. Who will build it? More intelligent robots? Okay, will it be able to fix its own flat tire? Get itself unstuck from a snow-filled ditch? Figure out a way to make do with what tools and supplies it has on hand when the it's missing what it should have had? Will it be watching for that kid hiding behind the bushes who's about to run up and jam a piece of pipe through its wheel?
If war is the ultimate contest of both strength and intelligence, most guerrillas would love to be up against "intelligent" machines. Also, every ruling class must know that is is dependent upon the proletariat. Machines, the modern slaves, have always been attractive to such people, but all that does is push the dependency down to the people who build and maintain the machines. Any way you figure it, the world will always need human workers.
I was involved in research into artificial intelligence in the early days. Consequently, I found this article fascinating as it details progress that was over the horizon at that time. It shows that some of the dreams of the pioneers in this field have been realized and the bright sparks fostering developments have moved on to more advanced objectives.
There is talk about these machines doing most of the advanced thinking for society. There is, however, no mention in the article of certain practical limitations that will ensure that the developing dreams cannot be realized. These machines cannot replace humans for a fundamental reason. People reproduce naturally. Input of air, food and water enables their operation and development and these are readily available. These inputs provide the materials and energy necessary for this purpose. The food provides the wide range of chemical compounds and trace elements that drives metabolism. Limitations on food and water supply could well limit future global population growth but that does not affect this natural reproductive principle. This process that has operated for eons and there is no reason why it should stop.
On the other hand, the machines that provide artificial intelligence are not naturally reproductive. They are made of minerals that have to be extracted from ore bodies and processed. This can be done in the future by using the flow of energy from insolation but the current driving force is energy from irreplaceable crustal materials such as oil, gas and coal. The state of these natural resources is the result of constructive natural forces operating slowly over eons. The machines are built using some of these irreplaceable natural resources. These machines age due to the impact of natural forces. They can sometimes be replaced by other machines having the same fundamental weakness. The production of thinking machines is an unsustainable process. I wonder whether these thinking machines will inform the human species of this fundamental truth before their demise!
I am still wondering whether we could one day invent a computer with human feelings?