Online edition of India's National Newspaper
Thursday, January 11, 2001

Front Page | National | Southern States | Other States | International | Opinion | Business | Sport | Science & Tech | Entertainment | Miscellaneous | Features | Classifieds | Employment | Index | Home

Science & Tech | Previous | Next

Artificial intelligence is not dead - just recycled!

Finally 32 years after Stanley Kubrick filmed the seminal `Space Odyssey' based on Arthur Clarke's prophetic vision of a world taken over by computers, we are in 2001. So why is no one worried? Anand Parthasarathy examines how classical Artificial Intelligence is having to swiftly reinvent itself in an increasingly commercial environment

MANY OF his writings foretold future technological developments with stunning accuracy. In a 1945 article in the magazine ``Wireless World'', he foresaw geostationary communication satellites - and by the mid 1960s, they were up and working. In similar unerring fashion he predicted mobile phones, men landing on the moon - and inhabited space stations. In the final stunning moments of ``2001: A Space Odyssey'', the film that Kubrick made in 1968 based on a short story by Arthur Clarke, the computer, HAL, takes over the space ship - and kills off the human crew.

We are three weeks into 2001 - and while it can be said that many of us are the willing slaves of our PCs, we do not log on nightly in fear of our lives. So is this the only major prediction implied by Arthur Clarke that is just a bit out of sync with the timeline? And if so why?

In the last days of 2000, Clarke now 83, and a long term resident of Colombo, Sri Lanka, maintained in an interview with the Associated Press that his vision of intelligent machines may not be all that far-out. ``When someone says that something is possible - he is almost certainly right. When he states that it is impossible, he is very probably wrong'' - is Clarke's dictum.

If machines which can match humans in intelligence are not yet in even a tentative state of achievement, the reasons could be the changing fortunes of - and the lack of clarity - of the Artificial Intelligence business. Remember the Japanese Fifth Generation Project, The Alvey Project in UK, the American thrust into AI led by the Bhishmacharya of AI, EA Feigenbaum - all part of the exciting 1980s? Where are they today? For that matter what about the natural language processing and voice recognition research spearheaded for decades at IIT Madras and TIFR? On a warm March morning 14 years ago in the lobby of what was then the Adyar Gate Hotel in Chennai, I was interviewing a bubbly roly poly of a man - Dr. Douglas Lenat - head of a consortium of US high tech companies called Microelectronics and Computer Technology Corporation (MCC). He had just emerged from the conference hall where he was a key speaker at an International Conference on Fifth Generation Computing. ``We aim to put the world's common sense on the computer!'' Lenat, a man not given to understatement, told me. In my writeup the following Sunday in ``The Hindu's'' ``Meet...'' column, I could not quite curtail my own enthusiasm at the Lenat vision. Intermittently, since then I have been monitoring the project on the MCC website - but nothing practical ever came of it. When I checked recently, Lenat was no more with MCC. Last week The New York Times was quoting Lenat, bashing the current breed of speech recognition softwares which required you to spell out punctuation. Today Lenat has moved on from trying to put all knowledge into one data base - and heads a Texas company called Cycorp.

Among the first people to realise that governments and financiers will not wait with breathless anticipation for decades to see the fruits of high sounding research like AI, were the scientists at MIT's AI Lab and at Carnegie Mellon's Robotics Lab - both in the US. Today they virtually set the agenda of what AI should be tackling. The mantra seems to be: take up tasks that will capture the paying public's imagination - and don't worry if people snigger: they are not the ones bankrolling the research.

The one's who are - include shrewd Japanese companies like Sony and Honda, who are laughing all the way to the back these days, raking in the yen that flow from millions of customers for their canny robots. The second improved version of Sony's robo-pup, ``Aibo'' was launched on January 2. The metallic puppy has 18 joints producing 250 movements. It crouches, plays ball, snuggles if fondled and sulks if slapped.Aibo which means ``partner'' in Japanese was an instant sensation with Japanese kids at first launch in June 1999 inspite of the stiff price ( the equivalent of Rs 60,000).Meanwhile car maker Honda, recently unveiled ``Asimo'', (Advanced Step in Innovative Mobility), a 120 cms tall human like robot. It can be a help in the house, opening doors, setting tables, loading the washing machine and cutting vegetables.

Not to be left out of the good thing, leading American hand held computer maker, Palm Inc, has last week licensed a company, Acroname, to sell a kit which will turn their Palm Pilots into the brains of six sided, three wheeled robot. The device developed at Carnegie Mellon is a quantum jump in the technology of robotics - for the first time a mass consumer computer can be modified to become the physical robot.``The last 10 years of the 20th century were dominated by the PC. In the first ten years of the 21st century, robots with independent movement will dominate'', says Sony's Vice President Toshitada Doi.

Dr. Rodney Brooks of MIT AI Labs is also Chairman of iRobot - a robotics company that makes robots for private missions to Mars. But for the butter on its bread, it also makes ``My Real Baby'', a $ 100, doll marketed by Hasbro, which smiles, cries, coos - and draws attention to a wet nappy.

So that's AI today - albeit less conspicuous. Remember the Paper Clip wizard in Microsoft's Word 2000? The fuzzy washing machine that selects wash cycles after feeling the clothes? It's AI at work.

And as for that old bogey about a machine overtaking Man - Rahul Sarpheshkar, an Indian researcher at MIT wrote in a recent paper in ``Nature'': ``We are not going to get close to the human brain in the next 50 or 100 years'' He admits that even the most sophisticated neural computer cannot do what a weeks-old baby does: walk across the room without tripping over obstacles. Get back into the pages of fiction, HAL, you still belong there.

Send this article to Friends by E-Mail


Section  : Science & Tech
Previous : Web site download manager
Next     : Towards third generation wireless technology

Front Page | National | Southern States | Other States | International | Opinion | Business | Sport | Science & Tech | Entertainment | Miscellaneous | Features | Classifieds | Employment | Index | Home

Copyrights © 2001 The Hindu

Republication or redissemination of the contents of this screen are expressly prohibited without the written consent of The Hindu