A Peek at the Computing Future
Computing has come a long way in a short time, but it seems unlikely that we’ll reach science fiction writer Arthur C. Clarke’s vision of an all-knowing, voice-activated HAL computer system by the year 2001.
Nonetheless, computing five or 10 years from now could be pretty exciting, as revealed in a sneak preview recently given by IBM. The world’s largest computer maker showed application research that may find its way into future products.
One project, NewSelector, shows how far we can hope to go in software for sifting through documents.
NewSelector is designed to scan newspaper stories and determine which would be of interest to each subscriber using the program.
In the example shown by IBM, the research program was given the profiles of three readers, one an investment banker, another a building construction executive and the third a private investor concentrating on computer-related securities.
Next, NewSelector was fed a newspaper article about Brazil suspending its debt payments. The idea was for the software to read the article and understand its content well enough to determine whether the article would be of interest to any of its three subscribers.
To do so, NewSelector had to take apart the news story word by word and sentence by sentence in a process known as parsing. Then it had to match the meaning of those words and sentences to the interests of its subscribers.
The banker would have been able to find the article with present-day text retrieval software because it contained such words as bank, debt and finance.
But NewSelector determined that the construction executive also would want to know about it because suspension of debt payments could influence interest rates for construction in the United States. It also decided that the investor would not be interested because there was no link between the Brazilian debt and computer companies.
Written in the artificial intelligence programming language Prolog, the project is being conducted by IBM’s Los Angeles Scientific Center. Project director Kathleen Dahlgren’s book, “Naive Semantics for Natural Language Understanding,” which describes how NewSelector combines common-sense knowledge and natural language to allow a computer to “understand” the meaning of text, will be published later this year by Kluwer Academic Press.
IBM also is studying the use of computers to translate languages at its Los Angeles Scientific Center and at its Thomas J. Watson Research Center in Yorktown Heights, N.Y.
The potential market for computerized translation is huge, given the need for product data, instruction manuals, governmental reports and scientific articles in many languages.
At the L.A. Scientific Center, researcher Paula Newman is developing a prototype system to translate English to Spanish and to Chinese.
She said her project seeks to improve on existing translation software.
The process involves parsing the original text to reduce it to its basic components of meaning and then rebuilding those elements into the other language. Translation software currently runs on a mainframe computer, but someday may be brought to desktop computers, Newman said.
But she said it will be many years before computer translation progresses to the point that it could, for instance, create a Spanish-language edition of this newspaper each day from its English text.
Another promising area is computerized tutoring. Musician Linda Sorisio led the design of the Scientific Center’s Harmony Intelligent Tutoring System to teach music theory.
The mainframe-based program is designed to provide individualized instruction to students.
It not only catches their mistakes and gives remedial training, it analyzes underlying misconceptions that lead to such errors.
It is a true multimedia system making use of electronic keyboards, synthesized sound, video, graphics and artificial intelligence. Upon completion it will be tested by university music departments to determine if it improves on traditional methods of teaching this difficult subject.
Apple Computer doesn’t give press briefings on specific development work, but it has publicized the purchase of a Cray supercomputer to aid its research. It has even given Kent Koeninger the job of encouraging research on the Cray and a job title that only Apple would bestow--”Cray Evangelist.”
The unconventional-looking computer is displayed behind a glass wall in all its purple-pillared glory.
The $15-million computer is thousands of times faster than Apple’s Macintosh, yet until Apple engineers reprogrammed it, the Cray couldn’t display graphics as well as a Mac.
Now it is the most advanced graphics machine in the world, Koeninger said.
Apple uses the Cray to simulate what could be done on a personal computer if it had sufficient power. Koeninger explained that Apple’s working assumption “is that someday there will be a machine sitting on your desk that is as powerful as the Cray is today.”
If Koeninger oversees the distant future at Apple, Joy Montford, manager of the human interface group, has a firm grasp of today’s reality in computing.
Her job is basically to make sure that before Apple markets a new product, the product works the way it’s supposed to.
She’s given a lot of thought to the way things work, compared to how they ought to work in a more technologically advanced world than now.
In a nutshell, Montford says the problem is that today’s technologies are a series of islands with no way to span the seas in between. Telephones don’t link up with television sets, and television sets don’t link up with credit cards, and credit cards don’t link up with computers and so on.
When and if that kind of unity is achieved, Clarke’s vision of HAL will have been realized. We can only hope the reality will be more benevolent than the fiction.
Computer File welcomes readers’ comments but regrets that the author cannot respond individually to letters. Write to Richard O’Reilly, Computer File, Los Angeles Times, Times Mirror Square, Los Angeles, Calif. 90053.
