On the frontier: graphics, sound, AI
| San Francisco
A computer would jiggle its electrons just the same, regardless of what we call it. But, as recent developments make plain, those early computer pioneers who wanted to call these magic boxes ''universal machines'' had an extremely important point in mind. Although that unwieldy name never caught on, it's becoming steadily more apparent that ''computing'' - i.e., ''number crunching'' - represents only a small fraction of what computers can do.
There are many milestones by which the computer revolution can be measured: numbers of machines now in use, their cost, their performance, their physical size. But one of the most interesting is functionality.
As reflected in the term ''computer,'' the machines got their start processing numbers. Although they filled entire buildings, the first computers were essentially glorified calculators. Next, the capability to handle written text was added. And a number of ingenious ways of manipulating both numbers and text were developed. Other than limited or highly specialized graphics capabilities, however, this is pretty much where computer technology remained until recently.
This life story of the mainframe computer was repeated with its desktop progeny, the microcomputers. The program that really launched the personal computer market was Visicalc, a then-new but fundamentally arithmetic type of program that combined the functions of paper, pencil, and calculator in a form that has become generically known as the electronic spreadsheet. And once people had computers on their desks, the units were quickly adapted for word processing , now a primary use.
But this general rule of thumb among technical people should be kept in mind: If a factor changes by a thousand, it creates qualitative as well as quantitative differences. Thus, in the computer realm, each change of a thousand in the price, speed, or data-storage capability should open new and radically different applications. It is no surprise, then, that the advent of inexpensive microprocessors is hurtling the application of computer power toward largely untracked informational frontiers.
One such frontier is graphics. Ways of manipulating and processing images are proliferating. Another frontier is sound, both music and voice. And recent applications of artificial intelligence - the attempt to get machines to mimic intelligent human behavior - involve more abstract and formal types of symbol manipulation. An explosion of graphics
Graphics have been around in various forms since the beginning of computers. But in recent years the machines' quality and power in this area has grown quickly.
Apple's premier personal computer, Macintosh (see review on page B2), allows a user to draw and display pictures in black and white that approach the precision of pen and ink - a far cry from the crude pictures possible with the early videogames. And it also allows the mixture of pictures and words on the same page.
The popularity of Macintosh has spawned a number of similar ''paint'' programs on other machines. The parade is led by IBM's ''Color Paint'' on the PCjr. and Mouse System's ''PC Paint'' for the IBM Personal Computer. These don't have the resolution of the Macintosh, but they are in color.
Reasonably priced, high-resolution color - mainly for graphics purposes - is coming. This summer Commodore International Ltd. acquired Amiga Corporation, a small Silicon Valley company whose personal computer has received rave reviews for its full-color graphics capabilities. Commodore has said it will soon be introducing a Macintosh-like product for under $1,000. Apple's basic machine retails for $2,195.
The standard graphics capability of the IBM PC falls short of the Macintosh. But third-party vendors have begun supplying the electronics to enhance that popular machine's graphics abilities.
''Color graphics is like another language, one which will become increasingly important to business,'' argues William Chu, founder of Verticom, a graphics terminal maker that is in the process of building high-resolution graphics into the IBM PC.
There is already a substantial market for business graphics. The production of color slides for business is by itself a $4 billion annual market. Only 3 percent are now generated completely by computer, but the percentage is growing. In addition, more and more businessmen are recognizing the value of graphic visual aids. A Wharton School of Business study found that such aids can cut the length of business meetings by more than 25 percent and can significantly increase the success of sales presentations.
Even more important, believes Dr. Chu, will be the use of graphics to help executives manage their companies. The use of color-coded charts and graphs to summarize the complex information generated by a modern corporation can lead to quicker, more accurate management decisions, he contends.
Studies have concluded that the market for business graphics should grow over 40 percent per year and will exceed $1.4 billion in 1986.
New graphics terrority is being pioneered by Datacopy, a company that is providing the IBM PC with an inexpensive ''eye.'' This is a camera system that allows the computer to store and manipulate visual images, opening up a whole new world to the microcomputer. With a camera eye, traditional graphics can be stored and manipulated by the computer; pen and ink drawings can be added to technical manuals; electronic facsimiles of signatures and pictures can be stored in computerized personnel files.
''Instead of word processing, we can now do composing. Instead of database management, we can do document storage and retrieval,'' explains Rolando Esteverena, Datacopy's president.
Not only does this technology allow companies to store electronic copies of documents a la microfilm, but Datacopy recently released software that will convert written text into a form the computer can understand. All for under $5, 000. Breaking the sound barrier
Like the conquest of visual images, the assault on the world of sound is surging ahead. For home use, this has involved the development of the electronics and software necessary to convert the computer into a rudimentary music synthesizer (see Musicalc review on page B7).
In the office, the move into sound involves the gradual merger of the computer and the telephone. The integration of the telephone and personal computer into a compact desktop machine will be the next big advance in office automation, according to Thomas H. Bredt of Dataquest Corporation.
This is the battleground of corporate elephants. AT&T, with its telecommunications expertise, has turned to Ing. C. Olivetti & Co. to bolster its computer expertise. The two firms have announced an agreement to jointly plan, produce, and market a range of work stations and personal computers aimed at challenging IBM in this field. Few details of their accord were released.
Despite lack of enthusiasm in the trade press, the result of the companies' first collaboration, the PC 6300, has exceeded sales expectations. If demand persists, it will put AT&T in a better position to rival IBM as ruler of the next era for the personal computer market.
IBM, on the other hand, recently acquired Rolm Corporation to get the private telephone exchange (PBX) technology it needs to complete its office automation strategy. It is IBM's first acquisition since 1964. Although ''Big Blue'' paid a premium $1.25 billion for the company, most industry observers seem to believe that it was a smart long-term investment. The PBX is evolving into the communications hub for the office of the future. Personal computers, printers, and even large computers can be interconnected using its phone lines.
This trend has concrete as well as institutional manifestations. Several companies have produced machines that integrate telephones and microcomputers. Other companies are producing hardware and software that turn personal computers into superautodialers. An example is the MacPhone (see Macintosh review, page B2 ).
One of the latest developments is voice messaging. This is something like an ultrasophisticated telephone answering machine, with verbal messages digitized and stored on computer disk rather than on tape.
An example is the $595 PC Dial Log by CMC International of Irvine, Calif. This can hold a phone list of 200 names and numbers and dial them automatically. It can, unattended, call each person in a list, play a message, and record responses. It can even record a numeric response keyed in on a Touch-Tone telephone. When other people call, the program can ask for an ID number and then dispense messages intended for individual callers. Calling in from another phone , the owner can have the device change functions and record a message for a specific person.
PC-based voice mail systems are miniature versions of larger systems developed for companies. Until recently, these have been minicomputer-based and sold for over $500,000; products are now available for as little as $15,000. Sales are increasing by 87 percent annually and should exceed $1 billion by the end of the decade, market analysts forecast.
Further in the future are workable voice recognition systems. Getting a computer to generate recognizable human speech is simple compared with the process of deciphering human utterances. However, Texas Instruments markets such a voice recognition system for their Professional Computer. TI's Speech Command system can be used to operate the computer (a verbal command can replace up to 40 keystrokes), record phone messages, update and maintain a daily calendar, or for standard dictation. It costs $2,600.
This fall IBM researchers reported the development of a similar experimental system that quickly and accurately recognizes spoken English sentences. It has a 5,000 word business correspondence vocabulary, and it identifies more than 95 percent of the words in these sentences correctly. It differentiates between words like to, too, and two by context. As the user dictates, the words, phrases , and sentences appear on the computer's display screen where they can be edited using a standard word processor. The artificial intelligence frontier
Such voice recognition systems use artificial intelligence (AI) techniques - an area of research that attempts to mimic intelligent behavior using formal logic. AI is currently bursting out of the laboratory into the marketplace. The five primary areas of AI development are expert systems, natural language processing, robotics, machine vision and other ''senses,'' and conventional software development.
Industry estimates suggest that the market for AI technologies and applications, although currently in an early stage, is likely to exceed $2.5 billion annually by the end of the decade.
Expert systems are programs that attempt to capture, in software, problem-solving techniques used by human experts.On large computers these have been successfully developed for everything from medical diagnosis to prospecting for precious metals. At the personal computer level, the first application of this approach has been software by the Human Edge Corporation, which purports to help salesmen close sales and managers manage more effectively by analyzing the psychological factors in their work.
A number of companies are also working on ''natural language systems'' to enable the computer to respond to commands in plain English. Natural language is a term that has been much abused. Several companies market systems where the commands are all simple English words. These systems ignore words they don't recognize and operate on the key word commands. While these are called ''natural language'' systems, purists disagree.
AI natural language systems do much more. Ideally, they can interpret speech much like another person can, asking questions when they cannot decipher something. Such systems, however, are severely limited by the capabilities of the microcomputer. TI's NaturalLink, for instance, restricts vocabulary choices only to those it recognizes in a given context.
Artificial Intelligence requires special computers that run a list-processing language called Lisp. These are suddenly in demand. Major corporations and software houses are snapping up the Lisp computers to develop and run first-generation artificial intelligence programs.
The three original vendors in this emerging market - Symbolics, Lisp Machine, and Xerox - are girding for a boom they believe will push annual sales past $1 billion within a decade. And major computer makers like TI and Data General are jumping in.
Many of these new developments have been facilitated by IBM's increasing dominance of the personal computer market. A number of the pioneers are deeply disappointed that the market has been captured by Big Blue.
But, as Byte magazine editor-in-chief Phil Lemmons wrote in a recent editorial: ''Standards sometimes deprive us of the latest and most innovative technology, but when a standard encompasses both hardware and systems software, it cultivates the ground for a flowering of third-party software and peripherals.''
IBM was so successful, he says, because its Personal Computer filled a void in the market that existed despite the large number of players.
It provided a processor that could address a lot of memory, an 80-column screen, a keyboard with upper- and lowercase and good cursor controls, and word-processing software that used correctly labeled function keys.
In addition, IBM took the unusual step of publishing all the technical specifications for the machine, making it possible for third parties to produce peripherals and software in abundance. Not to be underestimated, as well, is the power of the IBM name. It promised stability in a volatile market.