The past 30 years have seen about a 10-million fold increase in the cost-effectiveness of computers. This is a radical and astonishing advance—but it conceals a strange fact: While computer efficiency has risen exponentially, computer architecture has scarcely changed at all since John von Nuemann first defined it in 1948. Most of the gains have come from cost-reducing old ideas improving the von Nuemann components—the memory chips and CPUs and connecting busses...

Now the old von Nuemann structure is beginning to break down. Seeing this development, some observers declare that the industry is "maturing," that the technology is somehow "graying." But the real message of this era...is that computer technology is still in its infancy. After all, the computer can still barely see or hear, let alone walk.

Because of these handicaps, it is now fashionable to disparage Artificial Intelligence. "It's been going on for some 30 years, and apparently achieved little."

The computer industry is said to have thrived by doing well what human beings do badly. But AI seems to thrive by doing badly what people do well [laughter]. This charge is partly true.

But assume that you are a computer—essentially, a deaf, dumb and blind machine. If somebody allowed you to see and hear, even relatively inefficiently, wouldn't you hail 'em as the new Edison?

Marvin Minsky has said that a prime discovery of the AI Movement is that the bulk of human intelligence is acquired during the first year—all the perception and motor-skills and hearing, and all the perceptive mechanisms of the brain. By overthrowing the old von Nuemann agenda, AI will make the computer into a toddler, and change the world. And I think the result will be another 10-million fold rise in the cost-effectiveness of computing. And this time it may not even take 30 years.

A first step, I believe, is even now occurring at Kurzweil Applied Intelligence in Waltham (MA). The computer will introduce a Talk-Writer with an available vocabulary of some 250,000 words, an active vocabulary of 10,000 words, a learning faculty which takes dictation at an average pace of some 60 words a minute.

Using conventional devices, the Talk-Writer would have required 4,000 MIPS, 4 billion instructions per second, of computing power. [But] the Talk-Writer is the product of using computer-aided chip-design to create massively parallel, application-specific integrated circuits, dedicated, in a way, to what are essentially artificial-intelligence functions.

The key fact about this machine, however, is that it will sell for about $5,000... That's a little over one dollar a MIP. Similar breakthroughs in cost-effectiveness are occurring all through the industry, using computer-aided engineering technology to create massively parallel devices that can mobilize, essentially, faculties of artificial intelligence...

Today, while Kurzweil prepares to sell computing power for a dollar a MIP or so, IBM sells its mainframes for $150,000 a MIP. Even the new 386 general-purpose PCs give you 90 times more MIPS for the dollar than the new IBM Sierra. Used in networks and in parallel configurations, PCs will increasingly give mainframe performance.

But massively-parallel, application-specific devices will give thousands of times mainframe performance for specific functions, per dollar.

In the past, individuals have often been dependent on great organizations to achieve great goals. The individual entrepreneur relied on land, labor and capital—essentially territory and raw materials—to maintain his company and expand wealth. The New Age will free him from that dependency.

It will free him first of all, from raw materials. Secondly, it will free him from being tied to specific territories. Unlike natural resources, knowledge can be used without being used up. In fact, knowledge grows as it is shared and applied

With the new technologies—mostly software and chip-designs—they can be created by one person with a $25,000 workstation, anywhere in the world. In the past, to offer a chip, you essentially had to own $100,000,000 semiconductor fab. And indeed almost all the chips were authored by the owners of the fates. It's as if books were all authored by printers.

But, as the diffusion of computer-aided engineering [proceeds], the number of chips designed will probably increase from 10,000 last year to some hundred thousand in the next three or four years. And more and more of these chips will be designed by individuals and companies across the country, rather than by engineers in semiconductor firms.

Similarly, most of the new software products are productions of individuals or small teams. It's no longer necessary to have a huge capital plant in order to generate the critical products of the new era.

So the key lesson of the age of intelligent machines is that knowledge is power. But today that's true in a new sense. It means that other things are not power. And the things that no longer confer power are all the dreams and goals of tyrants and despots over centuries: The control of territory, the control of natural resources, the control of military manpower, and the control of national economic planning. This is an age when entrepreneurs can flash their capital down fiber-optic cables and bounce it off satellites, send it around the world in microseconds, and the entrepreneur can follow it in hours on a 747, and he can move his company in weeks.

Against this new power of individuals and entrepreneurs, the state is increasingly impotent and at bay. The good news for entrepreneurs is deadly news for the Socialist Dream. The only way the state can enhance its wealth and power today is to emancipate individuals to create new knowledge. Nations will no longer be able to seize power, they'll have to attract it and liberate...

An example is Cuba. People imagine that Cuba is an island "90 miles south of Florida." But Cuba, in essence, is no longer on that island. Cuba—the enterprise of its citizens, the Cuban culture, the Cuban economy—everything important about Cuba has moved to Miami, and to other parts of the United States. [Laughter]

And Castro thought he was capturing the essential wealth of Cuba when he expropriated the means of production, but the real means of production were in the heads of the people of Cuba—and they left.

We don't live any more in geographical time and space. We live in real time and space. And the nations that prevail in this new environment will be the nations that earn it, by liberating their people to create new knowledge, and thus new power for the world. Thank you.