Artificial Intelligence project proposed

-A A +A
By Roger Snodgrass

The title of Art Morse’s talk for the Computer Users Group at the Los Alamos Senior Center Tuesday was a little misleading.

As it turned out, “A Hundred Years of Artificial Intelligence,” was not about the last hundred years, but rather it was neatly divided between the first 50 years, when computers were just becoming accessible and 50 years that haven’t happened yet.

The next 50 years are important because that will be the time frame within which artificial intelligence will almost surely be upon us, according to many projections.

The enormous transformations to come were first described as a “technological singularity” by mathematician, computer scientist and science-fiction writer Vernor Vinge in 1993.

“We are on the edge of change comparable to the rise of human life on Earth,” Vinge announced. “The precise cause of this change is the imminent creation by technology of entities with greater-than-human intelligence.”

Despite a great deal of excitement, research and progress in artificial intelligence has surged and ebbed.

Technological progress may be inevitable and there is even a scenario in which mankind will wake up one day and discover that all the computers are connected and plotting our downfall.

While he is aware of those scenarios, Morse is more interested in exploring what he thinks are missed opportunities in computer hardware and software that would enable machines to “think” more like the human brain.

In the last 50 years mankind has plunged deeply into computing, Morse said, getting very good at mathematical problem solving and storing and manipulating data.

“Things were improving rapidly and problems could be solved thousands and millions of times faster than people could do them, fueling expectations that sooner or later machines could do everything we could do,” said Morse. “How much sooner or how much later was the question.”

Major efforts were made in the first half-century on behalf of precursors for artificial intelligence, like knowledge-based systems for medical diagnosis, or expert-based systems that tried to compress authoritative experiences into a usable database, but these systems fell well short of a machine that could think like a person.

Meanwhile, a second front began to open, related to rapid advances in knowledge about the brain, how it worked and how complex and even counterintuitive many of its functions and processes turn out to be.

The first fifty years have passed, Morse said, and we still don’t have a way to measure machine intelligence.

Or maybe it could be said another way, Morse suggested, that there is not enough machine intelligence yet to measure.

We specify how many instructions per second the computer can perform and how many numbers beyond the decimal point it can calculate. Yet even the fastest and most powerful high-performance computers have yet to make a blip on the scale of demonstrating any real smarts much less intelligence or genius.

A problem Morse worked on while he was at Ramo-Woolridge (later TRW), a conglomerate that included an advanced technology and information systems division, had to do with pattern recognition.

How can an overhead observer identify a black aircraft on a white sheet of ice?

A person can see it quite easily, as even a trained dog or cat can see it, but a computer can’t see it, because there is no sync, no set depth or alignment in the image.

“Brains use a relational or associative approach,” he said. “One thought triggers another.”

Morse recently founded a new company Cognetix, LLC, to explore novel hardware approaches that would enable greater and faster associative memory by means of an integrated circuit known as a CAM (Contents Addressable Memory).

Morse believes that CAMs, given a more compatible computer architecture than is now available, can address every memory element in a given data set simultaneously, so they are better suited for finding the ambiguous, incomplete or scantily defined information that our brains are so good at.

In the discussion that followed there were questions about the dangers inherent in the search for artificial intelligence.

One member said, “ I am not a Luddite; I am a real naysayer.” (Luddites revolted against automated weaving looms in England in the early 1800s.)

Morse said he was taking a poll on such concerns. He said he had found that more people were more worried about the risk of bad decisions by incompetent leaders than they were about runaway artificial intelligence.

“The talk was at higher level and less technical than we usually have,” said Ralph Partridge, who leads the computer education group, after the meeting. “He mentioned things we do talk about. Some of us have been in AI.”

Morse is seeking technical and entrepreneurial assistance in his project and can be reached at 505-661-4832.

A website is under construction at http:www.cognetixltd.com.

EDITORS NOTE: It should be noted the writer is an officer in Cognetix, LLC, the company that was recently incorporated by Morse.