Intel touts progress toward intelligent computers

SAN FRANCISCO--I hope Intel warned the Luddites and pessimists away at the door, because the chipmaker had a lot of bullish statements Thursday about its belief that computers will become smarter than humans.

At the Intel Developer Forum here, Intel Chief Technology Officer Justin Rattner showed off a number of technologies in computing, robotics, and communication that he cited as evidence that Ray Kurzweil's concept of "singularity," when machine intelligence surpasses human intelligence, is impending. Demonstrations spotlighted the wireless transmission of electrical power, dextrous robots with new sensory abilities, a direct interface to the brain, programmable materials that can be used for shape-shifting devices such as resizable cell phones, and silicon photonics that enables chips to communicate with photons rather than electrons.

"We're making steady progress toward Ray Kurtzweil's singularity," Rattner said.

Intel of course remains at its heart a chipmaker, and Rattner began with a brief tour, assisted by Mike Garner, senior technologist for Intel's emerging materials group, of various successors to the current complimentary metal oxide semiconductor (CMOS) process used to make processors. Future ideas that pack ever more computing capacity into a given volume include spintronics, quantum computing, carbon nanotubes.

Long live CMOS
And CMOS itself still has some legs, Rattner said, with recent progress shrinking the size of circuitry elements to their current size of 45 nanometers, or billionths of a meter.

Intel CTO Justin Rattner speaks at the Intel Developer Forum.

(Credit: Stephen Shankland/CNET News)

"When will silicon run out of gas? Can it fuel this exponential growth for 40 years to come?" Rattner asked. "We got very close to the limit at 45 nanometers. We were able to innovate our way out of what seemed an unsolvable problem...We've got some challenges ahead of us. It looks like 32 nanometers is on track, but you go beyond that and it looks a little bit iffy."

CMOS will benefit from "tri-gate" technology that will increase processing speed and circuitry density while drawing less electrical power, Garner said. Also on the horizon are new "3-5" materials that use gallium, indium, and arsenide to speed electron flow.

"We think CMOS will continue to be the workhorse for many years in the future," Garner said.

CMOS also will live on as a foundation for the post-CMOS era. It's not yet clear what will replace CMOS for the brains of the operation, though, Garner said: "It's like being back in the 1940s trying to invent the new transistor," he said.

Wireless data, wireless power
Wi-Fi has freed computing to some extent from pesky cables, and Rattner believes the power cord could be next to go. Intel showed wireless power transmission technology that used resonant properties of magnetic fields to transfer power from one antenna to another. The apparatus lost about 25 percent of the energy in transmission, but retained enough to light a 60-watt bulb.

Zachary Drake of Emotiv Systems demonstrates an interface that monitors external signals of brain activity with a special headset, then translates the data into actions in a basic video game. Here, he makes a lifting motion to lift objects in a video game.

(Credit: Stephen Shankland/CNET News)

The wireless technology loses less energy than current inductive transmission techniques over longer distances, but it still requires the broadcasting and receiving antennas to be relatively close. Still, Rattner said, it could be useful for charging a laptop or mobile phone that's placed on a properly equipped desk, and if it gets pervasive enough, it raises the prospect of devices that can be charged quickly through use of electrical capacitors rather than comparatively sluggish batteries.

Rattner also brough Jan Rabaey, a professor at the University of California-Berkeley, on stage to describe his vision for the future of radio communications. He believes each person will have about 1,000 radios soon, most of them vanishingly small.

Radio devices, Rabaey said, will become "cognitive," so they can automatically sense where there's uncluttered radio spectrum available and which communication protocols should be used at a given moment. He also thinks they'll become more collaborative, able to link together in a mesh network that collectively can transmit data faster, in greater quantity, more efficiently, and more reliably.

Indistinguishable from magic
Rattner wasn't afraid to trot out the oft-quoted observation from sci-fi author and space travel fan Arthur C. Clarke: "Any sufficiently advanced technology is indistinguishable from magic." Rattner's case in point was what he called programmable matter.

Rattner eyes an apple grasped by an Intel robot that senses nearby objects through changes in the surrounding electromagnetic field.

(Credit: Stephen Shankland/CNET News)

Intel envisions devices constructed from tiny programmable elements called catoms, a term coined by Carnegie Mellon University engineers that's short for claytronics atoms. Each has sensors, processors, and electromagnetic components that can control how far apart the catoms are from each other.

With the shape-shifting technology, a mobile phone could shrink for unobtrusive storage in a pocket, then expand to a more convenient size when in use, said Jason Campbell, an Intel Research senior staff scientist. Campbell showed some catom prototypes a few centimeters tall and wide but predicted miniaturization.

Catoms that could be used to build a fluidly reconfigurable model of a car, were 50 years ago recently, but Campbell said he now thinks the technology is only "a couple years" away now.

See also: From our colleagues at ZDNet, the video below shows Intel CTO Justin Rattner talking to Emotiv Systems President Tan Le about new interface technologies that are making humans more like machines.

Click here for full coverage of the Intel Developer Forum.

[TiVoJoe]

Better not tell Sarah Connor.

[DaveUssell]

Couple of years? At best this stuff is twenty years away for anything practical. This is really hard stuff. It still requires too much expertise on the part of the operator. That is hardly having the machine follow our whims. Who is training who?
Perhaps at some point the machines will be the next stage of evolution. Already people believe results on their calculator screen without thinking about it so the machines definitely have a toehold. They are not subverting us. We're submitting!

[spothannah]

It seems to be a matter of when, not if. How will be adjust? Will work mean anything? Will money mean anything? What will be the new metric? Hopefully someone thinks about kindness and caring. Hopefully.