Humans and Machines Converge at ACM1

May 8, 2001 by Amara D. Angelica

Humanoid robots aren’t perfect, but they may have a thing or two to teach computers.

Originally published May 8, 2001. Published on May 8, 2001.

“His love is real but he is not,” reads the teaser for the forthcoming A.I. movie, referring to David, a lifelike robot boy who wants to be a real boy.

But speaking at the recent ACM1 (Association of Computing Machinery 2001) conference in San Jose, Michael Dertouzos, professor and director at the MIT Laboratory for Computer Science, said computers should act less like robots and adopt more human-like qualities.

Instead of freeing us up, he said, today’s technology often enslaves us, as anyone who has filled out a complicated order form on an e-commerce site or dialed a toll-free number recently has discovered: “Press 1 for sales; press 2 for customer service; press 3 if you’d like to wait on hold for another 10 minutes while we pipe mind-numbing Muzak into your ear…. The average executive spends 1.4 hours daily reading and responding to e-mail. This is absurd.”

Office work now represents approximately 60 percent of costs, or $14 trillion, in the economy, he pointed out. Automating those tasks or deploying them to workers in developing countries over the Internet, he said, could achieve productivity gains similar to those achieved by the industrial revolution.

The two billion people in the world who can’t read or write are currently shut out of using the Internet, he added. Speech recognition and synthesized speech could open it up to them, as well as make the Internet more accessible to the roughly one billion people who use ideogram-based languages, such as Chinese. “We are not exploiting this technology revolution. We’re hardly scratching the surface.”

“When the computer finally vanishes under the human services that it provides, then the revolution will be finished.” Dertouzos offers a roadmap for getting there in his recently published book, The Unfinished Revolution: Human-Centered Computers and What They Can Do for Us.

Rod Brooks, director of the Artificial Intelligence Laboratory and Fujitsu professor of computer science at MIT, also speaking at ACM1, pointed to the current generation of consumer robots as a step in that direction. For example, Furbies, which learn words that a child “teaches” them by repetition, help to educate people about the potential of robots, he said. In the near future, robots will also play a role as domestic help.

“We’re talking about the emotional coupling between the robot and the human,” Brooks said. Robots will have to learn to emote and receive and transmit nonverbal cues. “We’re pushing on having a human form just to…understand the relations between robot and machine. If the robot has humanoid form then it will be both easy and natural for humans to interact with it in a human-like way.”

He showed a video of Kismet, a humanoid robot being developed at MIT that recognizes some human emotions and behaviors and has a primitive face that can express its own moods. It learns how to behave through interaction with people. Kismet is so real that when it’s turned off, people in the lab feel sad, he reported.

But Brooks doesn’t stop there. In the future we will “become machines, with leg and arm prostheses, retinal implants, and controls over the body’s biological processes.”

Vint Cerf, Senior Vice President, Internet Architecture and Technology, WorldCom, offered some examples of where this might go, including an in-the-ear telephone (voice picked up by bone conduction), glasses that project an image directly onto your retina, neural implants, memory upgrades (for humans–a solution for Alzheimer’s, he noted), and nanodevices inside the body, such as an insulin pump. In 2050 there will be an estimated 11 billion people in the world and 100 online devices per person, he predicted.

Ray Kurzweil, a pioneer in pattern recognition and artificial intelligence, took it every further, describing a future in which humans and robots are so alike it’s difficult to tell them apart. Within 20 years, he says, computers will appear to be conscious, feeling beings demanding the rights and privileges of humans. By the end of this century, we will spend more time in virtual reality than we spend in reality, he predicted.