Originally published June 6, 1999. Published on KurzweilAI.net May 31, 2001.

Q: Do you have any doubts that a superior intelligence will emerge in the next few decades?

A: No. It’s inevitable. For example, nanotubes would allow computing at the molecular level. A one-inch cube of nanotube circuitry would be about a billion times more powerful than the human brain, in terms of computing capacity. That raw computing capacity is a necessary but not sufficient condition to achieve human-level intelligence in a machine. We also need the organization and the software to organize those resources. There are a number of scenarios for achieving that. The most compelling is reverse-engineering the human brain. We’re already well down that path, with techniques like MRI. But we’ll do better because the speed and resolution–the bandwidth–with which we can scan the brain are also accelerating exponentially.

One means of scanning the brain would be to send small scanners in the form of nanobots into the blood stream. Millions of them would go through every capillary of the brain. We already have electronic means for scanning neurons and neurotransmitter concentrations that are nearby, and within 30 years, we’ll have these little nanobots that can communicate with each other wirelessly. They would create an enormous database with every neuron, every synaptic connection, every neurotransmitter concentration–a precise map of the human brain. So we’ll have the templates for human intelligence.

Once we can embody human thought processes in a nonbiological medium, it will necessarily soar past human intelligence for several reasons. First, machines can share their knowledge electronically. With humans, you spend years teaching language to each child. [But] once any one machine has mastered something, it can share that knowledge instantly with millions of other machines over the global wireless Web, which we’ll have by then. So a machine can become expert at any number of disciplines.

Complete article at Businessweek.com