Goodbye biology, hello software - conscious machines

UNESCO Courier, July, 2001 by Ivan Briscoe

Do we really need our flesh and blood? Ray Kurzweil, guru of Artificial Intelligence, believes that the conscious machine is only decades away--so get ready to download

How close is Artificial Intelligence to creating something that resembles a human?

There are broadly two levels of Artificial Intelligence (AI).The narrow level is about getting non-biological systems to perform activities that require human intelligence when we perform them. Strong AI is about applying the full range, depth and subtlety of human intelligence in a machine.

We have many examples of narrow AI today. Machines can diagnose blood cells, guide cruise missiles and play games likes chess better than any human. They can read books and they can understand human speech. In comparison to humans, their memories are more accurate, they are faster and they can share their knowledge instantly by quick downloading.

There are two basic requirements for moving onto the next stage. One is just the brute force hardware capacity. Already, we are working on certain circuit technologies that will be a million times more powerful than the human brain, at least in terms of raw capacity. But the more important issue is the software of intelligence. In this regard, the most compelling project underway is one to reverse engineer the human brain, possibly by using billions of tiny high-resolution scanners or nanobots that can travel through capillaries and scan the brain from inside.

My feeling is that by 2030, we'll be able to recreate how several hundred regions of the human brain work in non-biological mediums.

Would that non-biological medium then be equivalent to a human?

We already have technology called a neuron transistor--an electronic circuit that can communicate wirelessly in two directions with biological neurons. We also have examples of intelligent machines placed in human brains via neural implants. A treatment for Parkinson's Disease, pioneered in France by Dr AlimLouis Benabid replaces the biological cells destroyed in the first seven or eight years of the disease. The doctor demonstrated this by controlling the device from outside: when turned off, the patients remained in an advanced stage of Parkinson's--they were rigid like statues. He flicked a switch, and it was as if they had suddenly come alive.

By 2030, we will be able to send billions of nanobots inside the human brain to communicate wirelessly with billions of different points in the brain. They will give us the ability to combine our biological intelligence with non-biological intelligence. Ultimately you will have entities that are completely non-biological, with copies of human brains derived from reverse engineering, and you will have biological humans that have billions or trillions of nanobots in their brains augmenting their intelligence or enabling them to exist in virtual reality. So we will no longer be able to say humans on the left, machines on the right.

One of your most startling ideas is that of downloading the brain of an individual onto a computer system. Is this truly possible?

I discussed in my book some thought experiments of what that would be like. Is that the same person, for example? One simple thought experiment argues that it's not, that if you scan my brain while I'm sleeping and then reinstate it in a non-biological medium, I would not even necessarily know about it. I'd wake up, feeling nothing was different. Someone would come up to me and say: 'Good news, Ray, we don't need your old brain any more.' I would probably not accept that perspective. If you spoke to the new Ray, he would have a memory of having been me, but at that point he is a different person.

There's also the issue: is he a person at all? Is he conscious, or is he just acting that way? But the core essence of consciousness is ultimately not penetrable by scientific examination. We have a shared assumption about humans that they are conscious, but that shared consensus breaks down if you go outside humans. I'm really making a political prediction that humans will accept non-biological entities as conscious because they'll be so convincing in their behaviour.

What would be the benefits of being transferred onto a non-biological medium?

A lot of profound scenarios emerge, one of which is full immersion in virtual reality. These virtual reality environments will incorporate all of the senses, and will also be able to augment human intelligence. Right now we're restricted to a mere hundred trillion neuron connections per second, millions of which are needed to operate one chunk of knowledge. We'll be able to vastly expand the human brain--think faster, think bigger, more complex thoughts, have more knowledge and download knowledge.

But many neurological theories argue that consciousness is in fact very much mixed up with emotions, with a sense of being in the body and having objects act upon the body.

Intelligence is the ability to solve problems using finite resources, including time. And the problems we solve have much to do with our body--protecting it, feeding it, clothing it, providing for its needs and desires. A lot of our thinking has to do with our bodies: a disembodied human intelligence would quickly become depressed.