Tuesday, April 21, 2009

Transcendent Man

One of the films at this year's Tribeca Film Festival is Transcendent Man, a documentary that looks at the life, ideas and cult-like following of futurist Ray Kurzweil.
Ray Kurzweil is one of the world’s foremost inventors. At age 15 he was designing programs that were adapted by IBM and soon after machines that allowed the blind to read. Today he is hailed by some as a modern-day Nostradamus and dismissed by others as a crackpot. The “futurist” and best-selling author is a leading theorist on the “technological singularity”—a time when humans and machines will fuse in the next phase of bio-technological evolution, creating superintelligent, godlike beings that could conceivably live forever. The kicker is, Kurzweil claims that this monumental change is destined to happen in just 30 years.
And Kurzweil has said in an interview with Brooke Gladstone for On the Media that the change will come in closer to 20 years:

RAY KURZWEIL: By 2029, we'll have finished the reverse engineering of the human brain. There’s already 20 regions of the brain we've modeled and simulated and tested. We'll have very powerful and very small computers by that time. Most of the computers in the world are not yet in our bodies and brains, but some of them are in our brains. If you’re a Parkinson’s patient you can put a computer in your brain. It’s not blood cell-sized today, it’s pea-sized.

And if you take what we can do today and realize these technologies will be a billion times more powerful per dollar in 25 years, a hundred thousand times smaller, you get some idea of what we'll be able to do.

It seems to me that Kurzweil overestimates our current understanding of how the brain works. The Blue Brain Project - a major research to create a supercomputer-based functional model of a mammalian brain - used 15 years worth of experimental data to map the connections of 10,000 neurons in the 2-week-old rat brain. That's a tiny fraction of the 100 billion neurons found in an adult human brain. Considering that the experimental procedures currently used to map the rats' brains would be unethical to use in humans, a substantial increase in computing capacity is not the only technical hurdle to overcome. I do believe the human brain will be completely mapped eventually, just not in 20 years. And even once an "average" brain is completely mapped, there is the separate problem of recreating unique individual brains. I suspect it will be significantly easier to construct a computer-based intelligence from scratch than to build a computer that models a specific brain with all its memories and personality intact.

I personally believe Kurzweil's speculations are more science fiction than science, but maybe he will turn out to be the prophet people make him out to be. Watch the trailer for Transcendant Man:

(via Popular Mechanics)

Tags:, ,

1 comment:

  1. on the lag for modelling neurons, rat embryos, humans
    I agree with you, or did, until the memristor came along allowing synaptic like electronics (and it can be analog, and solid state, and can be turned off and on keeping its old state)

    so amazing I wrote a story about it the week I heard about it.


I've turned on comment moderation on posts older than 30 days. Your (non-spammy) comment should appear when I've had a chance to review it.

Note: Links to Amazon.com are affiliate links. As an Amazon Associate I earn from qualifying purchases.