Saturday, January 13, 2007

Nick Bostrom Says We Are A Computer Simulation

The most foolish article I have ever read in a serious publication was by Nick Bostrom, claiming that we are probably a computer simulation.

More precisely, he argues that either 1) most civilizations destroy themselves before they become technologically mature or 2) only a tiny fraction of technologically mature civilizations are interested in creating computer simulations of their ancestor civilization or 3) we probably are a computer simulation.

If 1 and 2 are not true, then future civilizations will want to devote some of their immense computing power to simulating their ancestor civilizations. Calculating the probabilities, Bostrom finds that there will be many far, far more computer-simulated minds in simulated ancestor civilizations than natural non-simulated minds in non-simulated civilizations. Therefore, it is much more probable that we are simulated minds living in a computer simulation than it is that we are non-simulated minds.

What is wrong with this argument? Obviously, the problem is Bostrom's assumption that computer-simulations of people will have consciousness, as we do.

We may develop so much computing power that we can do precise simulations of the weather, even calculating the effect of each butterfly flapping its wings. But no matter how powerful our computers are, standing near a computer simulation of a rain storm will not get you wet.

Likewise, it is possible that, no matter how powerful our computers are, no matter how precisely we can simulate people's behavior, a computer simulation of a person will not be conscious - any more than the figures in video games are conscious.

This is a major issue in contemporary philosophy of consciousness. I think that philosophers such as John Searle have shown that computer can never develop consciousness. Cognitive scientists sometimes ignore the question, sometimes deny that the question is meaningful, and sometimes claim that computers can be conscious.

How can Bostrom do detailed calculations of probability based on the number of natural humans and the number of computer-simulated humans that we can expect to exist, while completely ignoring the improbability that computer simulations of people could ever be conscious?

This combination of mathematical astuteness and philosophical blindness has earned Bostrom the position of director of the Future of Humanity Institute at Oxford University, where he spends his time arguing that we should use genetic engineering to change what it means to be human. No doubt, the Nick Bostroms of this world have the wisdom needed to decide exactly how humanity should be reengineered.