Daniel Dennett is the hacker's favourite philosopher. Since shocking traditionalists with the claim that there is no essential difference between human and machine minds, he has gone on to work with some of the key thinkers in both biology and artificial intelligence. Together with MIT's robot guru Rodney Brooks, Dennett is building a robot called Cog which, he hopes, will grow older and wiser like a human child, learning from the experience of exploring its environment. His latest book is Darwin's Dangerous Idea, a passionate and controversial defence of natural selection as the only valid explanation for both life and consciousness.
Wired: What makes Darwin's idea so dangerous?
Dennett: Darwin's idea completely reverses a traditional way of thinking. Before Darwin, we had a top-down or mind-first theory of the design of the world. In the beginning, there must have been a cogitative being. Darwin turned that around and said all the minds in the world can be the effect - not the cause - because mind can be created by mindless processes. A lot of people are very unhappy with that.So what do we gain from this new perspective?
For the first time, we can explain some of the things which are simply left hanging in the air by the old tradition. Any theory of intelligence has to break it down in the end into things which are not intelligent, mindless mechanical processes. Any other sort of explanation simply begs the question. There's no way to ground a theory of intelligence that isn't ultimately biological or physical.Back before the mid-19th century, vitalism was a very popular doctrine. It held that living things had some special property which set them apart, that there was an unbridgeable gap between living and non-living. Today, the processes of life get explained in terms of fundamen-tal biological and physical processes. There's no need for any special life force. But, surprisingly, people are not comfortable extending that principle to mind or consciousness.
Which leads us into the question of ethics. You produce Darwinian ethics. How does that work?
The main difference between what you call my Darwinian ethics and earlier versions is that I don't claim to deduce ethical views from Darwinism. I claim something much more modest - that any ethical theory that is sound has to be consistent with evolution. That's a tough demand to meet because it means you can't assume that there is such a thing as good will or ethical truth. If you're going to appeal to ethical concepts like cooperation and selflessness, you have to show how those phenomena could persist in our world; how they could be stabilised by evolution. That's a hard challenge to meet. It gives a different look to a moral theory.One of the differences that you mention is that ethical decisions are made online. What do you mean by that?
One of the fundamental insights of evolutionary theory is that the processes which work are not foolproof. They have to take risks. They cannot take all possibilities into account. That's true of evolution and it's true of ethical decision making. So, there is no dichotomy between two sorts of choices: one that considers all things and is divine and another that doesn't consider all things and is evil. There are "good" processes of discovery and ethical theory.All intelligent processes are at bottom what computer scientists would call generate and test. They make a lot of candidates via a relatively stupid process. They test those candidates through some sort of filter, some sort of evaluation process. They take the results and do it again and again. The Darwinian algorithm is generate and test at the genetic level. Skinner proposed that intelligence was the generation of behaviours which were either reinforced or not. I show that in fact there are generate and test processes everywhere.
Of all the generate and test processes which have so far existed on this planet, science - by which I mean, broadly, rational inquiry - is the first one which is deliberate and full of foresight. The process is undertaken to winnow out the truth and is a much more powerful and swift process as a result, because it is guided by foresight - imperfect foresight, of course. It follows that, since science puts everything under the microscope, it engages in a con-tinual redesign of itself. It is con-stantly re-evaluating its own methods. It evolves just the same as all the processes it studies.
How close is the meme to the gene? Do our ideas evolve?
Both genes and memes are informational units, but memes are much more promiscuous in the media that carry them - languages, pictures, practices, concrete artefacts of various sorts. So it's much less likely that a strongly mathematical and predictive theory of memetics can be achieved. But it's not out of the question. If we look at the areas of human culture which most closely resemble genetic systems, then memetics has a good chance. For instance, it is often noted that human minds are not slavish copiers of ideas but creative, critical combiners of ideas. They don't usually take the ideas they get and make copies of them, like genes are copied. But there are areas of human culture where that's not true, where the rule is not to criticise, not to change - religions, for example. So in areas such as religion, memetics has a much better chance of yielding interesting results.So is evolution not really useful in studying ideas?
The fact of cultural evolution is just as uncontroversial as genetic evolution. Everyone has known for a long time that civilisations evolve, cultures evolve, practices, agriculture. The only question is whether the explanation of this evolution is Darwinian. That is what Richard Dawkins proposes with the concept of the meme.When you look at cultural evolution from the memetic perspective you appreciate that the persistence of an idea in our culture may not be dependent on its value to us, because it may have its own independent way of encouraging its replication. The memes that survive are the ones that make lots of copies; they may not necessarily be good for us.
Is science better for us than religion?
I'm not hostile to religion at all. I am alarmed by fanaticisms which take their vision of the law into their own hands and promote violence. I find that perhaps the most dangerous social force on the planet today. Aside from that, I'm not hostile to religion; I think that religion has been a wonderful thing. Throughout the world, we have already recognised that there are very clear limits to what we will consider civilised in a religion. I simply draw attention to the fact that we already recognise that religious freedom has its limits.Do ethics have limits too? If there is no sharp and clear distinction between living and non-living, do we have ethical obligations to machines?
It's possible in theory - if difficult in practice - to make a robot which is a sentient pursuer of its own projects, a self-protective and comprehending agent. So it is in important ways a living thing, a living thing that has not just needs and desires but also values. I think that as soon as one has created such an entity - whether the old-fashioned way, procreation, or by modern technology - one has a responsibility to protect its rights and to treat it as more than just another artefact.Hari Kunzru is a section editor at Wired.