Cat’s the one who started me on DNA computers (we share a grad student office). My AI Otto is struggling with my need for speed in his computations and his need for energy to complete the work. When I ask him a question, he sorts through a datasphere the size of the digital Library of Congress (all public sources on the internet. Imagine if you searched ‘Homo erectus’ on the internet and then read and absorbed the one million hits–that’s what Otto does just to get started) to create the simulated reality required for his movies. You can see the importance of speed.
Here’s what I know about DNA computers. They weigh almost nothing, carry their own energy pack, can perform ten trillion operations at once and store an amazing amount of information–all in a drop of water with room to spare. The mechanics are deceptively simple. A high school senior won a scholarship by programming the Declaration of Independence into a DNA molecule. Here’s a link to How Stuff Works if you’d like more information.
The problem, from what Cat’s explained, is the amount of error in DNA computing. In our human genome, we call them mutations and they’re considered part of our uniqueness. The average child has around 6.3 billion base pairs of DNA with around 277 mutational differences from his/her parents. Many are noninvasive because 1) cells have built-in redundancies, 2) parts of our genetic make-up are inactive. Maybe they used to be active, but with H. sapiens sapiens, they aren’t. 3) some have nothing to do with how we get along in the world.
But, for traditional computing needs, we need more accuracy than that. The theorists believe that within highly-structured uses, they can be controlled. Taiwan has already created a chip out of DNA.