Burton MacKenzie

Today J. Craig Venter Institute scientists announced they have built the first bacterial genome from the raw chemical components of DNA.

If this process can be done entirely by machine instruction, we could build a programming interface to assembly of DNA. In other words, we could make a C/C++ to DNA Compiler. (also from the other language, LISP) We could literally design a new functioning organism from the ground up.

I'm not looking forward to the bugs in the code. (eg. "remember that bacteria we created to eat oilspills? it turns out it also attacks the oily animal fat in mammals and now we're all going to die")

Somebody's going to write a literal software virus that targets biologicals, and it will knock out the population. It will be the work of a script kiddie.

Maybe somebody will write a really useful organism, like something that targets amyloid beta plaques in the brains of Alzheimer's patients. Then, the helpful djinni will mutate to attack our immune cells and destroy us all.

If we ever actually create artificial intelligence, we're not going to do it first in silicon. We need the efficient raw processing power of a biological system to get the computational complexity we need. My money's on the first artificial intelligence arising, if anywhere, in artificially designed flesh and blood.

It is probably a muchos bad idea to design an unrestrained intelligence that lives in our ecosystem and could outcompete us. It would be like unleashing a new kind of cockroach into our ecosystem, except the cockroaches would be smarter than us.

All in all, as much as I'd love to have a C/C++ to DNA compiler, Pandora's Box seems mighty big. It's probably inevitable, but even still, I have hope.

Burton MacKenZie www.burtonmackenzie.com

Update: Here's some related informational links:

• From the J.Craig Venter Institute overview on Synthetic Genomics, "Synthetic genomics combines methods for the chemical synthesis of DNA with computational techniques to design it."
• DNA 'Fabricator' constructs walking DNA, "Just as computer languages let programmers create any number of applications, the researchers behind the approach predict that biochemical programming "languages" inspired by their work could let bioengineers create any number of desired molecular products and processes". (At time of this update, the newscientist original was down, but you should try it anyway)
• From Nature 451, 318-322 (17 January 2008), Programming biomolecular self-assembly pathways.
• Start Hacking Life: Desktop DNA Synthesizers, "your new DNA compiler ... fits on your desk and can take genetic code that looks like this, gatcctccat, and turn it into an actual DNA sequence. "