I've just finished reading Steve Grand's second book, Growing Up With Lucy (how to build an android in twenty easy steps). He points out on the page just before the introduction that he lied about the number of steps, but never mind that. I'll even give away the ending. By the end of the phase of the project described in this book, his orangutan-inspired android "daughter" Lucy has learned to (sometimes) point at a banana.
This may not sound like much, but it's actually a good step beyond Steve's earlier artificial life project, the computer game Creatures. The development of that game and of Steve's ideas about the nature of life and intelligence were described in his earlier book, Creation, which I wrote about in 2006. It's pretty amazing stuff - all the more so because the creatures in Creatures actually learn to do everything they do in their fairly complex 2D world. They have brains and immune systems and so on. Richard Dawkins has said that Creatures could be the closest that anyone has come to actually creating artificial life.
But Steve was not content to stop there. As complex and realistic in some respects as the artificial life forms of Creatures could be, they still developed in a very limited 2D world. Steve makes the very logical point that we (humans and other mammals, mainly) do not exist as isolated "intelligences" but that intelligence and other attributes are consequences of having bodies that have to learn to survive in a real 3D world. With Lucy, he set out to create a "creature" that would actually have use for intelligence. In the Lucy book, he describes his ideas and his self-funded "non-disciplinary" research in more detail, in addition to describing the trials and tribulations of Lucy and her software-emulated muscles (driving real and very troublesome hardware) and brain structures (modeled on the overall architecture of the mammalian brain, with a visual and motor cortex and so on). It's a "synthesis" approach that makes a lot of sense, and his telling of it is quite clear and often quite funny.
At the end of the book, Steve reports that his self-funding had just about run out when he fortunately received a UK foundation grant to continue the work for another year, allowing him to start work on a "Mark II" version of Lucy. That was around 2000 I think, and to learn more I'm afraid you will have to explore his web site (which has frames, alas, the de-framed Lucy part is here). It seems that Lucy is again (still?) on hold, but maybe things will change, and maybe someday Lucy-type robots will be common and will be called grandroids in Steve's honor. Then again, maybe not.
In any case, this is a very clever, sometimes funny, and always thought-provoking book, and I highly recommend it. Now I have to pack for a lightning trip to Taiwan, leaving for the airport in about 11 hours.
No comments:
Post a Comment