|Growing up with Lucy|
|publisher||Weidenfeld & Nicolson|
|summary||The design and development of Lucy the robot and especially her brain.|
Steve's goal is to build an intelligent android inspired by his understanding of the human brain. This book is the story so far of the creation of Lucy the robot (named for the famous fossil hominid). It's an experiment to circumvent what Steve sees as an impasse in current progress in AI which he describes as being "stuck halfway up a dead end creek without a paddle." Now Steve is not a neurologist, or a biologist, nor even an electrical engineer. He describes himself as a 'non-disciplinary' thinker. He's an ex-schoolteacher and a computer game designer, admittedly one so renowned for his advanced thinking that he received the Order of the British Empire in acknowledgement of his work. The game he made is called 'Creatures' and represents a peak in artificial life software- it's about cute little beings called Norns that you raise from eggs and have to teach and train (and if you feel a bit godlike you can tinker with their software genes). But still this is not the sort of background one expects to lead to a career in robotics.
If you read his previous book Creation: Life and How to Make It, also reviewed by me on Slashdot, you'll be aware of how radical his ideas can be. And perhaps not be so amazed at this next step in extraordinary ambition. But as he says, you can't jump to the moon incrementally. Reading this book is like trying to learn neurology and electrical engineering at the same time, with a bit of how to fly a plane thrown in for good measure. But it's so readable you can do it and laugh at the same time. There's something about Steve's writing style that's reminiscent of P.G. Wodehouse. This is a book that makes you feel inspired and despairing. Inspired that one man can have so many brilliant insights, the skill to make them into real working mechanical inventions and the courage to go it alone; despairing that our academic and funding resources have been such a failure at support for his endeavors.
His project is to create a robot capable of developing a mammal-like intelligence (an orangutan is the current external model, mostly down to an ugly orange wig and long arms). Yet for most of the development time, Steve says he feels like a passenger on the Titanic, expecting the financial crunch of his life savings running out while still a long way from the end of the journey. He's made time to produce around 250 pages detailing the genesis of his ideas, the physical constraints of producing a robot on the cheap, an outline of his methods for reproducing neurology in software, and a discussion of some of the implications of advanced artificial intelligence and lifeforms. He does not offer us his code to review and as yet has not produced any technical papers to satisfy the curiosity of the professional reader. This book is an overview but one that provides plenty to chew on whatever your customary field of endeavor.
Making an intelligent android is not necessarily a hopelessly overreaching task. Steve believes the human brain uses "general purpose building blocks," each a variation on a basic design, rather than a spaghetti mass of all original wiring such as is found in simpler organisms. So when trying to divine the structure of the brain, it is, as Steve puts it, more like taking apart a lego house than trying to untangle a pile of Christmas tree lights. It could be tougher to model a worm.
But if seeing your brain as simpler than a worm's isn't worrying enough, how about having your whole sense of self undercut: "being of one mind does not imply that all the information passes through a single controlling structure." Steve has no time for the concept of a person sitting inside your head that is "you." In his view it is an illusion that there is either control or controller-- or even free will.
On the other hand, he does believe that emotion is essential for the development of intelligence. And that the very human ability to imagine is key to how the brain models and predicts the way the world will act and enables us to act upon it. We need it to match up our actions to the state of the world and bring it into line with our needs and desires. Two of the things that define us as human are pivotal to Steve's theories of brain structure and intellect.
Then there's the section on why it may be that we dream. Both the REM and the slow wave parts of sleep are explained by Steve's theories of how the brain wires itself up in the first place and then maintains its connections and infrastructure during sleep. His idea of a sort of mental test card signal that enables the wiring to set itself up originally and then reinforce itself later is useful, indeed vital when you realize that without this maintenance function our brains would, in his view, likely revert to mush. It also raises questions about what would happen in the sort of long sleep needed for extended space flights. According to Steve's theory we would have to keep dreaming or we wouldn't still be ourselves when we woke up.
Even if the entire project does not succeed there are the spin-offs: the new ideas about how our brains might work based on how he's making Lucy. Steve has to simplify (or at least ply Occam's razor enthusiastically) in order to cull things he can use from the mass of conflicting writings in neurophysiology. For example he thinks he knows how our visual system does a number of neat tricks. From using fuzzy images to increase visual acuity to extracting the visual essence of an object: a mental image with no rotational, positional, or size data attached to it. That may lead to breakthroughs in image recognition.
Steve theorizes that every cortical map must be thinking about something all the time. And if there are no signals demanding its attention then the map will generate some. Perhaps this is the explanation for the endless monologue that runs in everyone's head. And the visual day dreaming we do in vacant moments. Without these our brains would have to micromanage to keep busy or lose their connectivity as the circuits fade out from disuse.
At the stage where this book breaks off the saga of Lucy, she is a one-eyed, legless agglomeration of springs and servos perched on a desk full of computers. She can only grunt and on a good day point at a banana if you ask her to. Yet she is one of the most advanced research robots in existence. With so many breakthroughs in understanding how our brains work in phase one, I'm sure there are going to be plenty of people out there rooting for Steve to get enough funding to continue his work.
It would be excellent if a Brit could be awarded a MacArthur Fellowship. But for the time being Steve is subsisting on the dregs of a NESTA (the UK's National Endowment for Science, Technology and the Arts) grant to him as a 'Dreamtime Fellow' more on the artistic merits of his work than on its scientific promise. How weird is that?
You can learn more about Steve's work on his website. This book is available for now only through amazon.co.uk.