Putting the life back into technology

Forget Artificial Intelligence: the future of intelligent entities lies in synthetic vitality. Enter Artificial Life. Digital master Steve Grand waxes techno-philosophical as he unveils his plans to put the life back into technology.

  • E-Mail
By  Kate Concannon Published  October 8, 2001

Introduction|~||~||~|The public mind has been fascinated for generations by the prospect of artificially intelligent entities. Little known, however, is an extraordinary field of research and development that has emerged in recent times; a science that holds all the mystery and magic of the creation of artificial ‘things’. Better than mere, albeit remarkably ‘clever’, machines, this field strives to deliver results that can rightfully claim to have attained consciousness, ‘intelligence’ and actual life.

Artificial Life is an interdisciplinary science that brings together biology, chemistry, physics, computer science and engineering in its quest to realise the ultimate human dream: to imbue artificial creatures with the vital essence. And the results are surprisingly promising.

The distinction between AI and A-Life is far deeper than first meets the eye: their approaches, ambitions and means are at considerable remove. Whilst both use computer science to devise computationally implemented models, AI is concerned with analysing cognitive processes, such as logic and reasoning, to create machines that can think in their own right.

A-Life, however strives to create actual, if initially virtual, living systems. In order to bring about artificially the processes characteristic of living systems — namely learning, adaptation, evolution, reproduction, metabolisation and self-organisation — A-Life employs a parallel bottom-down approach that further divides these spheres of artificial pursuit.

Steve Grand is a computer programmer by trade, who came into the public eye through his work as chief technologist on the computer game, Creatures. Again he was thrown into the limelight with the launch of his extraordinary book “Creation: Life and How to Make It” — a manifesto of sorts describing his theories on life and the possibilities (and methodologies) for its synthetic development.

Since Creatures, Grand has turned his ‘jack of all trades’ understanding of various spheres of scientific endeavour and computing fait-accompli to exploration of the possibilities of A-Life, and has started his own R&D company, Cyberlife Research Laboratories (CRL).

Disillusioned with the top-down approach embraced by AI — which is yet to yield truly compelling results — he is tackling the issue of creating A-Life (and so AI) by modelling living biological systems through complex algorithms. The former approach, according to Steve Grand, produces at best a simulation of the outward appearance of intelligence — a parroting of sorts — whereas the latter allows systems to arise and behaviours to emerge in their own right from various life-modelled structures, the result being an evolving, robust and intelligent being: “The trick is to eat the meal, not the recipe.”

||**||A-Life Approach|~||~||~|In real, practical terms, this bottom-up strategy means devising artificial systems that resemble those found in organic life-forms. The evolution of organisms and their functions seem to have arisen from very basic building blocks, and the key to development of these systems, according to Grand, stems from feedback loops.

Feedback loops are either positive or negative (definitions which carry no corresponding connotations in this field). A negative feedback loop is a system that tends to discourage change, and, rather, encourages maintenance of the status quo; positive feedback loops, on the other hand, strive to reinforce a change.

The basic building blocks essentially consist of neurones, chemoreceptors, and electroreceptors. These form the basic platform for our own functioning, and that of fellow higher organisms. Grand’s project is to create an entity that, equipped with artificial building blocks, will learn and evolve of its own accord (in response to artificially created ‘drives’ such as hunger and propagation), resulting in a ‘being’, possessing characteristics that, when found together, define life.

Intelligence too is a concept difficult to pin down. Grand feels that our mistaken exclusion of capabilities such as emotion and imagination from the intelligence equation — where more ‘solid’ features of mental processes such as logic sit uncontested by AI tradition — may just be the very thing keeping us from achieving artificial intelligence.

What makes behaviour intelligent is its adaptability and its use of imagination. The two are crucial to an organism’s survival, as they allow it to choose appropriate action in the face of variable circumstances, and provide, at a basic level, the ability to both foresee consequences and empathise — and thus act accordingly. Without these skills, a human — and, too, any AI machine — would be sorely lacking, autistic even in its interaction with the surrounding environment. A holistic approach thus comes to be of inevitable and paramount importance to A-Life and AI’s success.

In essence, Grand aims to create from software robust systems that resemble those that gave way to evolution of intelligent organisms, humans among them. The first real (rather than virtual) ‘being’ involved in this endeavour takes the form of Lucy Matilda, the orang-utan baby. It is hoped that she will be the first leap in developing sophisticated and intelligent artificial brains that will, eventually, inhabit robots and other ‘bodies’ in the real world.

In effect, Lucy will be alive and real, just as we are alive and real (despite the fact that the events occurring to make us real, living, persistent phenomena can ultimately be reduced to basic electrical and chemical reactions and interactions — a confronting prospect to some). Digital master Grand takes a very human approach to his work, and it may just be this bottom-up, human strategy that proves the key to unlocking the elusive dream of synthetic living beings.

||**||Meet Steve Grand |~||~||~|Introduce us to Lucy, in the context of CRL's mission.

Lucy is a research platform designed to help us develop our ideas about how the human brain may work and, more specifically, how to make artificial brains that will allow future machines to be more intelligent, robust, flexible and user-friendly. She is the embodiment of our long-term research programme to develop an artificial life form with a mind of her own.

Our ultimate aim is to help Lucy work her way through nursery school, learning to coordinate her muscles, to form spoken words and eventually to paint pictures. Lucy will have a radically new kind of artificial brain. She will not be as smart as a human or ape baby of the same age, but she will learn for herself and she will have something that no robot has ever had before — an imagination.

In the meantime, you can expect to see Cyberlife technology turning up inside simpler, everyday devices: consumer equipment, PCs, vehicles, artificial limbs. In every case, we aim to include the whole creature — its brain, nervous system, drives and emotions — because we believe in thinking holistically. After all, there’s just no such thing as half an organism. Try cutting one in half and you’ll see what we mean...

Creatures has become a hugely successful computer game. What methodologies, applied in Creatures, will carry over to Lucy, as a real world entity analogous — but far more intelligent and sophisticated in her behaviour — to the virtual Norns?

Creatures taught me how NOT to create artificial life, in a sense. All the central principles that I wrote about in “Creation” still apply, but I was very disappointed with the specific way I’d built their brains. It showed me what was wrong with a lot of current theory and with my own naïve understanding of intelligence. Lucy is my attempt to put that right, and this time I don’t have to do it in a commercial environment, tied to the needs of a product.

Nonetheless, in Creatures the computer is not programmed to behave like a living thing at all — it is programmed to behave like neurons and chemicals. The phrase ‘programmed to behave as it were an intelligent creature’ is a contradiction in terms: anything whose behaviour is explicitly pre-programmed cannot be considered intelligent.
From a huge network of these interacting virtual chemicals and nerve cells ‘arises’ lifelike and intelligent behaviour.

||**||Interview cont...|~||~||~|The Norns are not just lines of computer code, they are networks of nerves and chemical pathways. Just because the nerves and chemicals are fake doesn't mean the creatures are. The main reason Lucy’s a physical robot is simply that intelligence requires a complex environment. To be intelligent you have to have something to be intelligent about — imagine what would have happened to your brain if you were born and brought up in an isolation tank.

Creating complex virtual environments gets harder and harder as the complexity rises, and, on the other hand, in a virtual world it’s easy to bypass the really hard problems, like vision. I wanted to avoid the temptation to cheat.

It’s true that people will be far more willing to accept Lucy as being ‘real’ because she
has a physical body, despite the fact that all her brain and everthing else that really matters will still be virtual. Try as I might, I can’t get people to let go of their misplaced obsession with matter!

If your aim is to have Lucy learn how to use her own skills, to attain empathy and imagination, what is your proposed method?

In a nutshell, the idea is to develop an artificial brain made from virtual nerve cells, connected to each other in such a way that they embody the capacity to imagine. Almost everything we care about — our dreams, our fears, our hopes, even caring itself — stems from our ability to imagine things.

Thanks to the technicolour virtual world we all hold inside our heads we are able to form hypotheses, make predictions, plan a course of action, rehearse contingencies and realise our intentions. Imagination is so crucial to human (and primate, and probably at least most mammalian) intelligence that it begs to be explored.

It’s not some airy-fairy capacity that arises spontaneously whenever you bung a few million neurons together in a heap — there must be a specific neural mechanism that gives us this ability, and the things that we imagine must be represented somewhere in our brains.

I’ve been thinking very hard about how and where imagination arises, and how it helps us to survive, and I have some hunches that I hope will enable me to reproduce it inside Lucy. If I’m right, then Lucy will have the proper tools inside her head from the moment she’s born. But like a human baby she won’t know how to use them. The “software” will be in place but it will lack the “data” to make it work properly.

So, like any baby, Lucy will have to develop and learn for herself, starting with the crucial discoveries that we all have to make, like where our bodies start and end, and how to work our arms properly.

We eventually go on to learn to dance and to play chess by building on these discoveries, but Lucy’s not going to get that far.

Do you plan to have Lucy grow up in your own nest, so to speak?

All being well (if I can find a way to pay for all this!), there will be a whole series of ‘Lucys’, as my ideas develop and I need new hardware and software in order to explore them.

But in my head there will only be one Lucy, who will develop and grow, shedding her skin from time to time but remaining the same ‘person’ throughout. (We humans are no different, when you think about it — I’m hardly wearing the same body that I had when I was four!).

So to me, my wife and my son she’s already like a rather frail and dependent member of the family. In my dreams, I hope eventually to send her to nursery school; maybe one day she’ll fly the nest and make me proud.



Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code