牛牛资源

Artist and computer scientist Professor William Latham recently gave a TEDx talk on evolutionary art and computers to an audience in Oxford of 1,600.

Read on or watch the video below to find out how his work - initially inspired by the exhibits in the Natural History Museum - has developed over the decades. What can we expect to see in years to come as the relationship between science and art continues to develop?

鈥淛umping back in time to about 1982, I was a student at the Royal College of Art. Whilst all my friends were producing vigorous paintings in the style of Baselitz and Julian Schnabel I was spending a lot of time in the Natural History Museum observing natural forms, such as snake skeletons and butterfly eggs. The natural world is fantastically rich from an artistic point of view,鈥� Professor Latham explains.

鈥淎fter spending much time in the museum I started to devise my own evolutionary rules, these were simple rules that would tell me as the artist what to draw and I used these rules to produce huge evolutionary charts. Some of these charts were up to 30 feet long. These enormous drawings showed evolutionary variants becoming more complex with each generation. I鈥檇 created my own natural system 鈥� a humanised version of what I saw in the Natural History Museum."

The work was influenced by fractal maths and by Froebel shape grammars, used by architects and mathematicians at that time. Latham also took influence from American graffiti artists such as Keith Haring, and from Tantric art.

鈥淚n my work I鈥檝e looked to the Far East for inspiration rather than contemporary European or American art. As these drawings grew in time, the work coincided with Richard Dawkins鈥� brilliant work on the Blind Watchmaker, before he got bit side-tracked. I then became a Research Fellow at IBM. I went from art school to corporation, and found myself working in a computer laboratory using mainframe computers and started a long-term collaboration with mathematician Stephen Todd.鈥�

This was an absolutely fundamental phase in Latham鈥檚 work: 鈥淲e had software that could join things together, render them in 3D, and quite quickly we threw away the old evolutionary rules and developed a new set of rules that produced things that looked like animal horns, suggesting strange things that you might find them at the bottom of the sea or strange eggs.

"Fundamentally there was an underlying grammar, and the rules we designed determined the way forms would grow in time. Then our next novel step was to build an evolutionary fruit machine which would take all the numbers, that drive the rules, and randomly change them a little bit and generated hundreds of variant forms. The idea here was that the artist becomes a gardener, interacting with the computer, steering through a vast evolutionary space of possibilities.鈥�

The interface Latham and Todd developed could cross breed and breed between multiple 鈥榩arents鈥� to evolve extraordinary new forms, in effect an evolution driven entirely by human aesthetics.

鈥淚t鈥檚 a bit like the Rorschach Ink Blot Test. The viewer perceives what they want in the forms, because the forms have no meaning of their own."

"In effect, by viewing these variants one is sort of exploring a psychological landscape. When I was breeding forms I鈥檇 always pick anything that looked a little bit Rococo, a little bit Baroque, anything that looked like a Giger painting. My aesthetic was steering the evolution. Things that looked a little bit paisley, Greek helmets, pumpkins, viruses and other things that appealed to me. These were all evolutionary selection criteria that I was using to pick and breed.

鈥淭he world of plants had a massive influence on what I would pick, but also heavy metal imagery. I don鈥檛 like heavy metal music at all but the imagery is fantastic so anything that has a kind of heavy metal feel, I鈥檇 breed from that one definitely or maybe put it aside to breed from later, or cross-breed with a plant-like form.

鈥淪ome of the forms looked like they could be from an alien planet, they鈥檙e 3D, you can spin them around, and they鈥檙e continually evolving, subtly changing shape.鈥�

More recently he鈥檚 been working on a system where the viewer also takes part in the artwork.

鈥淭he idea is that in a gallery someone can pick and breed the forms themselves, it doesn鈥檛 require my aesthetic, someone can bring their own aesthetic in to the gallery and that will influence the type of forms that they will evolve. The idea is that the artist is not sacrosanct, and the public can actively take part in the works creation. We鈥檙e starting to do early work with virtual reality to take this into the VR space. People will be absolutely immersed, surrounded by these 3D forms. Even subtle body movements will change the dynamic of the evolutionary space they鈥檙e within.

鈥淎nother question I鈥檝e been dealing with recently with my collaborators is, what happens if you completely remove the human from the system? The machine has rules which decide which ones to breed from. A set of variants carry out calculations using a set of aesthetic filters.

鈥淭he challenge with this type of approach is that it鈥檚 fine if you can mathematically define what the rules are. So artists would often be concerned with such things as balance, size and composition. Because these are mathematical, you can programme them relatively easily. The computer can calculate the numbers. The machine generates the variants, picks the winner, and breeds new variants.鈥�

What Latham and Prof Frederic Leymarie at 牛牛资源 and their collaborators have found is that what computers cannot do is recognise forms that have got rich content to the human eye, like griffins or witches鈥� faces. A computer won鈥檛 be able to spot and 鈥渞eward鈥� forms that look to us like surrealist art, an alien spaceship or a bizarre lizard.

鈥淢aybe this is the area where AI and computer vision really needs to make a breakthrough. It seems you can computerise a certain amount but still the human does have a role. You can鈥檛 completely remove the artist or the public from the artwork."

Latham has spent a long time working in laboratories with scientists and more recently, as a professor at 牛牛资源, he鈥檚 begun a collaboration with Professor Mike Sternberg, Head of the Bioinformatics Group at Imperial College.

鈥淥ne of Mike鈥檚 PhD students had been a big fan of my work. A lot of the structures that we鈥檙e using resemble the way that proteins are bolted together. So this sort of crazy art system for creating evolutionary art forms does have some similarities to the natural world. For example, beta barrel -protein structures have some of visual and geometrical similarity to my horn-web forms.

鈥淚deally what we鈥檙e hoping for is a return to Renaissance art, with artists and mathematicians working together with programmers, engineers, experts in AI, computer vision. This to me is the direction art should take.鈥�

"We have art reduced to a form of real estate"

But what do we have right now? Latham says 鈥渨e have art reduced to a form of real estate鈥�, driven by auctions and the gallery system 鈥渨hich has at its core a philosophy which says 鈥業f something sells, it鈥檚 good鈥�. It鈥檚 about time this system was removed and we finally got back to our roots and started to be truly creative again鈥�.

鈥淭he role of the artist in the laboratory is very interesting. There鈥檚 a lot of pressure on scientists to be creative, for example the use of statistics has generated a lot of good scientific research rewards but arguably the barrel鈥檚 kind of empty right now.

鈥淥ne of the things I find very interesting when I work with scientists in laboratories, whether it鈥檚 at IBM or Imperial College, when you tag something as art it can be very, very, experimental because you鈥檙e not hindered by the scientific method.  So having an artist in the team, with anything that鈥檚 just a bit too experimental, you call it art for a little while, experiment freely and translate back into the science domain later. It鈥檚 a really good approach, and one of the benefits of the artist being in a scientific context.

鈥淲hat is the big picture here? If we鈥檙e going to redesign nature, which looks like it鈥檚 pretty much on the cards, even if takes a little bit longer than expected, it looks like ethics and aesthetics will play a major key role in that redesign. If artists are going to work with scientists, they already have to be engaged with software and engaged with some of the tools that scientists are using to be able to inject that aesthetic element into their redesign work. They already have to be in the lab.鈥�