Explorations > The Art-Machine

The Art-Machine

The Art-Machine (1990 – 2000)

I created The Art-machine while still at college in the late 80’s / early 90’s.  At that time I wanted to make works that reflected the complex and fragmented world I was living in but I just couldn’t find a suitable form or style that seemed dynamic enough. I was also studying on a particularly theoretical course, one that constantly questioned meaning, authorship and the effectiveness of any artistic expression.

Frustrated by these problems I set about trying to create some kind of system for generating art instructions. I had the idea of using techniques I’d developed when programming my ZX-Spectrum as a child. I would create a machine that could generate proposals similar to those of the conceptual artists I admired at the time such as Sol Lewitt. I would then physically paint, sculpt or perform the proposals with my own hand exactly as instructed.

I made the decision that I would only exhibit the manifested pieces and not the machine itself, which ultimately led to some debate over the machine’s existence, but despite this I felt it was necessary in order to give the Art-Machine’s results the status of independent artworks. The result of this 10 year experiment was a large collection of paintings, sculptures and performances that filtered out my own personal tastes, preferences and history.

The Art-machine was a real computational mechanism;  a set of algorithms, flowcharts and computer programs rather than a physical stand-alone machine like a sci-fi computer or a robot. It used random generators, existing grammatical structures and recursive networks to give instructions. I called each run through the system an “Art-Machine Iteration.”

This is how the process of generating an Iteration went.

Firstly I would consult the CPU. A massive flowchart of sub-charts that had a sequential set of variables that had to be defined. The first variable to be defined would be whether it would be a single or multiple run through the Art-machine algorithm.  These decisions would be defined by various programs i’d written for each task.

It would then to go on to decide whether a work would be a painting, sculpture, film, performance or some combination of these things. Next came the content, iconography, materials, size etc.


Flowchart showing the Art-machine’s 2 dimensional CPU in its centre


The programming was done in the language Prolog which was particularly good for handling lists and nested hierarchies. Once each of these variables had been established I would be left with a set of data that would be turned into a written iteration sheet

100 Artmachine Iteration Sheets 1994



Quickly It became apparent that having pre-defined databases from which to select variables was extremely limiting. I decided that I would try to use the entire world’s information structures as a database and find techniques for randomly locating the information required in each proposal. There were no search engines back then so I did this using timed interactions with live media or randomised selections from reference libraries and books.


Candlestick Maker (Applied Artmachine Decision Tree) Mixed media on Paper – Arts Council Collection of Great Britain


As I worked on completing the machine’s often bizarre instructions I wanted it to be more than just complicated randomising device. I began to use language structures to create more meaningful compositions. These were still unpredictable and filtered out my own preferences but bred together existing ideas to create new synthesised forms.





Art-Machine Iteration: Junction Box – Private collection New York

The collection of objects, images and events created by the Art-Machine were extremely diverse but nevertheless shared some common ground. They all had passed through the portal of the Art-Machine. From the fundaments of cosmological and biological evolution through to art historical precedents, technological invention and the decisions of the Machine itself, an infinite field of possibilities had been filtered into an unlikely but specific form.


Comments are closed.