Fred Trucks' ArtEngine

I have recendy had the opportunity to test drive Fred Trucks'
Art Engine, a robot that makes art (sic) The work is in thex form of a sizable piece of software written for the MACII computer. It concentrates on modelling the interaction of visual and associative memory functions in creative thought. To do this it utilises artificial intelligence techniques and aspects ofexpert systems design. It is not menu driven, but attempts a more complex interactivity: the operator can input data.

ArtEngine is not one of those familiar projects which simulates conventional forms or is formulated as a tool to aid the simulation of conventional forms on a video surface. These have generally been designed by technicians with a naive understanding of art practice, and the products have reflected that naivety, they tend to be somewhat mechanistic. In this category I place almost all the ‘user friendly’ graphics packages. These are of course subject to the lowest common denominator enforcement of the restriction ‘user friendly’ which is a major impediment to their complexity. One drives these programs in the same sense that one drives a car: its’ structure is set. There can be little argument that truly creative work with computers can only be achieved through custom programming. (These are the 'Hot Rods' and ‘Funny Cars’ of the software world.) There is no such thing as a general purpose machine, or to put it another way, the more generally useful it is. I am reminded of Harold Cohens distinction between the 'expertsystem’ and his ‘experts’ system’.

This is the debate into which the ArtEngine fits, and Truck is eminently qualified to join in. Fred Truck is one of the new breed, one of the very few qualified to explore the art-making potential of the computer beyond the 'automated Etch-a-Sketch’ format. Old enough to have seen Jasia Riechardts’ landmark exhibition Cybernetic Serendipity in 1969, and smart enough to have been lastingly impressed, Truck can discourse as eloquently on programming languages, communications theory and Boolean algebra as on the function of the self portrait from Titian to Warhol. Half of the manual to the work is an engaging interdisciplinary text the first section of which is titled ArtEngine(a symbolic machine): Portrait of the artist as an information processor.

What then is the ArtEngine? It has a dual identity as an artwork, and as a tool for making art. It is on these two criteria that we should examine it. We should not expect exclusive logic to apply; in this new medium it may possible to simultaneously be an artwork and a tool. As Truck explains in the manual, one function, in terms of the program, is the other one backwards.

The dual identity of the ArtEngine is an example of the level of conceptual punning that is built into the structure of the work. Truck mentions early on in the manual, that the operator (player?, artist?, interacter?, there is a language lag here) will find more aesthetic value in the implications of the program and the concepts embodied, than in the graphical and textual output. While this is faintly disappointing at the outset, and sounds like a cop-out for shoddy work, it does ground the viewer (?) in post-object aesthetics as a critical vantage point. This is singularly appropriate to software artwork which, by its nature, must be celebral, multidimensional and concerned with disembodied conceptual engineering.

The ArtEngine adopts a position midway between a user-friendly (read: foolproof) application and a complex programming task. It is easy to see the rationale for this. A user-friendly front end (HyperCard) serves to reassure the faint hearted. One hits the programming level soon enough. The ‘user friendly front end’ is a pragmatic strategy for easing the interlocutor into a complex meta-topography of ideas, a fluid universe Truck has designed as an interactive artwork. However, in order to derive the pleasure of fluent complex interactivity, one needs to be familiar with such arcane corners of the Mac system as the Quickdraw algorithms. Some familiarity with the ‘scheme’ programming language is also very useful.

None of us expect to be able to drive a new piece of software just by loading it, yet we expect that art works are that accessible. We forget that we have a lifetime of cultural education behind us that make that artwork accesssible, we’ve read the manual. The horns of the software art dilemma are the horns of the software dilemma as a whole: there is no such thing as a truly powerful user friendly application. Thus: any new piece of software demands a learning period like any other skill, an investment of time, in order to exploit it fully.

The ArtEngine purports to make art, it purports to be modelled on Truck’ own strategies for artmaking. The particular strategy embodied in ArtEngine is basically a dialectical process of deriving a synthesis of two juxtaposed sets of ideas. The program is so constructed that the ‘sets of ideas’ manifest as graphic information and text simultaneously, at the program level the same information gives rise to both. One creates the two ideas and the machine divides and multiplies, derives a synthesis of the two.

If this is the artistic process at work then it is once again faintly disappointing. For if the ArtEngine thinks that artmaking is simply a process of mixing and matching then it confirms our suspicions that artmaking is a very complex process and machines have along way to go to emulate it.

Artmaking in the late 20th century western society may be the extreme case of synthetic (as opposed to analytical) intellectual activity. It is a modernist cliche, and the modus operandi of avant-gardism, that art continually ‘expands the boundaries', ‘transgresses traditional forms' etc. It is exactly this that the computer is currently unable to do, because they are confined to a rule-bound world. This is a disability of machines, it is generally assumed that the human intellect is not subject to such limitations. Further, artmaking is an extremely ‘self conscious’ activity. The artist has some conception of what s/he is doing, and why. As yet, machines do not possess that type of ‘self awareness’ (see Hubert Dreyfus What Computers Can't Do (revised edition) Harper Colophon Books, 1979).

As an inventor of the art therefore, the ArtEngine stumbles on this block: it is confined to its rule-bound world. Truck is aware of this; in the manual, with respect to HyperCard, he says: ArtEngine is not nearly so freeform. It offers a particular, highly associative structure for information that, when adhered to, gives ArtEngine the potential to behave intelligently. In its defence, it should be remembered that the level of intelligent behaviour that a MACII is capable of, in terms of a loose neurological comparison, is about that of a worm or sea slug. That’s pretty creative for a sea slug. I’m sure that Truck would relish working on a similar project in a Cray Y-MP. Thus further limitations are externally imposed, by the technology as well as its socioeconomic context.

Pioneers in this new medium take on several large charges.
The technology is too primitive to allow user friendly realization of the project. A new audience must be educated to understand these limitations and be prepared to engage the skeleton of a work, at expense of some effort.

The potential for computer based interactive artwork is limitless. Other workers in the area have experimented with a variety of possibilities. The multi-narrative or hypertext has a group of explores. Robert Edgars' work Memory Theatre One experiments with the externalisation of a meta-geography as a facilitator of memory. In this he draws on the historical research of Dame Frances Yates and also the recent AI writings of Marvin Minsky. Truck has headed straight for the motherlode: to model the creative process itself. An impossibly ambitious project, as he implicitly acknowledges.

This is research work and the goals are visionary. I End myself frustrated by the work itself, but have only admiration for the enterprise. ArtEngine is a expedition into a territory where no art has gone before.