Blog Post

Interface Seminar: Neuroartist Extraordinaire

Bill Seaman is the head of Digital Media at RISD, and he's luckily coming over at the invitation of Kristine Stiles to speak to the Interface Seminar. Seaman's main interests lie along the intersection of digital media, electrochemical computing, and artificial intelligence: an eclectic and productive combination. While it is beyond my abilities to judge the aesthetic value of his art, what is clear from the onset is a series of very serious academic ideas drives his creative urges. In fact, he does not seem to view his art qua art, but as part of an interdisciplinary effort to create artificial intelligence in a wholly new form, the "Benevolence Engine."

First, Seaman is most well-known for his "Recombinant Poetics" and "Database Aesthetics", where he clarifies the nature of art in the digital age as one of Deleuzian recombination, touching in the aesthetic realm many of the same concepts that Ian Boghost arrives at through his study of gaming in "Unit Operations." In his "Recombinant Poetics", Seamna states "human processes become intermingled on the highest level with machine functionality in the service of human expression" and so this Deleuzian assemblage is the truly defining mark of art and media in our time. This does not dispose of the human body however, as "An embodied approach to computing acknowledges the importance of the physicality of experience as it falls within the continuum that bridges the physical with the digital" yet combines our sensibilities with the abilities of machines to produce ever new media combinations so that "database aesthetics put the poetic nature of composition, media configuration, sequence...in line with the constraint-based nature of combinatorics."

However, in his latest project Seaman, with his co-conspirator the famous physicist Rossler, is to create "Neosentience" - in other words a "sentient robotic entity to be a system that could exhibit the following functionalities: It learns; It intelligently navigates; It interacts via natural language; It generates simulations of behavior (it thinks about potential behaviors) before acting in physical space; It is creative in some manner; it comes to have a deep situated knowledge of context through multi-modal sensing; It displays mirror competence" No small order, and furthermore, Seaman's real contribution is his unique perspective on the simulation of the social part of intelligence, since the operations of each machine "can be stopped in an attempt to simulate in favor of the other's goals. In this case, the invention of an hallucinated other center of optimization occurs. This is the invention of benevolence." He goes into more detail in his paper on "Electrochemical Computing" where learning lessons for dynamical systems and (I assume) artificial life, he defines a "Thoughtbody" as "a mind/body unity built up through a lifetime of reciprocal forming/framing processes. He believes that an Engelbart-style bootstrapping of intelligence of a "thoughtbody" by designing the system such that language and memory interacts with even the most basic perceptions, and vice-versa: "Thought is embodied, that is, the structures used to put together our conceptual systems grow out of bodily experience and make sense in terms of it; moreover, the core of conceptual system is directly grounded in perception, body movement, and experience of a physical and social charcter."

Overall, I have sympathy for Seaman's project, and during his time at MIT he must have at least known Minsky since he seems to have definitely become enraptured by the silicon dream of AI. Yet, I have my doubts about embodiment being the "band-aid" needed to bootstrap AI. Seaman notes that his take on things is a "quite different perspective to earlier AI projects that were not embodied and did not see the importance of coming to a deep knowledge of context via multi-modal sensing systems that would be dynamically linked to their environment." However, on engineering grounds alone the technical feasibility of such a project is difficult at best. Lastly, I don't know if embodiment is really a silver bullet. After all, the real interesting things in human culture and art seem to be, while influenced by the body, not embodiment per se, and I wonder how much of this embodiment is merely a return to a sort of philosophical individualism ala Strawson (or worse, the transcendental individualism of Husserl). Furthermore, just as classic AI mistook the techniques of logic for intelligence, I also believe dynamicists and ALife seem to have mistaken the methodology of differential equations and complex systems for the mind itself, and as any epistemologist worth her salt should know, the gulf between methodology and subject matter is as wide as it is deep. While I used to be a "true believer" in embodiment, I'm more a believer now in rough consensus and running code, and while Kismet is cute, it's not even remotely intelligent. Embodiment is the baseline assumption of course we have, and the real interesting things happen embodiment - as Seaman hints at with his concept of "benevolence" only wish had written more about it. The actual advances made by dynamical systems and embodied robotics seem rather small, still sort of stuck on the "earwig" level. Where the action's at seems more to be the digital world outside our bodies and the increasingly fluid interface between our bodies and this world.

However, if anyone has the vision for making embodied artificial intelligence work, Seaman does...and his partner Rossler has impeccable credentials in making working physical systems. In fact, Seaman's realization about intelligence being embodied not just in material bodies but in electrochemical configurations is a real insight, one rarely acknowledged in AI and even in cognitive science. Breaking outside the box of traditional thinking is exactly what Seaman, and artists in general, could do for cognitive science.

75

No comments