Work with AI in a smart way
Ian Cheng’s methodology, and how to tackle Artificial Intelligence in a smart way.
The method used by Ian Cheng that he describes in the interview “A portal to infinity” is of great help for the research I will intend to produce for this course.
My project has widely to do with Augmented Reality (architecture and the city as interfaces to AR), which ideally would be done involving AI, possibly as a further development for the final project. The way I.Cheng presents his research method clearly stating the limitations of access to high end technologies shows how alternative, more modest ways of engaging with research can lead to relevant and satisfying approaches. Instead of aiming at building a perfect AI for his game, that might be an ambition for big tech companies like Google, he chose to focus on building a collage of different models of the brain based on different references. The strength could come from the way those models come together, what is relevant in each field when borrowing concepts as much from philosophy than from video games?
One model is based on how the Sims build a model on in one hand slightly urgent behaviours, and a narrative model, of more immediate sense of the world.
The evolutions come from what agent is taking the control over the others at a particular time.
He explains how this became a strength of the project by bringing in innovative aspects to the experience : “When you are hungry you are also hangry, (...) this is true to nature, but hasn't been seen in video games or more academic studies of artificial intelligence”.
Fiction as a research tool
Another aspect that I found very interesting in his description of the work Emissary in the Squat of Gods is how he doesn’t have full control over the narration that he is writing. By setting up parameters and guidelines, rather than building a total world and narration, you might also be learning from what happens with the parameters you set, and when you modify them. This is a novel way of approaching the design of fiction, and might be a particularly relevant one when used in research.
It could be seen as simulations of fiction, with an infinite possibility of observations of possible human behaviours.
The aesthetics of the experiences.
The visual choices made in those games relates to the discussion that were brought about during the week on Virtual Reality. The aesthetics of the pieces is far from being photorealistic, and yet, to many extents, it might have more to do with the physical world properties. Each object is unique, I.Chang worked on procedural methods to have all the elements of his universe have a set of common traits, but never being an exact replica. The non deterministic aspect of things is also inspired by how nature works, as opposed to cities or architectures, that are millimetric, determined objects.