Toward Computational Play

Advisor: Terry Knight
Duration: 2015 – 2018 (master’s thesis)

In my current research, I envision computational models that exhibit the same range of creative behaviors and abilities as designers. The problem-solving techniques of design thinking – which often escape traditional models of artificial intelligence – hold great potential for the development of creative computational systems. Design thinking is playful, but we have yet to develop a sufficient notion of how that can be applied to computation. For my master’s thesis, I am studying the topic of computational play as a way to research how and why designers roam as freely as they do, what the creative potential is of such exploration, and how such techniques might responsibly be implemented alongside traditional artificial intelligence methodologies. In order to complement and challenge my theoretical inquiry, I am implementing my proposals in an autonomous, playful drawing machine.

This machine uses an iterative cycle of physical plotting, machine vision, and computational processing to proceed through a drawing.

The machine is equipped with a dual draw/erase pen mechanism, enabling the construction of complex, layered drawings.

The drawing procedures are based on the drawing rules at the core of Shape Grammars.

The drawing machine embodies the four core characteristics that comprise my proposal for computational play: It is multimodal, generative, iterative, and autotelic. In my thesis I explore these topics in detail and explain why they are important, focusing especially on the topic of autotelism and the role it plays in design thinking.

Autotelism allows the subject to play simply for the sake of playing. It taps into both the objective and subjective qualities of the player‐at‐play. On the one hand, we have the objective number of possible drawing options available to the machine; on the other hand, we have the machine’s interest level in the activity. Treating these two conditions as independent variables and using them to construct a kind of autotelic decision‐making graph, we open the possibility for some of the interesting behaviors that underlie playful design – behaviors that are unpredictable, yet entirely attributable.