Date: April 19st (Friday), 4:00 PM (Warsaw time)
Join us online via Zoom: https://uw-edu-pl.zoom.us/my/hill.meetings
Co-Creative Art Making with the AI Drawing Partner: A Quantified Approach to Model Co-Creative Experiences
Abstract
Co-creative artificial intelligence (AI) is a field that studies and designs co-creative agents that collaborate with the user on a shared creative product or performance. Studying co-creativity can involve studying the person, process, or product. This talk focuses on techniques to quantify the process of co-creation, which has typically been challenging to quantify due to the open-ended and dynamic nature of co-creativity. The approach taken here is to utilize the co-creative system as a data collection platform to log all the interaction data between the user and agent, and visualize that data to reveal interaction patterns and trends. The AI Drawing Partner is an example of a system that not only collaborates with the user, but codes the user and agent’s interactions, logs action history, models collaboration dynamics, and visualizes these elements for the user.
The AI Drawing Partner is a co-creative drawing agent that analyzes the user’s line input and responds with a contribution of its own on a shared canvas. The user and agent engage in a turn-taking interaction, where meaning grows dynamically through time. The user draws a series of lines, and when they are done, the system analyzes those lines, tries to recognize any objects the user might have drawn, and then decides how to respond. There are different methods the system can use to respond: reactive algorithms that transform the user’s input (e.g. rotation, scaling, translation), and generative algorithms that sketch objects or generate images based on a text prompt. The AI Drawing Partner is also a quantified co-creative system in the sense that it quantifies, models, and visualizes the co-creation happening on the platform. The system uses the co-creative sense-making cognitive framework to structure the data collection processes. The framework has four main categories: cognitive dynamics, interaction dynamics, collaboration dynamics, and domain dynamics. The AI Drawing Partner quantifies elements from each of these categories to produce a model of co-creation that can be used to compare the interaction of the user to the agent, to compare between participants of a user study, and to make cross-domain comparisons of the co-creative process. The talk will include an artistic perspective of interacting with the AI Drawing Partner, data generated from the system, and a demonstration of the functionalities of the system.
This talk is a part of the Traincrease Lecture Series (D4.2).
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 952324.
