Learning to Engage in Interactive Digital Art
summary
A web version of The Plants is built with three randomized modes: single harp sound bank, predetermined rotating banks, and a learning agent using temporal difference learning with discrete actions/states (sound banks) and reward defined as number of touches per interval. Over 16 days with 178 participants (171 interactions logged) the RL agent raises average engagement by nearly 27% versus the other modes but does not increase interaction duration; the beatboxing sound bank emerges as the most preferred state.
AI-written summary · about
chapters (4)
- 0:00Start
- 1:11Temporal Difference Learning as the Reinforcement Learning Algorithm
- 1:39Reward for Learning
- 2:57User App Data and Survey Data
related talks
38mMaking Art with Artificial Intelligence: Artists in Conversation
20mLive Coding With Crowdsourced Sounds And A Virtual Agent Companion (second session)
20mIntelligent Agents and Networked Buttons Improve Free-Improvised Ensemble Music-Making on ...
20mLive Coding With Crowdsourced Sounds And A Virtual Agent Companion
26mInfinite AI
19mGenerative music, playful visualizations and where to find them
20mThe rules are the art: creating with AI
08mNo merge conflicts