Exploring Methods for Holistically Improving Drawing Ability With Artificial Intelligence

Abstract

Drawing is a highly useful skill that can make people better at solving problems, communicating ideas to others, collaborating, and producing more creative and novel ideas. It can be a difficult skill to master for many people, however. Like any learned skill, it requires many hours of practice for noticeable improvement, and sufficient motivation is also necessary to keep practicing consistently over a period of time.

Utilizing sketch recognition and other forms of artificial intelligence to assist in learning to draw may facilitate the necessary improvements in self-efficacy and motivation students need to improve their drawing ability. While similar tools have been explored, there has been little to no effort at designing a truly holistic approach for teaching drawing skills that includes the basic fundamentals and building blocks for drawing any 3-dimensional object.

This dissertation explored the potential of an intelligent tutoring system for teaching drawing skills called SketchTivity along with various other technology probes focused on drawing. We found evidence that individuals could build confidence, build motivation, make measurable improvements to drawing ability, and reduce fixation when ideating concepts through the various studies we conducted. We developed a flexible perspective accuracy recognition algorithm that can help individuals learn perspective. In interviews with students and teachers who used SketchTivity we discovered benefits and limitations of the system. Students were engaged by the interactive lessons, motivated by the gameplay, and saw it as a great warm-up tool. Meanwhile instructors loved that the system could offload grading tasks for them.

We hope the nuances of this potential will inform the future development and promise of the approaches described in this dissertation along with similar approaches to impact education at large.