Automated Reasoning and Adaptation for Assisted Experience Design

点击翻译按钮获取中文摘要

Many of our daily experiences are designed by one or more people who have particular goals in mind. These designed experiences could include visiting a museum or city park, following a fitness plan, playing a sport or a video game, and more. Sometimes designers aim to meet people's needs, like recovering from an injury, relaxing, or learning a new set of skills. A valuable experience is one that does what it was designed to in a reasonable amount of time, and without any unwanted side-effects. For example, a valuable training experience might fully train a new employee during their first week, without making them want to quit. While many practicing designers work to make our experiences more valuable (that is, better or faster at doing what they were designed to do), their time is limited, and the number of ways that an experience could be made better is enormous. The goal of this research is to study how practicing experience designers might work together with Artificial Intelligence (AI) systems to make more valuable experiences. One current example of an AI-based system is the writing software Grammarly, which supports a writer as they co-design a text, helping to create a more valuable reading experience. We want to understand how other kinds of experience design, besides writing, can be supported by AI systems. To gain this understanding, we will work with different kinds of practicing designers (like museum curators or game designers) to learn about their design processes and translate them into a machine-understandable form. We will also ask these designers about how their current processes might be improved, and for the most impactful (but still solveable) problems, we will build and test AI systems that help solve them. We will also consider the challenge of managing privacy, since people should generally have control over how and when their personal information is used to build or modify their experiences. We want our AI systems to support designers in making just decisions about how personal data will be used. For Canadians, this work is important because many kinds of common experiences could be improved: imagine having more engaging fitness plans, more effective training programs, or more exciting games. For researchers of AI systems, this work represents a key step toward more general AI systems, since reasoning about experiences is an important part of functioning in the world. Finally, this work will show an important way to make progress in AI research: every step forward will be made toward a goal of benefiting society, with tangible improvements to how experiences are designed.