About the course
As future innovators of the Artificial Intelligence (AI) frontier, we must be intentional about the impact of our creations (and if I may be a bit meta: our creations’ creations!) on our communities, our families, and ourselves. AI is only useful in how it positively augments the human experience. In this course, we explore how to harness the power of AI for the benefit of our world, i.e. to push technology towards human Intelligence Augmentation (IA).
While Human-Computer Interaction (HCI) researchers have long discussed and debated what intelligence augmentation (IA) should look like, this course will look at how those debates have and will continue to translate to concrete, immediate operational goals/tasks for AI researchers. Tasks play a vital role in the AI and machine learning community; some tasks ultimately channel the efforts of thousands of AI researchers and set the direction of progress for years to come. Some popular tasks over the years have been inspired by HCI, social science, and cognitive science literature: object recognition, scene understanding, explainable AI, interactive robot training, etc. Furthermore, while many such tasks have been worthwhile endeavors, we often find that the models they produce do not work in the wild or do not fit end-users' needs as hoped. Suppose the tasks that guide the work of thousands of AI researchers do not reflect the HCI community’s understanding of how humans can best interact with AI-powered systems. In that case, the resulting AI-powered systems will not reflect it either.
This course will explore the opportunities for HCI and AI researchers to begin closing this gap by collaborating to directly integrate HCI’s insights and goals into immediately actionable AI tasks, model designs, data collection protocols, and evaluation metrics.
Although, we are not enforcing any specific prerequisite courses, students are encouraged to have taken at least one AI course (e.g. CSE 415, 416, 446, 447, 455, 490G1, 546) or at least one HCI course (e.g. CSE 440, 510).
The class format will be a combination of lectures, light assignments, paper discussions, and a course project.
The lectures will focus on providing both a historical as well as a contemporary lens on human-AI interactions. Our learning goals are:
- Contextualizing artificial intelligence and human-computer interaction as two twin sub-fields of computer science.
- Awareness of the major researchers, terminology, historical debates, and long-standing problems that have shaped the two fields.
- Knowledge of the current state of the two fields, including key recent literature, outstanding research questions, and newer areas of interest.
- An understanding of different research and design methods within each, with a focus on the design of interactive systems, and which methods can practically best tackle which research questions.
The paper discussions will be centered around papers and connect the topics covered in the lecture to contemporary papers in computer vision, machine learning, and human-computer interaction. Students who take the class for credit will be expected to participate in discussions and conduct a quarter-long research project.
Two lecture slots will be dedicated to two light assignments. The assignments are primarily designed to create rapport amongst students and hack together quick ideas.