This week from Investigation & Connection: Debugging Imperfect AI with Sherry, and AI, Art, and Design with Deb Lee!
Happy Valentine’s day, UXA Community! We hope everyone is able to take a breather and spend some time with your loved ones (or someone new) 👀
Although we don’t have announcements this week, we do have some freshly baked content and conversations around AI in the field of UX! ⛅
<aside> 🧠 What does it mean to design human-centered natural language processing? What ways can we analyze and break down the different NLP development stages? How do we design and develop NLP systems that considers the needs of human users, and also the ethical and social implications of these systems?
Topics revolving around these two intersecting fields are explored in Human-Centered Natural Language Processing, and how we can meaningfully evaluate and design such systems.
</aside>
🤔 What is Human-Centered NLP?
Human Centered NLP explores the intersection between HCI and NLP, and ways of approaching both topics in different scenarios and how one may build useful NLP models/systems. The course allows students get access to, and understand, both HCI and NLP research papers and methods.
🧐 What do you do in Human-Centered NLP?
This class will be mainly project-oriented, with a half lecture and half seminar structure. Coursework includes lectures, paper readings, class presentations, and group projects.
😯 Who should take Human-Centered NLP?
Although there are no explicit prerequisites, students are expected to be proficient in Python (for completing assignments), and know basic ML concept — to the extent of understanding concepts like train/dev/test set, model fitting, feature, supervised learning, etc. (Not covered in the course)
<aside> 🌟 AI isn’t perfect. How can we identify where AI is going wrong, and how can we make sure these mistakes don’t create harmful consequences? This week, we spoke with Sherry Wu, an Assistant Professor in the HCII, to learn more about these imperfect systems and how we can still benefit from them.
</aside>
“…the models created will never be 100% accurate. They will always have some mistakes. So the question is how do we help people identify those mistakes and mitigate whatever problems that might be caused by incorrect models.”
— Sherry Wu