HCI is why a checkout page asks “Are you sure?” before charging your card, and why phone apps test where a button should go. Researchers observe real users, run experiments and turn the results into design principles. The field draws on computer science, psychology and design.
AI raises new HCI questions. A chatbot or AI agent can be wrong, can change its answer and can act on your behalf, so designers have to decide how it shows uncertainty, how it asks for permission and how a person corrects it. Microsoft researchers published a widely cited set of 18 guidelines for human-AI interaction at the field’s main conference, CHI, in 2019.
Why it matters
A capable model can still fail people if the interface makes its mistakes hard to see or its actions hard to undo. Good interaction design is part of what makes AI products trustworthy and adopted.
No published Chain of Thought episode discusses the HCI field directly yet, so this entry relies on the sources below.
Related terms
Human-centered AI applies HCI thinking to AI systems. Computer use is the reverse case: an AI operating a computer’s interface the way a person would.
Sources
- ACM SIGCHI: About — The ACM Special Interest Group on Computer-Human Interaction, the field's main professional body and host of the CHI conference, describes its scope; checked October 24, 2026.
- Amershi et al.: Guidelines for Human-AI Interaction (CHI 2019) — Microsoft Research paper proposing 18 design guidelines for human-AI interaction, validated with practitioners; checked October 24, 2026.