Carnegie Mellon's OATs: E Ink AI pets for your houseplant

Kim's project sits at the intersection of two trends the symposium discussion made explicit: the push to give AI a physical body, and the industry's reliance on cuteness to drive adoption. The body notes failed attempts like the Humane Ai Pin and forthcoming hardware from OpenAI and Jony Ive, as well as Meta's Muse Charm pocket gadget expected in December. OATs differ in scale—a $20 E Ink module rather than a smartphone replacement.

The demo's popularity also surfaced a debate about anthropomorphizing AI. Kim points to industry habits like calling chatbot errors 'hallucinations' and Anthropic's 'dreaming' feature as examples of human-like framing. Her response is to keep OATs' interactions limited to expressive emotions on the display, arguing that anthropomorphism can help change behaviors but should not be abused.

Kim's user testing found three patterns: checking whether an object is in use, warning people near hazardous devices like 3D printers, and tracking personal habits such as weightlifting. Those uses suggest the appeal is less about companionship than about making invisible states visible—though the short battery life keeps the project in proof-of-concept territory.