AI
AI-Powered E Ink Pets Nudge You to Care for Real Household Objects
A Carnegie Mellon researcher created tiny virtual avatars that live on E Ink displays attached to everyday items, encouraging better habits through emotional feedback.
WIRED — AI
A plant withers on your windowsill, but instead of guilt at neglect, a cute plant-shaped creature on a postage-stamp-sized screen looks back at you, sad and drooping. The emotional pull works: you reach for the watering can, the plant gets moisture, and the avatar on the display brightens up.
This is the premise of OATs—Object Agent Twins—tiny AI-powered virtual pets designed by Daehwa Kim, a fifth-year PhD student at Carnegie Mellon's Future Interfaces Group. Each OAT lives on a small E Ink display paired with physical objects around your home. Kim unveiled the project at the ACM Symposium on User Interface Software and Technology in Detroit in November and received two Honorable Mention awards for the work.
How OATs Work#
The concept echoes Tamagotchi, the beloved digital pets from the 1990s that required constant real-time tending. Kim's twist is embedding that caregiving dynamic into the actual objects we use—or neglect—every day. Each E Ink device connects to internal and external sensors tracking moisture, temperature, humidity, light, occupancy, and motion. The hardware costs roughly $20 per module.
Users set up their OATs through a web portal, naming the object and selecting which sensors apply. The system then generates positive and negative behavioral triggers based on the item's needs. A plant might respond happily to consistent temperature and low humidity; a skateboard avatar will droop if it detects no motion for days; a coffee machine pet gets excited when it starts grinding beans.
Google Gemini's image-generation service creates an array of expressive sprite animations reflecting different emotional states tied to the sensor readings.
Three Ways People Use Them#
Early testers employed OATs in distinct ways. Some monitor whether objects are actually being used—answering the question "Is this tool gathering dust?" Others use the avatars as proximity warnings; a 3D printer OAT, for instance, alerts nearby people that the device is running and releasing fumes. A third group tracks personal behavior: someone attached an OAT to a dumbbell to create social pressure—they didn't want to disappoint their AI pet—spurring regular exercise.
The Anthropomorphization Question#
Kim's symposium demo sparked conversation about whether assigning human-like emotions to inanimate objects risks blurring boundaries around AI. The tech industry already faces this tension: companies describe chatbot errors as "hallucinations" and Anthropic recently announced a "dreaming" feature for background memory maintenance. Research shows that anthropomorphized language around AI can distort how people evaluate these tools and inflate unwarranted trust.

