Training Through Tending*
*tending: the action of care both physically and emotionally


UX Designer/Researcher


Research for AI
Design for new public space
System design for future of agriculture


After Effect
3D modeling
3D printing
Physical prototyping

Training through Tending is a design research investigation that explores the process in which AI systems learn human bias through the act of tending plants.

Instead of thinking about labor shifting from the human to robot, I want to design the process that allows the user involved in the first step of agricultural production, rather than consuming the product in the end. What is the responsibility of the user? How could AI robots as an agency to actively care for plants like a human?

↑The installation for TTT

Training through Tending imagines a process through the care of a single plant, or the proxy plant, that in turn affects the care of thousands and millions of other plants in industrial scale farming–for every act of tending the proxy plant receives, the thousands of other plants in the large scale operation grow the same way.

The project imagines a new technofuture in the small-scale urban farm space by bring in a symbiotic relationship between technology and nature.

The project challenges the form and composition of the industrialized agriculture landscape by bringing together the experience of personalized community gardening and the vast scale of industrialized mass agriculture.

Multichannel video - please play two video at the same time

The first step for TTT

The second step for TTT

↑pruning robots in the field / security camera hidden in the grass to indicates the user's action of tending plants is under surveillance

landscape map

new public garden space

process sketches

©Xiaoxuan Liu 2020. All rights reserved.