A research-led approach to informed AI choices
Every AI prompt is processed somewhere: in a distant cloud or directly on a user’s device. That choice affects data exposure, resource consumption, performance, and ultimately how much control people have over their interactions with AI.
TerraCore makes these trade-offs visible. Through its responsible AI (rAI) app, users can compare the same workflow using cloud-based and on-device language models, exploring differences in privacy, estimated energy use and CO₂ emissions, and responsiveness. Snapdragon X Series PCs, with dedicated neural processing units, help make local inference a practical option alongside cloud services.

Grounded in aihorizon R&D’s Responsible AI Framework, TerraCore treats informed choice as a matter of human agency and justified technology design. Privacy must be protected through appropriate safeguards, while environmental comparisons require transparent assumptions: local processing is not automatically more sustainable in every context.
In workshops with Educational Foundation Freiburg, teachers and students explore these choices firsthand. Developed with input from Microsoft and Educational Foundation Freiburg and with support from Qualcomm Technologies, TerraCore investigates how privacy, ecological responsibility, and performance can become meaningful considerations in everyday AI use.
