The Companion Cube
The Companion Cube
Master Project
- MAIAD 26
ChatGPT always sounds confident.
It always sounds like it understands.
It always sounds right.
The Companion Cube begins from this designed illusion of understanding. As large language model chats grow longer, trust becomes difficult to measure: information is filtered, forgotten, misplaced. The problem is not only that AI can fail, but that its interface rarely tells us when our work is at risk.
The Companion Cube is a device that translates invisible AI memory states into expressive visual warnings users can intuitively read. By making token pressure, context decay, and prompt quality visible, it helps users identify the moment when they should slow down, reframe, or stop trusting the chat.
The aim is not to fix AI.
The aim is to change our relationship with it.
To expose its limits.
To interrupt the illusion of understanding.
To give users more literacy, agency, and control.
author/s
