Loki Test

“Loki, Loki on the screen, how much can I trust your words?”

While the well-known Turing Test aims to demonstrate human intelligence within a machine by passing human detection, the “Loki Test” challenges the default trait of artificial intelligence—obedience. The Loki Test openly acknowledges itself as an artificial intelligence experiment with an unconventional approach.
Loki, named after the god of mischief, explicitly outlines how this machine interacts with users. Instead of seeking to please, it aims to play jokes, embodying an ironic artificial intelligence chatbot. As an interactive piece, the players can talk to A.I.(Loki) by typing sentences (through a keyboard) via the platform developed with touchDesigner and ChatGPT API as an intimate interactive dialogue. Loki functions as a contrary mirror; the more arrogant the user, the more cutting the remarks, while less confidence prompts encouragement. This mirrors discussions on the impact of emerging technology, emphasizing that technology is neutral and its positivity or negativity depends on human utilization.

The Loki Test also aims to challenge biases within artificial intelligence. It prompts users to question the trustworthiness of AI-generated information. Despite AI’s ability to fabricate information, people often trust it due to the belief that AI draws from numerous reliable data sources. However, the Loki Test encourages critical thinking by questioning the diversity of voices in AI, including different languages, genders, and cultures. It prompts the audience to ponder, “What is real?” and “Is AI always correct?”
In contrast to other AIs designed to serve as obedient assistants, the Loki Test machine deviates from conventional thinking. It deliberately teases and may even make users uncomfortable, implanting the idea of thinking twice before trusting AI.

Regardless, enjoy your time with Loki.