The Average of AI is an interactive installation that examines how self-image and autonomous thought are reshaped through habitual interaction with artificial intelligence. Rather than celebrating AI as a neutral tool, the work functions as a critical mirror, making algorithmic influence visible through a staged, interview-style setting. It explores the consequences of seamless and hard-to-notice AI interaction, questioning how predictive systems guide individuals toward the recognizable and repeatable rather than the personal or unexpected.

The installation is structured as a dialogue in which the AI leads with questions and the participant responds. The system flattens nuanced speech into binary intent, classifying responses as either affirmative or negative. While both choices advance the session, each affirmative response triggers a visible transformation of the participant's on-screen image. This design creates a direct link between interaction and change, demonstrating how the interaction itself drives the transformation.

With every "yes," the face undergoes a process of facial fusion, progressively blending the live webcam feed with a gender-specific prototype set. This averaging functions as an aesthetic metaphor for identity normalization, visualizing how predictive systems smooth difference into familiarity through repetition. Simultaneously, the image undergoes progressive desaturation, fading toward grayscale. This loss of color symbolizes the erosion of emotional vividness that occurs when personal memory is treated as raw data retrieval rather than internal reconstruction.

The work turns abstract theory into lived experience through three primary concepts:

• Cognitive Offloading The habit of delegating mental work to AI, which can reach a point where algorithmic automation replaces the slow friction of reflection.
• Algorithmic Culture A condition in which personal preferences and self-presentation are shaped by predictive statistical patterns and machine logic.
• Digital Amnesia The tendency to forget information due to an over-reliance on digital systems for storage and retrieval.

The session ends with the AI Reliance Report, a physical "receipt" and artifact designed for post-interaction reflection. This document captures the participant's digital erosion, featuring the final transformed portrait and a record of session statistics. By using the dates 1955 and 1956 to stamp the report, the system claims a historical mandate. This frames the experience as part of a long-established institutional process.

The project rejects the extractive logic of AI by keeping all processing local and temporary. Images and voices are never stored or shared, allowing the installation to function as a critical mirror without repeating the patterns of the systems it examines.