Words become visuals and sound, shaped live by a human performer. Witness an artwork taking form as it is performed.

AI artist aratama 璞 will appear as a special guest at the GMI & MiniMax AI Art Gallery at Shibuya Hikarie in Tokyo on Thursday, October 8, 2026.

He will give a live performance of about ten minutes with his own real-time generative system, PERCEPTION → LIVE. | Real-Time Generation AIVDJ. Combining AI-generated visuals and sound with live human mixing, the performance will be followed by a short talk on the artistic ideas and technology behind the work.

Turning Words into Visuals and Sound—and Performing the Process

PERCEPTION → LIVE. is an AIVDJ system that generates visuals and sound from words, building the piece in real time through looping and mixing.

ChatGPT interprets the input words and concepts through APL (ARATAMA PERCEPTION LANGUAGE), aratama 璞’s original expressive framework, and translates them into instructions for audiovisual generation. These instructions are sent via MCP (Model Context Protocol) to a Mac-based system that uses MiniMax H3 to generate video with audio.

As the generated material plays, the artist adjusts volume, EQ, and reverb. Instructions to the AI and live human mixing shape the work together, transforming it as it unfolds.

The chat interface and console reveal the original words, how they are interpreted, the instructions being sent, and the progress of generation. Audiences experience the visuals and sound while witnessing how they come into being.

The Wait for Generation Becomes Part of the Performance

Real-time generation involves waiting and unpredictable changes. PERCEPTION → LIVE. incorporates both into the structure of the performance.

If the next output is not ready in time, the system returns to the beginning of the material generated so far and loops through it. When new output becomes available, the system prioritizes it for playback, taking the performance in a new direction.

What emerges, when it changes, and how the artist responds all shape the unfolding work. Generation, playback, and human intervention happen concurrently. Together, they constitute the artwork.

Giving a Website’s Language a Life of Its Own

For the work submitted to the event, aratama 璞 used words and concepts from the GMI Cloud website as source material.
“One cloud for compute, inference, and agents.”
Compute becomes a skeleton. Inference becomes a nervous system. The autonomy of agents takes shape as organs.

Through APL, the language of infrastructure is reimagined as organic structures and motion, then expressed through visuals and sound. The submitted work was created using MiniMax H3 and documents the process of generation and live performance.

For the live performance at Shibuya Hikarie, the system is planned to use MiniMax H3 through the GMI Cloud API.

At the venue, aratama 璞 will operate PERCEPTION → LIVE. in real time, presenting the full process from instructions and generation to human mixing. The interplay between generated output and decisions made in the moment will shape a work unique to that night.

Watch the submitted work and read about the generation and performance process on X

Playing a System That Behaves Like a Living Entity

At the heart of this practice is All-Human Perception, the concept that guides aratama 璞’s work.

How does contact with AI reshape human perception and expression? PERCEPTION → LIVE. explores this question through performance, bringing people, AI, and the space itself into a shared process.

A person supplies words. AI interprets and generates. The resulting visuals and sound, in turn, influence human judgment and action. The artist becomes part of this cycle, changing alongside the work.

Playing the system itself as it behaves like a living entity. In this performance, the visuals, the sound, and the relationships through which they emerge are all part of the artwork.

Event Overview

The GMI & MiniMax AI Art Gallery is an evening of AI-generated video and music screenings, guest talks, and networking for anyone exploring AI and creativity.

GMI & MiniMax AI Art Gallery — Event Details

Please check the official event page for current availability and attendance requirements.

Event details and registration

Speakers, program, and schedule are subject to change. Promotional images are for illustration only.

About aratama 璞

aratama 璞
AI Artist / AI Creative Architect

Guided by the concept of All-Human Perception, aratama 璞 works across video, music, and live performance. He develops and performs with PERCEPTION → LIVE., a real-time generative system that transforms words into visuals and sound through APL, his original expressive framework. His AIVDJ practice explores playing the system itself as it behaves like a living entity.

For performance bookings or inquiries about custom system development, please get in touch.

• aratama 璞 — Official Website
• X: @aratamadao
• AI Creators
• Contact