Muses ex Echoes (2022)

How do we envision creativity in the future? Is AI capable of generating something new and original?

In this work, two AI agents take turns generating images and interpreting them.
One agent generates an image and describes it aloud, then the other agent listens,
generates a new image based on the description, and articulates its interpretation.
This alternating feedback loop between the agents forms an endless cycle.

If allowed, we can anthropomorphize the system and say these AI agents emulate what human artists do: artists (AI) perceive and appreciate their peers’ work and create something inspired by them.
In 2022, we witnessed huge progress in AI image generation technology. While we benefit from AI-generated images’ high quality, we can still doubt their originality and creativity. These AI models were trained on art made by humans. So they directly reflect the biases and values of our society and culture. The fact that AI-generated images will likely spread online and be used again for training also implies the existence of another feedback loop. “AI-generated art” is really a long-lasting echo of human creativity.

But humans may have been in a similar feedback loop too. We create based on what past generations made, and our work will influence future generations. Still, history shows that we can make something new, original, and valuable.

So what’s the difference between creativity that arises from the interaction between humans and one that may emerge in AI systems?

The AI models here don’t just listen to each other’s words but also hear viewers’ voices and environmental sounds. Does this external noise affect the creative process in the system? Does it stop the system from being stuck in a loop?

Now, with AI models, humans aren’t alone in this creative echo chamber. We wonder what we’ll see when all of “us” resonate together and what will come out of the echoes of creativity or “muses ex echoes.”

画像に alt 属性が指定されていません。ファイル名: 2-1024x683.jpg
Exhibition at NTT ICC
画像に alt 属性が指定されていません。ファイル名: output-1024x1024.jpg
Examples of AI-generated images


  • Supervisor: TOKUI Nao
  • Technical Director: KOBAYASHI Atsuya
  • Concept Director: KOBAYASHI Yuga
  • Original Concept: Ryo SIMON
  • Lighting: OKAZAKI Keisuke, TAKAISHI Keito, SHIBUYA Kazufumi
  • Machine Learning: ISHII Asuka, SAWA Shoma
  • Sound: Ryo SIMON, TAKANASHI Dai, OBARA Kai
  • Visual: TAKAISHI Keito, SHIBUYA Kazufumi, ISHII Asuka, MATSUOKA Yuma
  • Concept: HANDA Sogen, NOBUSUE Ryuku, OKAZAKI Keisuke, INOUE Takumi
  • Support: NARUSE Santa, KIEU Quoc Thai, SASAKI Yuria