Broad Listening in 2025

- A new illustration was created based on the understanding from December 2, 2025.
- In the diagram of Broad Listening drawn in 2023, it was depicted like this:

- The understanding of the left diagram (E) and the right diagram (F) has deepened over the past two years, leading to a new illustration.
- A:
- When many people express their opinions, the recipient gets overwhelmed.
- This is represented by E.
- Issues like the Public Comment Overload Problem reflect this state.
- B:
- To avoid the situation in A, there is a practice of "listening only to a select few."
- This leads to dissatisfaction among those not included in the "select few."
- C:
Consequently, it leads to demands like "Listen to us too!"
However, doing this results in the state of A, making it an impossible choice.
An unexpected realization over the past two years is that some people think "Broad Listening" is about "turning B into C."
The "Listening" in Broad Listening is about "hearing," and the subject of hearing is the "listener."
- D: The form that has emerged as of December 2025
- The diagram F has become more comprehensible in detail.
- A system where AI Listens and Organizes Thoughts is emerging by 2025.
- In the February 2025 lecture Scaling Digital Democracy, a diagram of AI-Mediated Communication was drawn.

- At that time, it was expressed as "various forms are possible."
- In March 2025, Idobata System v1 was created.
- After repeated experiments, by June, it evolved into "Manifesto That Can Speak" by Team Mirai.
- By November, a structure called Mirai AI Interview emerged, where an AI Interviewer conducts one-on-one dialogues with people about what Team Mirai wants to know.

- The arrows in this D diagram being bidirectional is significant.
- In 2023, only a unidirectional view was apparent.
- In the current Mirai AI Interview:
- There is short-term "bidirectional communication" between people and the AI Interviewer.
- However, how Team Mirai utilizes the reports submitted as a result of these dialogues is not very visible.
- The bidirectionality between the listener and AI is quite weak.
- The listener inputs what they want to hear into a prompt.
- After running for a while and gathering opinions, they analyze and get reactions to "what they want to hear."
- Although it is technically bidirectional, one round trip takes 1-2 weeks.
While writing the explanation, I remembered writing about three types of stakeholders, not just two.
This page is a high-quality translation from /nishio/2025年のブロードリスニング. The original content is maintained by NISHIO Hirokazu.