Rethinking AGI Design: Let AI Prompt Humans Instead

I’m not a native English speaker, and the content below is refined and translated from my mother language by ChatGPT.

Introduction

This concept targets two main domains: artistic creation and psychological therapy.

Today’s generative AI systems — whether for text, music, images, or video — generally follow a one-way workflow: the user gives prompts, and the AI produces results.
While efficient, this paradigm overlooks something fundamental: the human need for exploration, expression, and self-projection in the creative process.

To unlock AI’s deeper potential, I believe we should reverse the direction — let AI prompt humans, guiding them to think, explore, and express.
In this vision, AI becomes a collaborator and inspirer, rather than a direct substitute for human creativity.

Moreover, psychology has long employed methods like music therapy and sandplay therapy, which use artistic creation as a medium for healing. Yet, these traditional approaches are often slow and inefficient. I envision a system where AI could serve as a dynamic facilitator in such contexts.


System Overview

Creative work usually progresses through three stages:
divergence (idea generation), organization, and realization.

Accordingly, the proposed system consists of three major components:

  1. Immersive Prompt Environment (divergent stage)

  2. Idea Organization Environment (structuring stage)

  3. Concrete Creation Environment (realization stage)

Through these stages, the system guides users from spontaneous inspiration to completed works — allowing AI to accompany, rather than replace, human creativity.


1. Immersive Prompt Environment: Let Inspiration Flow

This stage combines a dialog box with a dynamic “inspiration field.”

The user begins by typing what they wish to create. Instead of generating the work outright, the AI performs two key actions:

  1. Inquiry and Clarification – The AI asks guiding questions to help the user refine their ideas and uncover underlying psychological motivations.

  2. Inspiration Fragment Generation – The AI continuously produces multimodal fragments — floating keywords, short sentences, small images, video clips, and snippets of music — all displayed in a flowing, cloud-like interface.

The user can drag and drop any fragment that resonates into a temporary inspiration pool, forming a personal collection of ideas.

This mechanism is reminiscent of sandplay therapy — the fragments act like symbolic “building blocks,” while the inspiration pool serves as the “sandbox.”
It also draws from TikTok’s continuous engagement model, keeping users in a state of sustained curiosity and flow as they interact with AI-generated stimuli.


2. Idea Organization Environment: Turning Chaos into Structure

Once a user accumulates a set of fragments, the system enters the organization phase.

Here, the AI helps the user establish connections between fragments:
explaining each fragment’s meaning, how it relates to others, and what role it might play in the overall creative vision.

The interface functions like a note-taking or knowledge management app (e.g., Notion or Obsidian):

  • Fragments can be organized hierarchically or through bidirectional links.

  • A dialog box remains available for users to discuss and refine the structure with the AI.

This stage transforms scattered thoughts into a coherent creative framework.


3. Concrete Creation Environment: Shaping the Final Work

Now, the user begins the actual process of creation.
The system provides a project area where users can upload or embed their in-progress works — a few chapters of a novel, a musical composition, or an animation clip, for example.

The AI acts as a creative assistant, not a generator of finished products:

  • It offers suggestions when the user feels stuck.

  • It references prior interactions to provide contextual, personalized advice.

  • Suggestions may take the form of text, musical motifs, or even annotated video frames.

Importantly, the system discourages direct copy-pasting of AI-generated content.
If a user relies heavily on such content, the system will display a higher “AI Participation Index.”
This design choice reinforces the system’s true purpose: not to maximize production speed, but to enhance the human experience of self-expression and emotional fulfillment through creation.


Conclusion: The Personal Origin of This Idea

I’m an introverted person with a mind full of ideas but little drive to realize them.
My deepest joy comes from listening to music — during which countless images and emotions surface effortlessly.

I’ve often wondered: could creation feel as immersive and natural as listening to music?

That thought led me to envision an AI that acts as a resonant partner in imagination — helping me turn inspiration into form while preserving the joy and authenticity of the creative process.

This is why I designed this system:
to let AI guide, inspire, and heal, while keeping the human spirit at the center of creation.

I really like this idea. Especially the fact that it avoids simply asking the human a lot of questions because doing so is mentally draining. Instead, it lets them explore, which can be fun and engaging.