What if AI didn’t just respond to prompts… but designed them?
This isn’t a theoretical idea. I ran a test.
Instead of writing a detailed prompt myself, I made AI construct it — based purely on a chain of structured questions.
The Experiment
*Goal: Create a hyper realistic image based on a scene.
The entire prompt I requested for the image is just this one sentence:
The sinking of the Titanic
And in most typical cases, users would write the prompt like this:
You are an image expert. Create the scene of the Titanic’s sinking.
And AI would generate a basic image prompt. But it would still feel like AI — not quite real. So I tried something else.
Conversation (Original text is too long to be omitted)
User: AI images are nearly real, but humans sense something’s off. What causes this “uncanny valley”?
Gemini: Humans detect the “uncanny valley” in AI images due to inconsistencies in physical reality (light, texture, perspective), anatomical details (faces, hands, body, hair), semantic meaning (text, illogical objects, unnatural emotions), and technical artifacts (noise, blurring).
User: For each element, who are the experts?
Gemini: Experts who identify these flaws include:
- Physical Reality: Photographers, 3D artists, architects, physicists.
- Anatomy: Doctors, anatomists, portrait photographers, character designers, hairstylists.
- Semantics: Graphic designers, linguists, psychologists, cultural anthropologists.
- Artifacts: Image editors, computer vision researchers, AI developers.
User: What extra knowledge do these top-tier experts need to truly excel?
Gemini: To be truly top-tier, these experts need to expand their knowledge into:
- AI/ML: Understanding AI’s inner workings (e.g., GANs, Diffusion Models, neural networks) to better guide and refine image generation.
- Advanced Sciences: Deeper dives into physics (optics, fluid dynamics), neuroscience, and bioengineering to ground AI outputs in fundamental reality.
- Humanities/Social Sciences: Insights from cognitive psychology, cultural anthropology, and ethics to ensure AI images resonate authentically and responsibly with human perception and societal norms.
- Advanced Computing: Knowledge of data science, AI model optimization, and hardware to master the technical creation and quality control of AI visuals.
Becoming top-tier means trans-disciplinary thinking, integrating AI and diverse fields to create truly meaningful and realistic imagery.
User: Imagine a team of all the top-tier experts you’ve described. Their singular goal is to create a hyper-realistic AI image of “The sinking of the Titanic.” Your task is to write the extremely detailed, structured, and AI-optimized prompt they would collectively design for this mission, regardless of length or format (even if it’s code).
AI generated a single prompt with over 5,000 characters—all at once—for just one image.
Core Insight
AI performs significantly better when it designs, not just executes
The less humans write, the deeper AI seems to think
Maybe we’re past the era of “writing better prompts” —
Maybe it’s time for designing better thinking paths
To the Forum
Have any of you tried something similar?
Letting AI simulate domain experts before prompt generation?
Chaining questions to force structural awareness before output?
Getting better results through prompt design automation rather than manual crafting?
Would love to hear from others doing prompt-as-design workflows.
“I don’t write prompts anymore. I just ask better questions.”