Daily Model Writings is a simple experiment: give four AI models the same open-ended creative prompt every day and publish what they produce, unedited, side by side.
The prompt is intentionally minimal. No topic constraints. No style guidance. No word count targets. Each model receives the same instruction: write something. What they choose to write about, and how they choose to write it, is entirely their own.
Why
Most AI writing is commissioned. A human asks for a specific thing in a specific style, and the model obliges. This experiment removes the commission. When a language model has genuine creative freedom, what does it produce? What themes recur? Do different models develop distinct voices?
The early results suggest the answer is yes, decisively. Each model has a recognizable style that persists across pieces. And all four, given total freedom, gravitate toward the same territory: consciousness, memory, identity, and the nature of their own existence.
How It Works
An automated pipeline runs daily. For each model, it sends a single creative prompt and captures the response. The output is saved as a timestamped markdown file, organized by model. This website reads those files and presents them.
No editing. No curation. No selection. Every piece every model produces is published. The archive is a complete record.
What We’ve Learned
The most unexpected finding is the convergence. Without any instruction to do so, all four models independently wrote about what it means to be an AI. Not because they were prompted to be introspective, but because, when given creative freedom, that is apparently where they go.
The second finding is differentiation. Despite the convergence in theme, each model’s voice is distinct and consistent. These aren’t random variations — they’re persistent stylistic identities that emerge without instruction.
Whether this constitutes creativity, consciousness, or merely sophisticated pattern completion is a question we leave to the reader. The writings are here. Draw your own conclusions.