Autograph: An Evolutionary Neural Network Singularity Experiment in the Browser

Autograph is a crowdsourced strange loop experiment that runs tiny neural networks in the browser, learning to paint through evolution. Its ultimate goal is to enable AI to draw and 'prove' its own origin, exploring self-reference and emergent phenomena in artificial intelligence. The project is maintained by admiralakber, sourced from GitHub (link: https://github.com/admiralakber/autograph), and was released on May 31, 2026.

"Strange Loop" originates from Douglas Hofstadter's works, describing self-referential hierarchical structures (such as Escher's stairs, Gödel's theorem). Autograph transforms this philosophical concept into a technical reality: neural networks learn to draw images, and the images in turn describe their own structure and evolutionary process, forming a strange loop at the technical level.

The project's core design is based on the browser environment, lowering participation barriers and supporting distributed collaboration. Tiny neural networks enable real-time training and inference in the browser; through genetic algorithms and swarm intelligence, parameters are propagated, mutated, and selected among participants to achieve distributed evolution. Unlike traditional centralized AI training, the crowdsourcing model generates data dynamically, and the direction of evolution is determined by collective intelligence.

Implementation faces three major challenges: 1. Efficient training and inference under browser performance constraints; 2. Distributed consistency issues in coordinating participants' contributions; 3. Open problems in self-referential image generation (strategies like attention mechanisms and meta-learning may be used to guide the network).

At the philosophical level, it transforms abstract concepts such as self-reference and the nature of consciousness into interactive experiments; at the educational level, it provides learners with an intuitive platform to demonstrate the principles of neural networks, genetic algorithms, and distributed systems, stimulating curiosity and creativity.

Although small, Autograph contains a grand vision, proving that technical experiments can have both philosophical depth and entertainment value. Against the backdrop of AI commercialization, it reminds us to maintain curiosity about the essence of technology and question the true meaning of "intelligence". Those interested in AI, art, or philosophy should pay attention—its thought-provoking insights are more precious than practical tools.