Info & Ethics

A laboratory for critical exploration of computed history.

Methodology

Time Machine does not show the past “as it really was.” It shows how AI models imagine it was, based on the data they were trained on.

We use Gemini AI to generate hyper-realistic historical visualizations in real time. Every image is the result of a probabilistic process, not traditional archival research.

  • Standard Model: Reflects the average visual consensus of the web (mainstream bias). Produces “cinematic” and idealized images.
  • Conspiracy Model: Amplifies hidden patterns and alternative theories. Fine-tuned on alternative literature and counter-narrative forums.
  • Academic Model: Limits creativity in favor of archaeological rigor. Trained on museum archives and scientific data.
  • Fantasy Model: High creative “temperature.” Imagines speculative futures or mythological pasts with maximum artistic freedom.

Ethical Statement

Simulation, Not Reality

Every image produced by this site is synthetic. It is prohibited to use these images as documentary evidence or to pass them off as real photographs.

Bias Transparency

Our goal is to make biases visible. If a model generates stereotypical images, it is not a system error — it is a characteristic of the dataset that we want to expose and critically discuss.

Critical Reading Guide

1. Look for Anachronisms

AI often inserts modern objects or incorrect architectural styles. A wristwatch in 1500? An electric streetlight in 1800? These errors reveal the limits of the dataset.

2. Who Is Represented?

Observe the demographics of the crowd. Is the model over-representing or under-representing certain ethnic, social, or gender groups? Why?

3. Aesthetics vs Truth

The "Standard" model tends to create "cinematic" images with dramatic lighting. Real history was often more mundane, dirty, and less photogenic.

4. The Data Void

If the model generates a confused or generic image, it may indicate a lack of historical data on that place or period. Silence is also data.