MEDIENEXPERIMENT / AN OPEN EXPERIMENT
Seeing images. Understanding movement.
Five photographs and a short film begin a new experiment: what do search engines find, and what can AI systems actually recognise?
The files come from the selection supplied by Mario. We describe visible subjects only; locations and capture dates will be added once confirmed.
No results yet: this gallery is our starting point. We will examine discovery, reproduction of text and actual visual understanding separately.
Five perspectives
River and trees
A watercourse in front of green trees and a meadow beneath a blue, cloudy sky.
A railway through the fields
A straight railway track leads between fields towards the horizon, with poles and a signal beside it.
Traces in the wood
A tree stump with a cracked cut surface and moss at its base, against a blurred woodland background.
A sunflower in the foreground
A large yellow sunflower in front of more sunflowers and a blue sky.
A leaf in detail
A green, lobed leaf with reddish veins and a stem against blurred greenery.
What we want to record
- When a specific page or file first appears in a search result.
- Which information an AI reproduces from titles, image descriptions and other page text.
- Which visible details or movements it describes correctly, and what it adds or confuses.
We record the date, service and model where known, exact prompt, cited sources and answer. A matching description alone does not prove image or video retrieval. Comparisons will therefore include questions about details not already supplied in the page text.
Media is served by this website. Small previews keep the initial page light; larger files open on request. There is no visitor tracking. Source files remain locally preserved; web copies contain no inherited EXIF location metadata.