sound.fan
Editorial
Investigations and essays on what AI products do, how their claims can be checked, and where the evidence stops.
- Read the story →
What a stranger can verify about an AI business
We tested five XPRIZE Top 100 finalists using only public evidence. Each claim we could check opened a harder question, until we reached one that public evidence could not answer.
- Read the story →
Ask the model for something your code can check
Four AI projects handle wrong model output the same way. They change what the model returns so plain code can check it before anything happens.
- Read the story →
What AI did, and what stayed human
Four winning projects show a useful division of labor: AI interprets or proposes; people encode the rules that validate results and control execution.