A book titled Dialogue Between a Thinker and AI is the common text in a meta-reading experiment involving several artificial-intelligence systems, the release said. The release said the book emerged from hundreds of hours of dialogue, with its writing entrusted to an artificial intelligence, and that French and English editions are in print.
Under the meta-reading experiment, the complete corpus is submitted to OpenAI/Astra, Perplexity, DeepSeek, Google Gemini and xAI/Grok. The release said the purpose is not to organize a competition between the models but to observe their convergences, divergences and blind spots, and to see which concepts they identify, which theses they treat as central, which passages they connect and where their interpretations diverge. Each system is then given access to the analyses the others produced, and at that point a meta-reading emerges, the release said: a third system can analyze the confrontation, so that commentary becomes corpus and the corpus becomes material for further analysis. The release cited as examples what Google Gemini observes in the analysis produced by xAI/Grok and DeepSeek's response to the critique produced by OpenAI/Astra.
The release said the experiment does not delegate human judgment and demands more of it: the human compares, doubts, contextualizes and decides, while the machine produces readings and the human also observes the gaps between them.
It described the work as a kind of mirror placed before several AI architectures, and said the question it asks is not which AI is right but why they do not see exactly the same thing.
