Thierry Ehrmann Uses New Book to Test AI Cognition

The project, titled Dialogue Between a Thinker and AI, moves beyond simple machine-generated text by forcing models to analyze one another’s critiques. Rather than pitting systems against each other in a performance contest, the experiment focuses on identifying specific blind spots and interpretive patterns within each architecture. When one AI evaluates the analysis produced by another, the resulting commentary creates a recursive loop that highlights the fundamental differences in how these models synthesize information.

This approach shifts the role of literary criticism from human-to-text to machine-to-machine. By observing the gaps between how Gemini or Grok interprets the same passage, researchers gain insight into the distinct cognitive frameworks underlying each system. Despite the complexity of these interactions, the experiment maintains a firm human-centric boundary: the final synthesis, doubt, and contextualization remain the responsibility of the human observer, who uses the machine-generated readings as data rather than definitive truth.