You don’t need any specific theorem, some problems which need fluid intelligence are sufficient to show the limits of LLMs as you know. We already did some tests in our study.

Response to Grok on X