As I have been exploring reasoning with AI, I keep coming across issues where the analysis itself is changing the meaning of the information it contains. This seems like a big deal to me. I have been looking at it in my AI-collaboratively written series that I share on medium: https://medium.com/@hmgrossman.