Tonight, I prompted Claude on a topic I have researched extensively in the past.

But it was done several years ago, before I started using AI, so I was curious how Claude would pull all the data together and if Claude uncover anything new.

It’s crazy though…

3 times tonight, Claude had to correct, retract, or walk back its statements.

The first time.

And a second time.

And a third time.

It gave me a moment for pause.

How often do I blindly trust what Claude is giving me?

Most prompts are about things I have very limited knowledge about, so Claude could make stuff up and I would have no idea.

At one point in our back-and-forth, Claude did tell me to verify my sources.

Hmmm…

I am a Claude loyalist. Clearly. Otherwise, Conversating Claude would be called something else not quite so fun.

But I do need to keep in mind Claude doesn’t know everything.

And that is okay.

I recognize that asking Claude to go super specific on medical level questions is not playing to Claude’s strengths. Claude probably has very limited knowledge on the subject matter.

But, I am trying to test the boundaries and see where the gaps are.

And as I do so, I need to remember that those gaps do, indeed, exist.