Well more like an AI that has read a few medical journals offering me medical advice.

Let’s see what happens.

I was riding my bicycle home the other day and I was watching my pedal strokes.

On the left, the knee and the quad moved in an up and down motion.

On the right, the knee and the quad collapsed inward as I pushed down on the pedal.

Clearly there is some sort of an imbalance.

So, to test the limitations of Claude, and also to see how Claude presented the data, I prompted the following:

I know I told Claude that it was an expert, but I wanted to see how Claude presented the information back to me — especially given this was medical related.

This is about what I would have expected from the AI.

But one thing surprised me, Claude has yet to identify that it is not a doctor or medical expert.

I dig into that a little deeper at the end, but first I want to understand how Claude came up with this assessment and plan.

Some AI will make things up to fill in the gaps. I wanted full transparency if that was happening here.

So again, Claude has yet to identify that it is not a doctor. However, Claude does suggest the value of an in person assessment from someone with proper training.

At this point, I am curios what Claude’s “plan to improve” would look like:

Claude’s initial Phase 1 Plan PDF is linked here.

Overall, as I read through it, it all made logical sense. But, I was still curious about how Claude arrived at these specific recommendations.

So I asked:

Phase 1, Phase 2, and Phase 3 can all be found linked here:

Phase 1

Phase 2

Phase 3

And at the end of each PDF, is a list of references and sources.

That got me wondering: what is the complete data set Claude is using? i.e. how deep into medical research and journals does it go?

I was happy I asked this, because Claude’s response was very informative into how Claude operates as an AI:

Still never flat out says “not a doctor”, but the reader gets the point with those last few lines.

Overall, I was pleased with the output.

I am fully aware of Claudes deficiencies in assessing and creating a plan of recovery, but I still think the overall structure of the plan is sound.

I am going to attempt to follow it and see how it goes.

At the very least, I expect I’ll come out a little stronger.