AI told me one patient had 148 pregnancies

I was working with medical dataset (diabetes UCI data) and I was using AI data analyst . I asked AI to load data from my hard disk. It generated Python code to load data and display it. When I saw the first few rows I was shocked - it was showing 148 pregnancies for the first patient! So clearly something was wrong. The AI itself seen this as well! The AI data analyst sent additional prompt and it spotted this. AI computed mean Pregnancies in my data frame, which was 121 which is too high ... other columns had wrong values as well, for example Age 0 or 1.

thanks to automated additional prompt with ask to validate and analyze results, and thanks to displaying analyzed data I was able to quickly find halucination and fix it.

What was the core of the issue? There was additional comma sign in one of the rows in the data. Simple mistake, but it was producing crazy results.