I would also add the hatchet job the first FDA lady presented...

I would also add the hatchet job the first FDA lady presented showing FDA playing with MSC's in their lab and showing all kinds of variability (heterogeneity) and then showing a some of their software/AI was able to pick out various differences. Basically trying to show up SI/MSB and say 'Hey, we have shown the MSC's vary in potency and attributes and here is some software we wrote which picks it up, so why can't you do it in your manufacturing?'

That was a tough issue for SI to come up against when it was presented to an ignorant cohort. What SI would like to have said in response was:

1. Show me how you manufactured those MSC's and show me the validation of that manufacturing process, raw material inspection and qualification and final lot release documentation. And then I'll show you why you screwed up! We have spent thousands of man years developing our processes, show me how you developed such a process on your FDA budget!

2. Show me the software validation of the AI code you claim picks up these variances. The FDA has a around 100 documents on how to validate software when used as a medical device ( which is what you are claiming you are doing). That will take you hundreds of man years, like it does for industry when you make us follow those V&V processes, you fools.

Crap in, crap out. It's not the FDA's job or talent to produce that kind of research. The smart guys are in the industry and are paid accordingly.