Rasa Open Source testing

Hi there!

I am a QA mentor and a QA internship lead with 20+ years of experience in QA. During the last months, together with students and interns, we are trying to understand how to test AI applications. The materials on the Internet and ChatGPT don’t help much, but we have some practice testing AI-powered applications of our internship customers (e-commerce and others), based on the accuracy formulas.

So my question, is anybody here who test Rasa open source (as a blackbox or a whitebox)? Manually or automatically? Would be happy to share experience.