Machine Learning Systems

One thing I would kind of caution anyone in this field is, you know, there's machine learning systems. How do I make my machine learning system to have good enough quality that like it'll hit my 99% image detection threshold in real life or something like that? And then I build different ball defense of design, like defense design. HowDo I defend against adversarial input? And in general, like, that's really hard. If you told me a new fancy machine learning system that will like, test,you know, fraud or detect like, got will guard my system. It'd guard my system in a complex environment. I would be extremely skeptical. You can do fraud detection