The Importance of Human Subjects in Trusting a Model

Raj: I think that there's also kind of a use case mismatch. We want to use these in situations where we want some type of well calibrated probability and integrates a ton of information. If we're not in that scenario, we probably don't need these tools in the first place. Marzia: But when the goal is adding up lots of very small statistical contributions to disease, again, it's hard to reduce that to a simple explanation. And that's precisely the area where we want to use tools like this.