I have a background in expert systems in a medical context, and it is wild to see people reaching for LLMs in this space. Having a clear audit trail for how a decision was reached is a requirement, and these tools don’t provide it.
LLMs can be a component in a system. See them as better language handlers than existing algos for looking at what doctors write to see if there are mistakes or omissions. Flag things for review rather than make decisions. In the audit trail, there is less concern about what model you used to flag something for human review. If you can catch mistakes, you can save lives, and the hospital money
Can you provide more context on what you do? I’m curious about why you don’t think an LLM can provide that audit trail!