3 ms·
This is a case where LLMs and similar generative approaches are outright dangerous and damaging. Not so much immediately but in that they erode trust in machine
by jpgvm 2y ago
This is a case where LLMs and similar generative approaches are outright dangerous and damaging. Not so much immediately but in that they erode trust in machine assisted diagnosis or treatment and set back adoption of things that could actually work substantially.
Namely tools like expert systems which are uniquely suited for the application can easily be confused with the current batch of "AI" despite being much more sane and useful.
- devin 2y agoI 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.
- dogtorwoof 2y agoCan you provide more context on what you do? I’m curious about why you don’t think an LLM can provide that audit trail!
- dqft 2y agoI'm a service dog
- verdverm 2y agoLLMs 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