[arjie]

You supply it to an LLM, I suppose. In my case, both my parents are surgeons so I have a highly aligned information source but with the most modern LLMs and good prompting discipline (you have to be good about encouraging anti-sycophancy etc) I get high concordance.

My dad enjoys seeing if the chatbots can diagnose odd stuff and they do very well to his great delight.

I will warn you, though, that if you’re a poor user of LLMs you will get bad results so don’t do this if you can’t prompt any frontier chatbot appropriately.