Hmm, how does a language model "search for trends"?
Pages and pages of text that can't be read by any one person can be parsed in seconds by a machine with enough knowledge and understanding to make something meaningful of it.
The same idea is already being applied for law - https://twitter.com/ai__pub/status/1644735555752853504
What you described sounds like a way to search documents not mine for trends.
How is ChatGPT going to infer connections here without supervision? Are you saying the process is to paste every medical record into the context window and then prompt GPT to “find common patterns in the above?”
That doesn’t sound likely to come up with useful results on its own… Maybe I’m wrong though.
It’s seems more likely that GPT is being used to interpret the text into a standardised form that can then be mined using other techniques.
It'll be more implicit than that. For a toy example, paste some fake but detailed stats about certain treatments. then try to ask gpt to recommend a particular procedure for a certain patient. Those stats will undoubtedly influence gpt's recommendations...and in ways that make sense.
Searching documents is powerful here because searching documents influences GPT's response in much the same way it would an equivalent human.