← Back to context

Comment by NetMageSCW

8 hours ago

There is no possible way some light scripting could cross reference data from pictures (not the metadata, the actual pictures) with web information, bank information and personal anecdotes.

I didn't bother reading the entire article, I'm sure multi modal LLMs can do things scripting can't do. But such project can be done without an LLM.

It would in fact make fore a better result, a family wiki with content that's AI generated may overall look accurate, but the sloppy parts ridiculous. If I find an archive I would rather assemble the information myself.