I’ve been thinking about using LLMs along with personal note taking apps like Obsidian.

There are so many functions which seem to defeat the point of the process (e.g. Summarising is a great tool to help remember information, getting an LLM to do it saves time but reduces efficacy — though there are other benefits of summarising using an LLM such as identifying and eliminating content for further study on a topic).

However, I think I now have a good idea for using an LLM.

Critiquing an atomic note with an LLM

Let’s save you’ve read an article and came across something interesting — perhaps teleological ethics (why yes, I am studying Christian ethics right now). So you make a note on it — what it is, it’s strengths and weaknesses, key figures, etc. Now you get to the point where you’re quite happy with it but you want to make sure you’ve understood the topic right and not missed any key points.

Hello LLM.

By sharing your content with an LLM and asking it for feedback (did I miss anything? is anything incorrect? Is any part unclear or poorly written?) you get a reviewer of your note. You might realise you missed a key point OR the source you were studying from had missed something too!

Or at it’s most basic, you may get some good writing feedback.

Beware! Here be hallucinations

Of course, the danger of hallucinations is real and might be a big issue espeically if you are trying to cite a source or find a reference book that posits an opposing view or part that you primary source missed.

But that’s where you come in human.

You shouldn’t accept the LLMs answer as gospel truth, but it can help you check and know what to investigate further.

What do you think?

I’m not certain about this: I don’t really want to feed my private notes into an LLM so it’s gobbled into its training data and I can see some real risks of hallucinations.

But I’d love to know your thoughts.

I’d real love to know if you have found another good use of LLMs with a private note taking app.