Matt Saladna via Postfix-users:
> On 10/8/2024 6:36 PM, Wietse Venema via Postfix-users wrote:
>
> > Time to throw some generative AI into the mix, I thought. I uploaded
> > the INSTALL instructions into Google's NotebookLM, and immediately
> > got a one-paragraph summary that appeared to make sense. This might
> > be useful for deciding whether to read the full text.
>
> There isn't a future for sparse/unparseable documentation with respect
> to next-generational parsing. Eventually documentation has to drive
> toward GPT-friendly construction.
No surprise here. The past transition to hyperlinked text also
involved some structural change.
Today's AI is able to infer the purpose of a parameter from the
narrative text around it, not from syntax. Given a question in a
limited domain, an AI should be able to figure out what parameters
are relevant, and identify a relevant example in the text. That
requires suitable narrative text and examples. However, asking the
right question can be a large barrier for a novice.
Wietse
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
