Daniel P. Berrangé <[email protected]> writes: > On Wed, Jun 03, 2026 at 07:14:02PM +0100, Alex Bennée wrote: >> Paolo Bonzini <[email protected]> writes: >> >> > Hi Daniel, >> > >> > Thanks for the review. It will take a while to incorporate everything >> > and I'll wait for more feedback, in the meantime just a couple things >> > I can confirm or add... >> >> I mean you could just let the LLM handle it ;-) >> >> AI-used-for: collecting comments and updating patch >> Signed-off-by: Alex Bennée <[email protected]> >> >> I only include this by way of an experiment. I think the new text does >> cover the discussion although I think it has taken a fair amount of >> verbatim text from the source messages that were commentary rather than >> suggestions. > > Most importantly it has applied no critical thought to the review > comments, seemingly treated them all as mostly equally important > and incorporated them. The resulting text is more volumous than > Paolo's, has poor structure and in places is verging on what I'd > class as slop.
Well there is no "thought", although its pretty good at text extraction from the mailing thread. I guess that's the "attention" part of the model. It wasn't helped by the fact there was a full lift and shift of the AI section to a new document. In my other experiments with documentation I've let the model do the changes and then I can review each hunk and reword as I go. For code changes I've found it pretty good - enough so that it might replace my hand-rolled mechanism for saving review comments: https://github.com/stsquad/my-emacs-stuff/blob/master/my-org.el#L176 > > > With regards, > Daniel -- Alex Bennée Virtualisation Tech Lead @ Linaro
