A few things. I added the disclaimer out of respect for the community and the developers. I have been a professional developer for close to twenty years, and I largely echo John's perspective. He is spot on about larger shops adopting these tools already (one of my sons is a developer at one of the companies he listed and they are using LLMs internally).
Years ago I was a carpenter, and I carried an Estwing California framing hammer that I absolutely loved. I could drive a 16d nail with one swing, but that took practice. I was also the type of guy who reluctantly used a nail gun while still recognizing its effectiveness. Over time, I used my hammer less and less. The end result did not really change, and nobody ever asked whether I used a hammer or a nail gun. That is roughly where I am at with LLM tools. They are a tool, and I still need to understand the outcome. Does it work? Does it introduce issues? Can I maintain it? For this particular project, time was also a factor. I already knew the software I wanted to use, but I needed additional functionality. Reports felt like a relatively benign area to experiment while also helping me understand some of the inner workings of both GnuCash and Scheme. I still believe individual responsibility and understanding matter. I am ultimately accountable for the code, whether I typed every character manually or not. I understand your perspective and your choice not to use the reports. At the end of the day, I was simply trying to share something with the community that might help someone else in a similar position. Regards, Andre On Wed, May 27, 2026 at 9:34 AM John Ralls <[email protected]> wrote: > > > > On May 26, 2026, at 21:41, Adrien Monteleone < > [email protected]> wrote: > > > > I simply do not see how or why probabilistic word generators should be > used for anything but novelty kicks and entertainment > > There’s enough code and commentary about the code on the web that the > probability of generating what the user is asking for very high, as long as > the user is very specific about what they want, doesn’t ask for too much at > once, and carefully reviews and corrects the results. An LLM can “type” a > million or more times faster than a human can, so once the user gets adept > at writing good prompts the process is is much faster than writing the code > oneself. I’ve heard/read many senior devs describe it as programming rocket > fuel. > > Regards, > John Ralls > > _______________________________________________ > gnucash-user mailing list > [email protected] > To update your subscription preferences or to unsubscribe: > https://lists.gnucash.org/mailman/listinfo/gnucash-user > ----- > Please remember to CC this list on all your replies. > You can do this by using Reply-To-List or Reply-All. > _______________________________________________ gnucash-user mailing list [email protected] To update your subscription preferences or to unsubscribe: https://lists.gnucash.org/mailman/listinfo/gnucash-user ----- Please remember to CC this list on all your replies. You can do this by using Reply-To-List or Reply-All.
