[GNC] Sharing Some Changes With the GnuCash Community

[Adrien Monteleone]

Thanks John,

I understand. I'm following these developments closely. But I'm opting 
out as much as possible for now. Things may smooth out and become more 
rational at some point, but I'm not a bandwagon or Tulip Mania sort of 
fella.

This tech isn't what people think it is, and certainly not what its 
marketers claim it to be. (especially the ones fear mongering that they 
don't understand it themselves - that's pure conjob level stuff) They 
might get lucky and succeed in 'faking it until they make it' but I'm 
not holding my breath. These are not answer machines, and I simply do 
not see how or why probabilistic word generators should be used for 
anything but novelty kicks and entertainment, no matter how fantastical 
they appear to be.

Certainly, giving up individual agency is irresponsible and just plain 
nuts. I don't think these systems are dangerous in themselves. The 
danger lies in mistaking them, and using them, for something they are not.

I trusted the GnuCash team to not do so, but now that I know ya'll are 
already using this, that trust is challenged. I'm sure this isn't the 
only project in that boat. (indeed, I'm aware of Mozilla's use of it.)

Lots of folks and companies are being stung and bitten in the backside. 
That's going to continue for some time. I'm not interested in being 
collateral damage in the interim and want no part, as much as I can 
avoid or excise it, in anything to do with this tech. I'm going to sit 
this one out, watch the bubble burst, the mania subside, and ride out 
the destructive wake as best I can.

A full-disclosure of a breakpoint version/minor-version would 
respectfully be appreciated.

Regards,
Adrien

On 5/26/26 8:10 PM, John Ralls wrote:
> Adrien,
> 
> I think you’re being a bit too absolute. While we did get a few PRs full of AI slop a couple of weeks ago (not sure what the submitter was up to, most of the time he closed the PR a few minutes after opening it) and I banned the submitter after a week, other submitters—including Chris Lam—use AI responsibly to help them code. I haven’t yet started to do that, but I probably will soon. Responsible use isn’t vibe-coding: The developer has to understand the result and review it for correctness and sound design.
> 
> You can be sure that every major professional shop, Apple, Google, Microsoft, and the rest, are already moving quickly in that direction. No doubt you’ve seen the hype about Anthropic’s Mythos finding hundreds of bugs and being “too dangerous” to release generally. It won’t take long for that level of capability to come from other frontier labs. OpenAI is already claiming that they have it and they’re not restricting release. The Firefox folks say they’ve fixed everything that Mythos found and they also used AI assistants to get the fixes done as quickly as possible.