@simonmic Hi Simon, I just saw your hledger announcement and read a little, including your justification for AI coding assistance.
I must say that hl looks interesting and have noted it for the future, and from your feed I see we share values. Yet again though, I find someone like me, whose adoption of LLMs I don't get.
So I wanted to pick up on an area in your list of reasons. You may not wish to debate this, or even read and that's fine. I'm no bully or ideologue.
1/
@simonmic Thanks for reading and considering. I agree it's a big complex topic, which is why I don't believe it being useful for software development is necessarily sufficient justification for that use, or anywhere in fact that we find it has utility.
Whatever our views on that, I don't doubt it will have utility in some contexts. But the downsides of that use, whatever it is, go far beyond the immediate context.
1/2
@simonmic
So the complexity is in fact part of the difficulty for anyone trying to make an objective assessment of the overall justification for it's use in any particular context.
So yes it's a very hard one to debate. 😃
2/2
> I don't believe it being useful for software development is necessarily sufficient justification
Right you are @happyborg , I agree. I think we must try to judge the overall impact of a particular use - is it net positive or net negative ? This can be hard of course.
What do you think of other possible "justifications" mentioned in the doc, eg
- "The result provides enough value to society to outweigh the cost"
- "We use AI once to build an efficient tool so many people will use AI less"
@simonmic I think we agree that the first is the key question. While very hard to be definitive, I think the onus is on proponents to make the case and answer the issues raised by others who are skeptical.
First do no harm is a good maxim IMO, and I've seen enough move fast and break things to believe that is in general a wrong approach and the last people to be trusted are those who own and hype this tech. I look to independent academic research, but don't hold much hope (see climate change).
> I think the onus is on proponents to make the case and answer the issues raised by others who are skeptical.
I agree here too. I'm not a proponent of the current AI wave - if I had been asked if we should do it, I would have said no. Given that it is happening, does it make sense to try and operate within current conditions, and look for ways to have an incremental net-positive influence ?
At least, I can discuss and try to increase my personal clarity on the issues. Thanks for helping.
> "We use AI once to build an efficient tool so many people will use AI less"
I don't think this is credible and not something I'm seeing people saying, so tbh I haven't considered it. It doesn't seem to take account of either where we are or of human nature in general.
@simonmic Ah, thanks for clarifying. I understand the logic better.
I don't think many will use your tool over AI once they are in the habit. It's too easy to waste hours typing dumb prompts than to sit and think for five minutes, or find the right tool for the job.
You and I will do it, but we're a minority and let's face it, it's in the interests of purveyors to maximise token use (cf. the attention economy which has become a danger, recognition of which is resulting in age verification 🤦♂️).
@simonmic
> It's too easy to waste hours typing dumb prompts than to sit and think for five minutes, or find the right tool for the job.
I've seen anecdotal evidence for this and it makes totally sense based on my understanding of how most humans behave (and I've seen it in myself too 
, although pre LLM).
EDIT: a desired effect of those selling #LLMs is to undermine individual agency and reduce cognitive capacity in users. That's not necessary for this, but will make it much much worse.
@simonmic regardless, I see engagement with this as helping the owners of #LLMs, and when I consider who they are and what they are going to do (what's starting already is nothing to what's coming - see my boost about #Copilot inserting ads into GitHub PR descriptions)...
... I return to my base position that while there will of course be some benefits, this is a very bad thing for individuals and society in general, and in so many ways I would have to write a thesis to cover them.
@happyborg Sure, here's a few examples of people leveraging #hledger in their AI use:
https://forum.plaintextaccounting.org/tag/ai/12
https://x.com/CookResearcher/status/2032247670903828596
https://x.com/sudo_goreng/status/2036789525616414970
https://x.com/motdotla/status/2029247616828276838
@simonmic thanks. I can't follow folk on X having left years ago. I see the pta forum has several, but those are I expect a small fraction of hl users who are also using #LLMs such as Claud, while that's still an affordable (subsidised) service that hasn't begun #enshitification.
Anyone beginning to rely on these services, or on the models that their owners allow you to run locally, is setting themselves up to be captured and exploited as we have now seen happen many times.
happyborg
Simon Michael
