maybe, ig time will show, i still will try to use it when required, but it’s another tool, one day it will be forgotten and obsolete, and people will find something else to be more dependent on
It’s not a tool. It’s an exploit.
Not everyone is comfortable or going to lean on the easy solution
But you do what you feel is best for you, in the end you are the one affected
Except by how it affects the environment, or people’s jobs, or how the technology kills people and no one regulates it (or, for that matter, how it makes searching the internet increasingly difficult because of all the slop…)
But in the end I was trying to help and make suggestions on how it might affect you, and it’s up to you to decide what to do with that (just as long as you don’t use it to make issue reports on Godot, then it’s my problem)
and that’s why i give big thank you, cuz you’re trying to give new perspective, as i’ve said - i’m not completely dependent on AI, and i just use it as another tool that helps, but time will show if it was good or not…
Please remain civil and keep the discussion on-topic, aligned with the original post.
by myself, no need to make forum posts, no need to research more, just move on to next thing
this got me thinking some, another perspective i havent thought about is that it can speed up for looking thru information in docs and such. Not my cup of tea (i like the godot docs
i came from SDL2 so its much easier than SDL docs to look through for sure) but definitely a more unique use case
Though i guess the downside to that is there’s less information online for others/the AI itself to learn from
discord feels like a 0.5 of that, where information is no longer ‘online’, but in a small community and in long message threads, weird to think about
also didnt expect this to blow up so hard, been enjoying reading through all the msgs, trying on different shoes
i think something important to note is, you cant change someone on the internet very easily, its a lot more productive to learn their perspective, and practice constructive conversation. Hence the ‘open discussion’ title ![]()
My feelings i stated in the original post are the same, but its interesting to hear the other side of things, thanks even those who got shot down some for putting your perspective into the mix
a lot can be learned from leaving conversations open
Exactly, it’s always important to engage with different perspectives and not try to force someone to think a certain way
That’s what i said about speeding up searching information and looking through it, i don’t say that AI should write code, i’m quite opposite, but the moment you gather information - you can spend less time on it and focus on thinking than searching
I’m convinced that there are / should be better technologies out there that simply use indexed search (lucene core - a part of solr, elastic search, postgres token search - and quite possibly ported by google back in the day) and summarizing (?) without pretense prompting to do good research with.
But thank you for being a great sport in this conversation and thank you for listening & reading all responses fairly.
Another thing that is important to consider is how these tools filter information, and how much control one might lose over that
Search engines of course already filters a lot and controls what we can access, but they are also regulated in various ways
This is a risk even if it’s not done maliciously where the search engine or model filters out certain information
Solr/Lucene is fully open source and boots up using a basic java -jar command. From that point onward you can index and all the content you want, wholly under your control.
The thread was about polluting repos with slop. There’s a very slippery slopE (see what I did there) from “It’s just a search tool” to “I’ll generate just a little bit of example code” to fully vibe coded PR garbage. The more you’re of the “my head hurts when reading docs” disposition - the more likely you’ll slip on that slope.
I don’t buy any “trust me, I don’t use it to generate code” rhetoric. It’s straight up lying in most cases. This shit is aggressively advertised precisely for that purpose, it’s highly addictive and tricks you into false sense of “productivity gains”. It’s a hard to resist combo for average opportunistic lot. That’s what created the slop flood problem in the first place.
So what more do we need to look out for when (relatively) new to this forum? I cannot distinguish between a robot, a grifter and someone in need of a genuine explainer.
Alas, I know not which zombie thread to point to (yet)..
ok so maybe a bot..
Is not that difficult
Things to look out for:
- A post that starts with “Copy and paste this into a post on the Godot forum to ask for support”. (Yes, that happened.)
- A post that looks like a Readme and contains uncommon emojis.
- Wording that is too flowery. (Keeping in mind that some people just use LLMs to translate their questions to English. You can usually distinguish these because word choices are odd.)
- Code that is FULL of verbose, full sentence comments
- Code that is divided up into logical sections with comments that just contain dashes in them.
- Code that is overly complex for what it is doing.
- Different coding styles in the same script.
- Debug or print statements in code that include Emojis.
- Replies that are overly complimentary in a creepy way. (“You are so right, so what do you suggest I do?”)
- Replies that communicate a complete lack of understanding of their own code when you offer a suggestion on how to fix something.
- A new user who has a really attractive female avatar.
Some of these signs remind me of the (joking) tips that AlbertaTech gave on her video on how to attract a programmer boyfriend. I wonder if the AI has been trained on joke/satirical sources by mistake.
It’s trained on everything. I was hesitant to post that list because it’s going to make the LLMs smarter. Like one of those adaptive killer Sentinels in X-Men: Days of Future Past who can adapt to any power thrown at them after the first attack and you can only run from.