For Anyone Using AI in Their Workflow

zester

zester

To everyone on these forums who appreciates AI you have allies here. Don’t let trolls scare you away from sharing your work.

The reality is that many of the loudest anti-AI voices don’t even write code. From experience, AI can be the difference between a one-person project never shipping, or actually releasing something useful. In open source, you’re often working alone. If AI helps you get a port working, compile code, fix an error, generate documentation, create examples, explain unfamiliar code, or help you learn a new domain then use it!

In the professional world you will be using AI tools, alongside your coworkers. They save time, and nobody treats it as controversial. Most of the hostility only exists online. If someone gives you a hard time, ask them to share their own work before criticizing yours. Constructive criticism is valuable empty gatekeeping isn’t.
 
zester said:
Provide a link to your projects repo's
Click to expand...
On the other side of the coin, in many ways this is also a good way of exposing the LinkedIn "AI tech-bros" who can barely write a hello world program (with or without AI) and yet are still preaching about how Software Engineers are obsolete. I'm sure you have come across them in your travels. These guys honestly are funny, 6 months ago they were obsessively discussing about bitcoin before pivoting!

Meritocracy and a demonstration of ability always adds more weight to someone's argument.

(I'm not religiously for, or against what we are currently calling AI. I am liking the progress its making though)
 
This is how we respond to toxic people while protecting our peace.

1771281711193.png
1771281677470.png
 
I appreciate this perspective.

At the end of the day, AI is a tool. It doesn’t replace understanding, responsibility, or review — it simply accelerates certain tasks. In open source, especially when you’re working alone, time and energy are real constraints. If a tool helps you prototype faster, understand unfamiliar code, or draft documentation more efficiently, that’s practical, not ideological.

For me personally, as a non-native English speaker, this tool is extremely valuable for communication. It helps me express technical ideas clearly and participate more confidently in discussions. I’ve been transparent about using it (as I mentioned in this post), and I don’t see a reason to hide that.

These are the projects I’ve started while using it: [R4R & freebsd-6.2-r4r-experiments].
And this is what the project attempt looked like before using this tool.

Healthy skepticism is valuable, and tools should always be used critically, not blindly. But discouraging people from sharing their work simply because they used AI does not strengthen the community.

Quality, transparency, and review matter more than the method used to draft something.
 
Isoux said:
I appreciate this perspective.

At the end of the day, AI is a tool. It doesn’t replace understanding, responsibility, or review — it simply accelerates certain tasks. In open source, especially when you’re working alone, time and energy are real constraints. If a tool helps you prototype faster, understand unfamiliar code, or draft documentation more efficiently, that’s practical, not ideological.

For me personally, as a non-native English speaker, this tool is extremely valuable for communication. It helps me express technical ideas clearly and participate more confidently in discussions. I’ve been transparent about using it (as I mentioned in this post), and I don’t see a reason to hide that.

These are the projects I’ve started while using it: [R4R & freebsd-6.2-r4r-experiments].
And this is what the project attempt looked like before using this tool.

Healthy skepticism is valuable, and tools should always be used critically, not blindly. But discouraging people from sharing their work simply because they used AI does not strengthen the community.

Quality, transparency, and review matter more than the method used to draft something.
Click to expand...

That’s some seriously advanced work, well done! I’m going to follow you. Maybe you’ll follow me back.

This is what I used AI for today. I’m working on adding a Chromium browser widget to my C++ GUI toolkit, which uses Wayland, Vulkan, Skia, Yoga, and CEF3:
https://chromiumembedded.github.io/cef/

I couldn’t get the widget to compile and run without crashing. AI helped me track down the issue it turned out to be an undocumented API change now it works.

1771282543077.png
 
Thank you — I’ve followed you back.

Those are some impressive domains you’re working in, congratulations. That’s a pretty serious stack.

I’m mostly focused on low-level work, closer to the hardware side, so my experience with AI has been a bit different. I’ve found it generally quite helpful and often surprisingly accurate, but of course it does make mistakes. You need a solid understanding of what you’re doing, otherwise you can both go off in the wrong direction and end up spending more time chasing trivial issues instead of saving time.

Used carefully, though, it’s a very useful tool.
 
Isoux said:
Thank you — I’ve followed you back.

Those are some impressive domains you’re working in, congratulations. That’s a pretty serious stack.

I’m mostly focused on low-level work, closer to the hardware side, so my experience with AI has been a bit different. I’ve found it generally quite helpful and often surprisingly accurate, but of course it does make mistakes. You need a solid understanding of what you’re doing, otherwise you can both go off in the wrong direction and end up spending more time chasing trivial issues instead of saving time.

Used carefully, though, it’s a very useful tool.
Click to expand...


AI isn’t a fire-and-forget tool that's for sure. To use it effectively, you need a good understanding of what you’re trying to build.

I’ve found that if you want to build something complex "say a search engine" and you don’t know much about how search engines work, you can use AI to break it down into smaller pieces. Start by asking for the major components, then ask what each part does, then what those parts are made of, all the way down to the underlying algorithms.

Once you understand the lowest-level pieces, you can reverse the process: start from the algorithms, combine them into systems, and gradually build back up to a full application. How useful the final result is depends on your goals, but the real value is the learning process. Now you know how a search engine is designed, with a working example. This isn’t really about search engines it’s about a method. When your knowledge of a domain is weak, AI can help you deconstruct it, understand it, and then build it.
 
zester said:
The reality is that many of the loudest anti-AI voices don’t even write code.
Click to expand...
I like running efficient code; AI assistance on code doesn't target that, while usually only being initially sought for to make something easier. Easier =/= quality.

I'm not loud about anti-AI, but don't usually applaud it either :p
zester said:
In the professional world you will be using AI tools, alongside your coworkers.
Click to expand...
I guess I'd be fine with AI use while being paid while using it, but it's too unpredictable, financed, and centralized for me to consider using it personally. It wouldn't vibe right for me to implement code from some big company's bias machine for free or even advertise for em :p
 
Espionage724 said:
I like running efficient code; AI assistance on code doesn't target that, while usually only being initially sought for to make something easier. Easier =/= quality.

I'm not loud about anti-AI, but don't usually applaud it either :p

I guess I'd be fine with AI use while being paid while using it, but it's too unpredictable, financed, and centralized for me to consider using it personally. It wouldn't vibe right for me to implement code from some big company's bias machine for free or even advertise for em :p
Click to expand...

I don’t know, I’ve seen some pretty efficient code written by AI. I have an archive of GLSL shaders for generating shapes using signed distance fields; performance and quality wise, they rival some of the most efficient 2D graphics algorithms that exist.

I guess I'd be fine with AI use while being paid while using it, but it's too unpredictable, financed, and centralized for me to consider using it personally. It wouldn't vibe right for me to implement code from some big companies bias machine for free :p
Click to expand...

That’s not really how being an employee works. I don’t get to have an opinion about using AI at work, and even if I still had objections, I wouldn’t be allowed to share them. My point is if your employed and your job requires the use of AI, you really have no other option than to be fine using AI. Unless you would rather just be unemployed.

Unpredictable? What kind of AI are we talking about generative AI? Sure, but that’s a feature. And it’s not true that AI is centralized. See https://huggingface.co/ there are over 2 million open-source AI models. The main barrier is financial: training a new model from scratch (think GPT-3 scale) is like $35,000 Otherwise, the expense is mostly operational. Alot of peope when they think ai they only think of ChatGPT, Claude, Grok, ... but there are soooooo many more. And sooo many types.
 
zester said:
In the professional world you will be using AI tools, alongside your coworkers.
Click to expand...

I am a software architect in a German state enterprise, there is no mass usage of AI tools, and sharing of data with service based AIs is prohibited.
Developers can choose to use AI tools in an anonymous way. No code can be shared with these tools, it is a strict violation of rules of work.

The projects we develop go through rounds of multiple security certifications etc.
 
Zare said:
I am a software architect in a German state enterprise, there is no mass usage of AI tools, and sharing of data with service based AIs is prohibited.
Developers can choose to use AI tools in an anonymous way. No code can be shared with these tools, it is a strict violation of rules of work.

The projects we develop go through rounds of multiple security certifications etc.
Click to expand...

It's the exact opposite in the united states.
 
Zare said:
I am a software architect in a German state enterprise, there is no mass usage of AI tools, and sharing of data with service based AIs is prohibited.
Developers can choose to use AI tools in an anonymous way. No code can be shared with these tools, it is a strict violation of rules of work.

The projects we develop go through rounds of multiple security certifications etc.
Click to expand...

Oh a German state enterprise. So that’s government-owned. Is it federal, state, or municipal?
Even if your particular department doesn’t use AI tools, they are definitely being used somewhere within the organization.

1. Deutsche Bahn (federal government majority-owned)
  • Uses AI for predictive maintenance of trains and tracks
  • AI timetable optimization and delay prediction
  • Computer-vision inspection of infrastructure
2. Bundesagentur für Arbeit (Federal Employment Agency)
  • Uses machine learning to match job seekers with positions
  • Automated document classification and case triage
3. German customs & tax authorities
  • Fraud detection models
  • Risk scoring for audits and inspections
4. Healthcare (public insurers / hospitals)
  • Radiology image analysis
  • Hospital capacity prediction and patient flow modeling




 
Really, you're inputing my posts into AI? What the actual f.

I am not giving you any more information, this is the exact reason why our organization does not cooperate with United States based software services.
 
Zare said:
Really, you're inputing my posts into AI? What the actual f.

I am not giving you any more information, this is the exact reason why our organization does not cooperate with United States based software services.
Click to expand...

It was from google, but I guess technically your correct.
 
You must log in or register to reply here.
Back
Top