New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
It was easy, because you didn't write it. Duh.
Also, your point is moot because of:
So less vulnerabilities will be found, so lower quality commits will take place in the closed source. You're conflating quality and exploitation.
Fuck panicked pussies. First comment on Slashdot:
And...
It proves me right!
I see you're fighting reception problems, try rotating the antenna sometimes it improves reception.
You got it all wrong!
Using AI to make it better... that's a good one, clueless but funny. You're such a joker, unintentional but still funny.
From 2026 on AI will be harvesting close to zero new quality code from open source projects, the good quality code base available for AI training has been frozen pre 2026.
From now on AI bots will have to be trained using the shit produced by its previous versions and shit produced by other AI bots, it's coprophagy all the way!
A mix of shit with pre 2026 good quality but increasingly outdated code.
Proven more wrong?
Open source wouldn't stay unpatched at 2026 levels, it would get vulnerabilities fixed. It's open source, can't put the code back in the machine, it just gets forked and continued by the users.
My whole complaint is your terrible and confusing writing and you double down with fucking antenna reference. Take your medicine or just fuck off and let the adults talk.
Use it to cover my ass in healthcare
I see this trend happening more and more with COSS (closed open source software). AI can rewrite your software pretty effectively to a completely different language. Take that and growth opportunities afterwards, you'll see a lot of opportunists come along.
The digital market as we know it is about to shift, heavily.. sooner than later imho.
I am a copywriter and I definitely noticed I have less jobs and also, there are 60 copywriters applying for a freelance job that has been posted an hour ago.
I noticed some businesses cut back their budget for copy significantly. They just invest in AI more.
That said: I’m not going to be replaced. There are still customers who appreciate quality copy. Every month, I have at least a few requests to fix AI-slop copy. I refuse and only write entirely new copy. Fixing AI-slop costs more time and is not effective.
I do write AI copy for anyone who wants to pay me for it. You just can’t expect the same level of quality, leave alone results in terms of conversion.
On the other side, I use AI for my hobby projects for vibecoding. I have a Cursor and MiniMax subscription. Now I can build an in-browser radio automation I have been dreaming of. But the timing of the transitions still is not right sometimes and honestly, the code sucks.
Who is going to understand what a t1, t2, vi1 and a v2 is? Even when I set up the rule to only use variable and functions name that refer to a function, it messed up. It keeps shortcutting function names and just explains everything is a comment above it.
Fun fact: MiniMax fails to understand the code Cursor wrote. Developers who write complex or reliable software, are not going to be replaced anytime soon. You don’t want to have your software down at 3AM at night and debug without sleep, without having a clue what you are doing. Especially if that software needs to be online again when offices are open.
Do you know how software licensing works?
The copywrite owner has the power to duo license, apply as many different licenses as he wants, change the license terms or even remove it altogether. He holds the legal rights to it and can do whatever he wants.
So, take note:
The AI free ride on Open Source is over!
OVER
EDIT: that final remark you made was very adult, do you know the meaning of the word irony? There's no better irony than the involuntary kind and you just incurred in it.
EDIT2: Please accept my apologies in advance for using concepts that, apparently, are too advanced for you, that is a problem that is up to you to fix so start working on it.
The bigger issue is that if people rely on AI to do all the work for them rather than using it as a starting point from which they then do their own research (i.e. Google / StackOverflow / other comp sci resources), they lose their technical abilities over time because they’re not actively practicing anymore.
I think there is an analogy to be drawn with the US military industry’s policy of replacing parts rather than fixing them. Companies like Boeing, Lockheed Martin etc refuse to give engineers in the US army access to their detailed schematics so when something breaks down they can’t repair it, even if it’s just one small capacitor or IC, they have to order the whole part from the manufacturer and swap it.
So over the years talented senior engineers who have superb electronic troubleshooting and soldering skills get off the force due to frustration with this situation, better pay in the private sector or normal retirement, while the juniors that come in don’t get to learn and practice this technical skill set because they’re “taught” the new way of replacing without repairing and the seniors are gone.
This actually becomes a national security issue because what happens when you’re deployed on a mission and you have to repair something on the fly but can’t, and you can’t order parts either?
(Louis Rossmann has 2-3 great interviews on the topic of right to repair with engineers in the military where they mention all of these problems. Highly recommend watching them on his YouTube channel.)
Circling back to vibe coding, you can see the same thing happen when AI generated slop full of security vulnerabilities gets deployed in production. Even here on LET we’ve had some people showcase their vibe coded web apps and they got pwned.
So yeah, AI can be useful if you don’t switch your brain off and if you use a variety of non-AI resources to check what it outputs. But it seems that only a minority of people are doing that and it’ll probably get worse as time goes on.
Life is good when using them for any work that gives low return as reality how one work priced in SEA market.
Do you? One of the main points of open source is having the source code to modify it yourself. The limitations, if any, are generally on distribution.
You are conflating open source with specific types of licenses. Stop wasting our time.
Edit: in this case, the source is allowed to be redistributed.
I don't work in the tech industry and i lost my job because of AI. The AI hype is real
Honestly AI changed everything for me personally.
My programming background before AI was bash scripts. That's it. Not a developer by any stretch.
In the last month I completed two complex WHMCS plugins that I would have never been able to build on my own. Not "it would have taken longer", I literally could not have done them. The gap between what I could build and what I needed to build was too big.
I also built areyoureplaceable.com from scratch in a few days. It is a free tool that gives you a personalized AI risk score for your career, business, or website, not a generic percentage based on job title, but something that takes into account your age, location, financial situation, skills, and then gives you actual advice about what you can do in the AI era.
So when people ask how AI affects my work, it gave me capabilities I simply did not have before. For someone who is already a strong developer the answer might be "it makes me faster." For me the answer is "it made me a builder."
*vibe coder
I wouldn't call myself a vibe coder. I provide the logic, the algorithms, the fixes and the improvements. The AI does one thing: writes the code.
In my IT career I've had the pleasure of working with very talented programmers. Their code is good, but their logic is sometimes flawed. Many times I've had to sit down with a programmer and explain that what they're doing is technically correct, but it's the wrong way to solve the problem or there is a better way to do it.
Vibe coding is when you hand the model an outcome and hope "make it work, make it pretty, fix it." I work the other way around. I decide the architecture, the data flow, the constraints, the edge cases. The AI implements against that spec. When something breaks, I diagnose the actual cause; I don't just re-prompt until the error goes away.
Call it AI-assisted, call it logic-led but i don't think it's vibe coding.
Well yes it's helping me and my team push code way faster. We also basically stopped hiring juniors
It already has> @TimboJones said:
You just don't get it, do you?
After a license change you'll still have access to old versions of the source code but not to newer versions, you will not get access to code for new features and improvements, you'll be stuck with the the old versions of the code and that's it.
A side note, I still find you amusing... for a while. You are way too simplistic and not very agreeable but I find you amusing.
The hype is real but the accomplishments are falling short.
EDIT: Struggling Shoe Retailer Allbirds Pivots To AI, Stock Explodes More Than 700%. Quite an accomplishment,
.
So now you have a canon to shoot yourself in the foot instead of a slingshot.
Someone can come up with a high-level algorithm but implementing it properly, and by that I mean efficiently and securely, is a whole different beast. Two prime examples: high-performance computing and any encryption or digital signature scheme (@forest can chime in with relevant horror stories). Or generally just having a web app that's not just pretty but really fast, that takes a lot of careful tweaking and it's not something that generative AI can spit out, one example I like referring people to is https://www.mcmaster.com/.
tl;dr your point that they're stuck on outdated code ignores the point of using open source in the first place, taking over updates yourself. That isn't prevented with AI, it's ENABLED more so now than previously. Non-coders can update their own code or fork it and/or other interested parties take over future updates. This happens ALL THE FUCKING TIME when the original dev fucks off or does something so dumb the community forks it. Devs lose interest all the time. You haven't made any argument AI is speeding this up in any way. Other people take over abandoned repos or people move to currently supported products.
Take your meds.
Just switched from boring coding to sleeping coding !
You manage to be simplistic and anal-retentive at the same time, amazing! It's a bit like sucking and blowing at the same time. it's a delicate balance.
I wonder how you pull that off.
Blah, blah, blah, whining like a bitch because you failed to make a point.
Exactly. An LLM can vomit out basic apps pretty well. It can whip up UIs, implement parsers, etc. But if you're actually trying to do anything even remotely sophisticated, it will be worse than nothing. Try writing an efficient CUDA program with Claude, or doing kernel development without introducing a million bugs and abusing the ABI. It's not going to happen.
My current favourite thing while I do security audits is I can immediately tell by "vibe" (xd) whether something was written by a LLM and is using a borderline insane unauthenticated supa/firebase backend that will gladly both vomit out the entire contents of the database, expose sql queries over fetch(), and also gladly accept modifying any field including my own user fields to admin. Record was admin AND rce in 3 seconds, entirely human, hand editing a request from f12 panel
That's the biggest issue with LLMs, in my opinion. People can spit out code so fast that seems to work, but statistically, even the best LLMs produce code with almost 4x more vulnerabilities than humans. And we're not talking humans who are actually good programmers, but average programmers. Just think about how god-awful the average programmer is in terms of their security discipline, and realize that even Claude is nearly 4x worse.
It's probably even worse than that considering LLMs also have a tendency to pull in massive, bloated frameworks with a huge attack surface on their own. They'll be the ones to pull in fucking Electron when you're making a file manager or terminal emulator.
Now, I do think LLMs are good for code review. Not alone, but as an extra set of eyes doing a really strange mix of static and symbolic analysis with quasi-human reasoning. They tend to find bugs that humans don't (and vice versa).
You did miss the whole point.
Your future clients wont need you anymore.
They can just code it themselves.
Or a toddler can own a businesss to challenge you in the future.
And you wont have jobs anymore.
The funniest part is some of these people at API based token pricing would have spent more yelling repeatedly at opus bad "fix every security bug look at every file" prompts than to have paid someone in a low cost of living country $50 to build the thing in the first place with wordpress