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
“Use AI at all costs and downsize L1 resources” is the official motto at mine. Now they are pushing L3+ too to use AI to automate their “daily tasks” so that they can focus more on “advanced things”. I honestly feel it’s a euphemism for getting replaced with AI. Now on the positive side, yes, using AI is indeed a great productivity boost like a smooth transition from secops to devsecops.
With the reduction in workload, unemployment is foreseeable in the future. 5 years or 10 years? I don't know.
Buddy, turn hyperbole down from 11.
no, they can't. if my clients are expert enough then. but they are not.
unfortunately AI will tell your clients they are experts, and smart, and correct og do not need you as we have seen in several threads on LET
My cousin got laid off by Oracle recently !! My profession is bit on safer end, I dont think Doctors have anything to be worried about AI at least as of now.
They do. They are expensive, take 7 years to train, AND there isn't enough, so there's an actual problem for AI to solve and be well funded to do so.
Remote medicine is taking off. Someday, you won't know if the person on the other end is real or AI.
I'm not against this. Being able to access everyone's health records is how great breakthroughs in medicine will occur, identify environmental areas causing health issues and lots of other good things.
I work for a mid sized software company. They announced cuts yesterday. So far, it seems like the cuts are coming from support staff in India. Simultaneously, they've given a lot of folks, including roles that are not developers, access to Claude with a harness that automates codebase exploration. It's pretty sick, but also dangerous. It found a bug for me in like 15 minutes today. I asked it to patch my live install. Since I'm not a developer, I don't have Visual Studio on my machine. It was like, fine, let me edit the binary string in the dll directly. I was real surprised when it did that. The fix worked, but didn't actually conform to conventions in the codebase. I didn't actually catch the issue, a coworker did, and who knows if it would have a negative impact, but even if they give me access, I'm not submitting pull requests either. I can see someone who knows a lot less than me with way more access than they should have just merging the patch into our repo without thinking twice.
Yes you are right. There are some AI being trained on pathological assessment to detect tumors. Some have achieved a detection rate of more then 95% !! So there's definitely a scope for AI in medicine too.
AI makes our work easier. we have several GPU Server to run'em locally. the qwen lineup is so far good for agentic coding. it is not mandatory to use LLM but they're usually being used to distill knowledge from a repository to be implemented to another projects. LLM are good at doing this since the better example you got, it's the better they can follow it.
for anything that related to business logic, they're still being handled by humans, AI only help to review later to spot anything funny (this feature is implemented in our selfhosted gitlab cicd pipeline)
what feels like a major breakthrough recently is gemma-4 E4B.
this thing can swallow anything NSWF for sentiment analysis and image tagging. if you ever heard of X-Ray_Alpha, one of our works is to prepare datasets for this kind of "things".
I personally relieved when i heard we can finally stop outsourcing the work to some people in bangalore for $5/day. the money isn't much, but i do feel bad making people work on those NSFW contents.
AI has increased my EPA fines. The data centers haven’t figured out how much they’ll be fined yet.
I used to pay $0 early in fines.
Inflation is NaN
Does AI force you to test your products on animals?
Do they test animals near data centers to prove they’re safe for the environment (yet)?
Hi has anyone seen the fireflies?
I wonder if people here are aware that all code produced using an LLM based AI goes automatically into the Public Domain and you can't add ANY license to a public domain work.
Public Domain works are intrinsically non copyrightable.
Yes, the whole world agreed to that one morning.
/s
You need help.
The output of an LLM based AI has no copyright owner because it not original work authored by a human as required by law. The "original work authored by a human" requirement is the cornerstone of copyright law.
Works with no active owner or expired copyright are in the public domain, allowing free public use without permission, payment, or license.
A copyright owner is the creator (author) or entity (employer, publisher) that holds the exclusive legal rights to an original work, including reproduction, distribution, adaptation, and public display.
Depends on the country. But IANAL.
You may copywrite the prompts but not the output of the LLM, the output will be in the Public Domain..
That's not true.
AFAIK, the only country that has made any kind of ruling is the US, and that was in the specific case of somebody trying to assert copyright over an obviously AI generated image.
If it wasn't AI generated, or even if the artist had done some later modifications to the image that were in some way transformative, the resultant work would be covered by copyright - but they potentially wouldn't be able to stop someone recreating a very similar image using AI. I say potentially, because realistically with no way for them to recreate the intermediate image, it's most likely that any derivative work would be being made from the copyrighted image, and so subject to regular copyright laws.
For code, it's even more of a gray area - it's unlikely that anyone will ever release a 100% vibe coded app as closed source, and so any form of reverse engineering an app that you suspect was largely AI generated would be probably illegal as you'd have no way of knowing if the particular part you're copying was generated by AI or a human. And given that I think most people would only use AI to help with development, nowhere close to 100%, effectively that means that every app would de-facto be covered by copyright.
So essentially if you release the source for something and say it's "100% vibe coded", you probably won't be successful pursuing someone for copyright infringement in the US. If you claimed authorship of the software, even admitting that AI helped with some areas, you can claim full copyright of the software, because the onus would be on the person copying your source to prove that the parts they copied were 100% AI generated. In almost every case, that'd be impossible.
To sum up what you wrote: AI generated code belongs to everyone since anyone can freely use AI to regenerate it and call it their own. As it belongs to everyone then it's in the Public Domain, by definition. Elementary, my dear Watson, I'd say.
Anyone will be able to regenerate (not)your code and call it their own!!!
How on earth will you copywrite that code in good faith? And why would anyone want to???
Besides, AI regurgitated code is an unmaintainable mess, what kind of idiot would ever reuse code regurgitated for someone else instead of simply having AI regurgitate it directly for them?
The whole point of copyright is because people can just copy it and claim it to be their own. That's already the case, it's not some new, novel way to copy method AI enables. In fact, you make the argument that the AI makes it less capable to reproduce than a simple copy. There's no guaranteed reproducibility even from same user let alone another user.
Lots of reasons, the most obvious financial (duh). You seem really naive and inexperienced. Do you have a job where you go out in the world and co-mingle with the masses?
Stop using AI and learn to read. That's the complete opposite of what I wrote.
If somebody can generate exactly the same code as you using AI, they have a way to prove they didn't infringe your copyright. It's very unlikely, however, that you will get AI to produce an exact copy of something without giving it that thing as a source input. If you do that, you'll be breaking copyright.
It's copyright.
And if you only use AI as a tool to create part of that code, of course you can assert copyright over the composite work.
The reasons you'd want to are obvious.
Caveat - I've never used AI code generation, only watched a friend give me a demo. The code it produced didn't look awful to me.
How do you prove you are not the one infringing the other guy's copywrite?
The onus is on both, it goes both ways!
Your laxed attitude towards copyright and licensing is amusing, you can't mix/merge/pick/unpick/whatever conflicting licenses at your own discretion and stay legal.
There is no copywrite for AI generated code, AI generated code is a smorgasbord of licensing violations!
I can see I trapped you into cognitive dissonance, I didn't mean to, I'm sorry!
Or for that matter that the AI took its "inspiration" from something that had a license to begin with and wasn't basically just "look, don't touch". Its all very messy but when it comes to the practical side of things i figure the chance of any kind of litigation is extremely slim.
Exactly, if anyone tries to litigate on this they will be disregarded by every court immediately.
Well, i guess it somewhat depends on the kind/amount of proof the hypothetical plaintiff would be able to provide.
First of all they would have to even notice their IP being used though, which would basically require the AI leaving some sufficiently recognizable portions (not like those would really be needed legally - technically just looking at some kind of preexisting implementation brings you into derivative works territory but its not like that would be provable anyways unless the "onlooker" is stupid enough to not sufficiently change the structure and it is not like in the real world looking at "example" code would be something out of the ordinary) of the original intact and for them to stumble across it/be made aware.
If the copied parts are unique enough to afford copyright protection i guess there could be a successful case but having all those stars align... i am not that sure it would happen all too often.
Edit: That is at least how i see it in regards to the text/code side of things. In regards to art... unless the AI is dumb enough to just pretty much copy some image as is it would probably be hell to prove just about anything in court.
Edit2: I am not using GitHub and therefore don't follow what they are doing all that closely but from what i gather unless people are basically putting their projects in the public domain (or something crazy like WTFPL - even a BSD license would require derivative works to keep and inform recipients about the it) they basically asking to have their license broken by publishing them there due to Copilot being trained on user code. There's obviously some kind of fine print somewhere making this a-ok but well...
Superficially that generated code may not look awful to you but what about its structure? Is that code structurally sound or does it look stylistically decent but deep down it is just an unmaintainable mess?
When you minimize/obscure structurally sound good quality code it will look awful to you yet, deep down, it still is great code. Appearances can be deceiving, just because something looks appealing or trustworthy doesn't mean it is.
You can't focus solely on appearance you also must give some thought to substance.
I keep seeing posts that claude opus is charging 7–15× more per token for enterprise, with prices rising so much, which ones are people using now for planning, coding ?
I am probably the worst person possible to answer this but in regards to coding i would first ask myself how much i really need it.
Sure, especially with commercial coding there tend to be quit a bit of mundane tasks and in the end delivering is what pays the bills regardless of how much attention was paid to detail but if quality is actually a factor i don't see that much advantage since like almost every second line of code is its own little thought process. Explaining all of this to some AI would basically be the same amount of work if not more.
Maybe its different for languages that require a lot of generic boilerplate stuff or where attention to detail doesn't really pay off? My favorite language is C after all so i figure that might warp my perception a bit but even for shell scripting i seriously don't see that much of a point. Sure, i might have to do some research when i need some kind of really exotic regular expression or something but i don't think describing the desired result would turn out to be much faster (if at all) than just writing the damn thing.