- cross-posted to:
- [email protected]
- [email protected]
- [email protected]
- cross-posted to:
- [email protected]
- [email protected]
- [email protected]
You must log in or register to comment.
If this is true, then we should prepare to be shout at by chatgpt why we didnt knew already that simple error.
ChatGPT now just says “read the docs!” To every question
Hey ChatGPT, how can I …
“Locking as this is a duplicate of [unrelated question]”
And then links to a similar sounding but ultimately totally unrelated site.
Stack overflow was the pioneer of hallucinations.
Nah, it just marks your question as duplicate.
Already had that happen with perplexity, like, no mate, I’m asking you.
Honestly, that wouldn’t be the worst thing in the world.
You joke.
This would have been probably early last year? Had to look up how to do something in fortran (because fortran) and the answer was very much in the voice of that one dude on the Intel forums who has been answering every single question for decades(?) at this point. Which means it also refused to do anything with features newer than 1992 and was worthless.
Tried again while chatting with an old work buddy a few months back and it looks like they updated to acknowledging f99 and f03 exist. So assume that was all stack overflow.
This message brought to you by chatgpt bot.
Take all you want, it will only take a few hallucinations before no one trusts LLMs to write code or give advice
[…]will only take a few hallucinations before no one trusts LLMs to write code or give advice
Because none of us have ever blindly pasted some code we got off google and crossed our fingers ;-)
It’s way easier to figure that out than check ChatGPT hallucinations. There’s usually someone saying why a response in SO is wrong, either in another response or a comment. You can filter most of the garbage right at that point, without having to put it in your codebase and discover that the hard way. You get none of that information with ChatGPT. The data spat out is not equivalent.
That’s an important point, and and it ties into the way ChatGPT and other LLMs take advantage of a flaw in the human brain:
Because it impersonates a human, people are more inherently willing to trust it. To think it’s “smart”. It’s dangerous how people who don’t know any better (and many people that do know better) will defer to it, consciously or unconsciously, as an authority and never second guess it.
And the fact it’s a one on one conversation, no comment sections, no one else looking at the responses to call them out as bullshit, the user just won’t second guess it.
Your thinking is extremely black and white. Many many, probably most actually, second guess chat bot responses.
Split segment of data without pii to staging database, test pasted script, completely rewrite script over the next three hours.
When you paste that code you do it in your private IDE, in a dev environment and you test it thoroughly before handing it off to the next person to test before it goes to production.
Hitting up ChatPPT for the answer to a question that you then vomit out in a meeting as if it’s knowledge is totally different.
Which is why I used the former as an example and not the latter.
I’m not trying to make a general case for AI generated code here… just poking fun at the notion that a few errors will put people off using it.
We should already be at that point. We have already seen LLMs’ potential to inadvertently backdoor your code and to inadvertently help you violate copyright law (I guess we do need to wait to see what the courts rule, but I’ll be rooting for the open-source authors).
If you use LLMs in your professional work, you’re crazy. I would never be comfortably opening myself up to the legal and security liabilities of AI tools.
Yeah but if you’re not feeding it protected code and just asking simple questions for libraries etc then it’s good
I feel like it had to cause an actual disaster with assets getting destroyed to become part of common knowledge (like the challenger shuttle or something).
Maybe for people who have no clue how to work with an LLM. They don’t have to be perfect to still be incredibly valuable, I make use of them all the time and hallucinations aren’t a problem if you use the right tools for the job in the right way.
The last time I saw someone talk about using the right LLM tool for the job, they were describing turning two minutes of writing a simple map/reduce into one minute of reading enough to confirm the generated one worked. I think I’ll pass on that.
confirm the generated one worked. I think I’ll pass on tha
LLM wasn’t the right tool for the job, so search engine companies made their search engines suck so bad that it was an acceptable replacement.
Honestly? I think search engines are actually the best use for LLMs. We just need them to be “explainable” and actually cite things.
Even going back to the AOL days, Ask Jeeves was awesome and a lot of us STILL write our google queries in question form when we aren’t looking for a specific factoid. And LLMs are awesome for parsing those semi-rambling queries like “I am thinking of a book. It was maybe in the early 00s? It was about a former fighter pilot turned ship captain leading the first FTL expedition and he found aliens and it ended with him and humanity fighting off an alien invasion on Earth” and can build on queries to drill down until you have the answer (Evan Currie’s Odyssey One, by the way).
Combine that with citations of what page(s) the information was pulled from and you have a PERFECT search engine.
That may be your perfect search engine, I jyst want proper boolean operators on a sesrch engine that doesn’t think it knows what I want better than I do, and doesn’t pack the results out with pages that don’t match all the criteria just for the sake of it. The sort of thing you described would be anathema to me, as I suspect my preferred option may be to you.
So my company said they might use it to improve confluence search, I was like fuck yeah! Finally a good use.
But to be fair, that’s mostly because confluence search sucks to begin with.
You’re describing Bing Chat.
And google gemini (?) and kagi’s LLM and all the other ones.
Yeah, every time someone says how useful they find LLM for code I just assume they are doing the most basic shit (so far it’s been true).
This. I use LLM for work, primarily to help create extremely complex nested functions.
I don’t count on LLM’s to create anything new for me, or to provide any data points. I provide the logic, and explain exactly what I want in the end.
I take a process which normally takes 45 minutes daily, test it once, and now I have reclaimed 43 extra minutes of my time each day.
It’s easy and safe to test before I apply it to real data.
It’s missed the mark a few times as I learned how to properly work with it, but now I’m consistently getting good results.
Other use cases are up for debate, but I agree when used properly hallucinations are not much of a problem. When I see people complain about them, that tells me they’re using the tool to generate data, which of course is stupid.
The quality really doesn’t matter.
If they manage to strip any concept of authenticity, ownership or obligation from the entirety of human output and stick it behind a paywall, that’s pretty much the whole ball game.
If we decide later that this is actually a really bullshit deal – that they get everything for free and then sell it back to us – then they’ll surely get some sort of grandfather clause because “Whoops, we already did it!”
Have you tried recent models? They’re not perfect no, but they can usually get you most of the way there if not all the way. If you know how to structure the problem and prompt, granted.
People keep saying this but it’s just wrong.
Maybe I haven’t tried the language you have but it’s pretty damn good at code.
Granted, whatever it puts out needs to be tested and possibly edited but that’s the same thing we had to do with Stack Overflow answers.
I’ve tried a lot of scenarios and languages with various LLMs. The biggest takeaway I have is that AI can get you started on something or help you solve some issues. I’ve generally found that anything beyond a block or two of code becomes useless. The more it generates the more weirdness starts popping up, or it outright hallucinates.
For example, today I used an LLM to help me tighten up an incredibly verbose bit of code. Today was just not my day and I knew there was a cleaner way of doing it, but it just wasn’t coming to me. A quick “make this cleaner: <code>” and I was back to the rest of the code.
This is what LLMs are currently good for. They are just another tool like tab completion or code linting
We already have those near constantly. And we still keep asking queries.
People assume that LLMs need to be ready to replace a principle engineer or a doctor or lawyer with decades of experience.
This is already at the point where we can replace an intern or one of the less good junior engineers. Because anyone who has done code review or has had to do rounds with medical interns know… they are idiots who need people to check their work constantly. An LLM making up some functions because they saw it in stack overflow but never tested is not at all different than a hotshot intern who copied some code from stack overflow and never tested it.
Except one costs a lot less…
This is already at the point where we can replace an intern or one of the less good junior engineers.
This is a bad thing.
Not just because it will put the people you’re talking about out of work in the short term, but because it will prevent the next generation of developers from getting that low-level experience. They’re not “idiots”, they’re inexperienced. They need to get experience. They won’t if they’re replaced by automation.
First a nearly unprecedented world-wide pandemic followed almost immediately by AI taking over the world, man it is really not a good time to start out as a newer developer. I feel so fortunate that I started working full-time as a developer nearly a decade ago.
Dude the pandemic was amazing for devs, tech companies hiring like mad, really easy to get your foot in the door. Now, between all the layoffs and AI it is hellish
So, the whole point of learning is to ask questions from people who know more than you, so that you can gain the knowledge you need to succeed…
So… if you try to use these LLMs to replace parts of sectors, where there need to be people that can work their way to the next tier as they learn more and get better at their respective sectors, you do realize that eventually there will no longer be people that can move up their respective tier/position, because people like you said “Fuck ‘em, all in on this stupid LLM bullshit!” So now there are no more doctors, or real programmers, because people like you thought it would just be the GREATEST idea to replace humans with fucking LLMs.
You do see that, right?
Calling people fucking stupid, because they are learning, is actually pretty fucking stupid.
Where did I say “Fuck 'em, all in on this stupid LLM bullshit!”?
But yes, there is a massive labor issue coming. That is why I am such a proponent of Universal Basic Income because there are not going to be enough jobs out there.
But as for training up the interns: Back in the day, do you know what “interns” did? And by “interns” I mean women because sexism but roll with me. Printing out and sorting punch cards. Compilers and general technical advances got rid of those jobs and pushed up where the “charlie work” goes.
These days? There are good internships/junior positions and bad ones. A good one actually teaches skills and encourages the worker to contribute. A bad one has them do the mindless grunt work that nobody else wants to. LLMs get rid of the latter.
And… I actually think that is good for the overall health of workers, if not the number (again, UBI). Because if someone can’t be trusted to write meaningful code without copying it off the internet and not even updating variable names? I don’t want to work with them. I spend too much of my workday babysitting those morons who are just here there to get some work experience so they can con their way into a different role and be someone else’s problem.
And experience will be gained the way it is increasingly being gained. Working on (generally open source) projects and interviewing for competitive internships where the idea is to take relatively low cost workers and have them work on a low ROI task that is actually interesting. It is better for the intern because they learn actual development and collaboration skills. And it is better for the staff because it is a way to let people work on the stuff they actually want to do without the massive investment of a few hundred hours of a Senior Engineer’s time.
And… there will be a lot fewer of those roles. Just like there were a lot fewer roles for artists as animation tools stopped requiring every single cell of animation to be hand drawn. And that is why we need to decouple life from work through UBI.
But also? If we have less internships that consist of “okay. good job. thanks for that. Next time can you at least try and compile your code? or pay attention to the squiggly red lines in your IDE? or listen to the person telling you that is wrong?”? Then we have better workers and better junior developers who can actually do more meaningful work. And we’ll actually need to update the interviewing system to not just be “did you memorize this book of questions from Amazon?” and we’ll have fewer “hot hires” who surprise everyone by being able to breath unassisted but have a very high salary because they worked for facebook.
Because, and here is the thing: LLMs are already as good, if not better than, an intern or junior engineer. And the companies that spend money on training up interns aren’t going to be rewarded. Under capitalism, there is no reason to “take one for the team” so that your competition can benefit.
This is already at the point where we can replace an intern or one of the less good junior engineers. Because anyone who has done code review or has had to do rounds with medical interns know… they are idiots who need people to check their work constantly.
Do so at your own peril. Because the thing is, a person will learn from their mistakes and grow in knowledge and experience over time. An LLM is unlikely to do the same in a professional environment for two big reasons:
-
The company using the LLM would have to send data back to the creator of the LLM. This means their proprietary work could be at risk. The AI company could scoop them, or a data leak would be disastrous.
-
Alternatively, the LLM could self-learn and be solely in house without any external data connections. A company with an LLM will never go for this, because it would mean their model is improving and developing out of their control. Their customized version may end up being better than their the LLM company’s future releases. Or, something might go terribly wrong with the model while it learns and adapts. If the LLM company isn’t held legally liable, they’re still going to lose that business going forward.
On top of that, you need your inexperienced noobs to one day become the ones checking the output of an LLM. They can’t do that unless they get experience doing the work. Companies already have proprietary models that just require the right inputs and pressing a button. Engineers are still hired though to interpret the results, know what inputs are the right ones, and understand how the model works.
A company that tries replacing them with LLMs is going to lose in the long run to competitors.
Actually, nvidia recently announced RAG (Retrieval-Augmented Generation). Basically the idea is that you take an “off the shelf” LLM and then feed your local instance sensitive corporate data. It can then use that information in its responses.
So you really are “teaching” it every time you do a code review of the AI’s merge request and say “Well… that function doesn’t exist” or “you didn’t use useful variable names” and so forth. Which… is a lot more than I can say about a lot of even senior or principle engineers I have worked with over the years who are very much making mistakes that would get an intern assigned to sorting crayons.
Which, again, gets back to the idea of having less busywork. Less grunt work. Less charlie work. Instead, focus on developers who can actually contribute to a team and design meetings.
And the model I learned early in my career that I bring to every firm is to have interns be a reward for talented engineers and not a punishment for people who weren’t paying attention in Nose Goes. Teaching a kid to write a bunch of utility functions does nothing they didn’t learn (or not learn) in undergrad but it is a necessary evil… that an AI can do.
Instead, the people who are good at their jobs and contributing to the overall product? They probably have ideas they want to work on but don’t have the cycles to flesh out. That is where interns come into play. They work with those devs and other staff and learn what it means to actually be part of a team. They get to work on really cool projects and their mentors get to ALSO work on really cool projects but maybe focus more on the REALLY interesting parts and less on the specific implementation.
And result is that your interns are now actually developers who are worth a damn.
Also: One of the most important things to teach a kid is that they owe the company nothing. If they aren’t getting the raise they feel they deserve then they need to be updating their linkedin and interviewing elsewhere. That is good for the worker. And that also means that the companies that spend a lot of money training up grunts? They will lose them to the companies who are desperate for people who can lead projects and contribute to designs but haven’t been wasting money on writing unit tests.
-
deleted by creator
I got an email ban.
1609 hours logged 431 solved threads
Reddit/Stack/AI are the latest examples of an economic system where a few people monetize and get wealthy using the output of the very many.
Technofeudalism
Mmm this golden goose tastes delicious!
First, they sent the missionaries. They built communities, facilities for the common good, and spoke of collaboration and mutual prosperity. They got so many of us to buy into their belief system as a result.
Then, they sent the conquistadors. They took what we had built under their guidance, and claimed we “weren’t using it” and it was rightfully theirs to begin with.
You really don’t need anything near as complex as AI…a simple script could be configured to automatically close the issue as solved with a link to a randomly-selected unrelated issue.
So vanilla stack overflow?
That’s the joke
I’m slow.
Based and same-here-often…pilled
Oh I didn’t consider deleting my answers. Thanks for the good idea
BarbraStackOverflow.I’d be shocked if deleted comments weren’t retained by them
I think the reason for those bans is that they don’t want you rebelling and are showing that they don’t need you personally, thus ban.
Of course it’s all retained.
Isn’t it amazing that places like this built on user support and contribution turn around and pull a “we don’t need you”?
Isn’t that illegal in most countries?
In Europe GDPR gives you the right to have all your data deleted. All you do is send in a request and SO has to remove everything of yours, not just anonymize it. There are some exceptions for legal reasons, eg where financial transactions are involved, but comments should not be exempt.
They have been un-deleting after they ban.
They are also retained by anyone who has archived them., like OpenAI or Google. Thus making their AIs more valuable.
To really pull up the ladder, they will have to protest the Internet Archive and Common Crawl, too. It’s just typical right-wing bullshit; acting on emotion and against their own interests.
Maybe we should replace Stack Overflow with another site where experts can exchange information? We can call it “Experts Exchange”.
Expert Sex Change?
Yes, next to Pen Island
deleted by creator
I mean that’s just been a schoolyard joke for ages
Also a market there. Especially among programmers. You might be onto something.
Among Rust devs? Absolutely!
I agree with your idea. I will be launching a website where users can share content. It will be free once knowledge should be free and we will make money by selling data…umm selling user data…umm selling T-shirts I guess. That should be enough to keep the servers running.
You don’t want that shit done by an amateur
codidact … Stack overflow had a mass exodus of mods a 2-3 years ago and a some of them made codidact.
Any discussion on making it ActivityPub enabled?
I didn’t see any, but would be curious if anyone else had.
Maybe there I can ask where to find a good pen supplier.
Messages that people post on Stack Exchange sites are literally licensed CC-BY-SA, the whole point of which is to enable them to be shared and used by anyone for any purpose. One of the purposes of such a license is to make sure knowledge is preserved by allowing everyone to make and share copies.
That license would require chatgpt to provide attribution every time it used training data of anyone there and also would require every output using that training data to be placed under the same license. This would actually legally prevent anything chatgpt created even in part using this training data from being closed source. Assuming they obviously aren’t planning on doing that this is massively shitting on the concept of licensing.
CC attribution doesn’t require you to necessarily have the credits immediately with the content, but it would result in one of the world’s longest web pages as it would need to have the name of the poster and a link to every single comment they used as training data, and stack overflow has roughly 60 million questions and answers combined.
IF its outputs are considered derivative works.
Ethically and logically it seems like output based on training data is clearly derivative work. Legally I suspect AI will continue to be the new powerful tool that enables corporations to shit on and exploit the works of countless people.
Maybe but I don’t think that is well tested legally yet. For instance, I’ve learned things from there, but when I share some knowledge I don’t attribute it to all the underlying sources of my knowledge. If, on the other hand, I shared a quote or copypasta from there I’d be compelled to do so I suppose.
I’m just not sure how neural networks will be treated in this regard. I assume they’ll conveniently claim that they can’t tie answers directly to underpinning training data.
Share Alike
I can’t wait to download my own version of the latest gpt model
It does help to know what those funny letters mean. Now we wait for regulators to catch up…
/tangent
If anything, we’re a very long way from anything close to intelligent, OpenAI (and subsequently MS, being publicly traded) sold investors on the pretense that LLMs are close to being “AGI” and now more and more data is necessary to achieving that.
If you know the internet, you know there’s a lot of garbage. I for one can’t wait for garbage-in garbage-out to start taking its toll.
Also I’m surprised how well open source models have shaped up, its certainly worth a look. I occasionally use a local model for “brainstorming” in the loosest terms, as I generally know what I’m expecting, but it’s sometimes helpful to read tasks laid out. Also comfort in that nothing even need leave my network, and even in a pinch I got some answers when my network was offline.
It gives a little hope while corps get to blatantly violate copyright while having wielding it so heavily, that advancements have been so great in open source.
See, this is why we can’t have nice things. Money fucks it up, every time. Fuck money, it’s a shitty backwards idea. We can do better than this.
Hear me out. Bottle caps.
Nah, I can’t imagine the Fallout that would cause
'Nuff said!
You can be killed with steel, which has a lot of other implications on what you do in order to avoid getting killed with steel.
Does steel fuck it all up?
Centralization is a shitty backwards idea. But you have to be very conscious of yourself and your instincts to neuter the part that tells you that it’s not to understand it.
Distributivism minus Catholicism is just so good. I always return to it when I give up on trying to find future in some other political ideology.
This has nothing to do with centralization. AI companies are already scraping the web for everything useful. If you took the content from SO and split it into 1000 federated sites, it would still end up in a AI model. Decentralization would only help if we ever manage to hold the AI companies accountable for the en masse copyright violations they base their industry on.
Anarchosyndicalism ftw.
I despise this use of mod power in response to a protest. It’s our content to be sabotaged if we want - if Stack Overlords disagree then to hell with them.
I’ll add Stack Overflow to my personal ban list, just below Reddit.
Once submitted to stack overflow/Reddit/literally every platform, it’s no longer your content. It sucks, but you’ve implicitly agreed to it when creating your account.
While true, it’s stupid that things are that way. They shouldn’t be able to hide behind the idea that “we’re not responsible for what our users publish, we’re more like a public forum” while also having total ownership over that content.
Eventually, we will need a fediverse version of StackOverflow, Quora, etc.
Those would be harvested to train LLMs even without asking first. 😐
At this point I’m assuming most if not all of these content deals are essentially retroactive. They already scrapped the content and found it useful enough to try and secure future use, or at least exclude competitors.
They scraped the content, liked the results, and are only making these deals because it’s cheaper than getting sued.
Can they really sue (with a chance of winning) if you scrape content that’s submitted by users? That’s insane.
Honestly? I’m down with that. And when the LLM’s end up pricing themselves out of usefulness, we’ll still have the fediverse version. Having free sites on the net with solid crowd-sourced information is never a bad thing even if other people pick up the data and use it.
It’s when private sites like Duolingo and Reddit crowd source the information and then slowly crank down the free aspect that we have the problems.
The Ad sponsored web model is not viable forever.
The Ad sponsored web model is not viable forever.
a thousand times this
I’d rather the harvesting be open to all than only the company hosting it.
Assuming the federated version allowed contributor-chosen licenses (similar to GitHub), any harvesting in violation of the license would be subject to legal action.
Contrast that with Stack Exchange, where I assume the terms dictated by Stack Exchange deprive contributors of recourse.
SO already was. Not even harvested as much as handed to them. Periodic data dumps and a general forced commitment to open information were a big part of the reason they won out over other sites that used to compete with them. SO most likely wouldn’t have existed if Experts Exchange didn’t paywall their entire site.
As with everything else, AI companies believe their training data operates under fair use, so they will discard the CC-SA-4.0 license requirements regardless of whether this deal exists. (And if a court ever finds it’s not fair use, they are so many layers of fucked that this situation won’t even register.)
But users and instances would be able to state that they do not want their content commercialized. On StackOverflow you have no control over that.
You can state what you don’t want, but no one will be paying attention. Except maybe the LLM reading your posts…
Yup. Laws are only suggestions until you get caught.
I suspect it isn’t even illegal, but I’m not an expert.
Not fediverse, but open-source and community run: https://codidact.com
Oh this looks decent. British non-profit, I like it. Registering.
Smells too much like duo-lingo. Here, everyone jump in and answers all the questions. 5 years later, ohh look at this gold mine of community data we own…
This was actually the whole original point of Duolingo. The founder previously created Recaptcha to crowd source machine vision of scanned books.
His whole thing is crowd sourcing difficult tasks that machines struggle with by providing some sort of reason to do it (prevent spam at first and learn a language now)
From what I understand Duolingo just got too popular and the subscription service they offer made them enough money to be happy with.
Everything you write on here is public. There’s nothing stopping anyone from using that data for training
We needed it a few years ago.
Can we pass on quora?
Federated yahoo answers.
how is feddi formed
Arguably, they need to do way instain mother> who kill thier babbys. becuse these babby cant frigth back?
It’s important to remember that it was on the news this mroing a mother in ar who had kill her three kids.
Too much, can’t figure it out
We already have the SO data. We could populate such a tool with it and start from there.
primary use for AI is self destructing your website.
Pleasing tech illiterate sharholders
Remember when adding the word blockchain to an Iced Tea company’s name caused share prices to jump?
I dunno. AlphaFold 3 is pretty big.
The enshittification is very real and is spreading constantly. Companies will leech more from their employees and users until things start to break down. Acceleration is the only way.
Accelerationism is like being on a plane and wishing it crashes when one of the engine fails.
That’s a terrible analogy, implying the wish that everyone on the plane dies if one engine fails.
It’s like an airline company has been complete shit for decades, wanting to see them fail fast so that a better airline company can take their place.
Begun, the AI wars have.
Faces on T-shirts, you must print print. Fake facts into old forum comments, you must edit. Poison the data well, you must.
I mean we aren’t even fighting AI, we are still fighting greedy little turds
Problem is, it still results in turning the Internet to shit. We just do it manually to preempt the AI doing it.
