Although we have seen a small decline in traffic, in no way is it what the graph is showing (which some have incorrectly interpreted to be a 50% or 35% decrease). This year [2023], overall, we’re seeing an average of ~5% less traffic compared to 2022.
And yet they didn’t show us a rectified or accurate graph.
They’re lying
Says who? ChatGPT?
ChatGPT is lying too, and is also really confident about it.
Says corporations trying to protect their investment.
Is this a facts or feelings response?
Is this a trust a company to announce it’s dying response?
You got evidence for that bold statement, ooooor
An ML trusting a corporation, orrrr?
It will recover when AI learns sneering condescension.
Wonder where chatgpt will get its training data in the future, as it’s known not to extrapolate well. Where will it learn new frameworks, languages, … from?
Its going to starve itself.
Let it die
I doubt it ever scraped SO, otherwise all the answers would be smth along the lines: “I cannot answer this question due to low quality effort!” closes browser window
The documentation?
The docs. It’s what it does now a lot of the time I’ve noticed.
Auto generated docs since devs don’t document?
Chatgpt, look at this repo and write docs
Honestly it’s petty good about doing that. Already had similar tooling options but it does a generally good job of making docs for non devs assuming good naming are used in the methods
Yeah the smaller the project the less effective this is.
But even learning from the source code is pretty effective.
That works when the docs are good and clear. Otherwise, we’ll have to revert to communicating with each other for brief periods while the chat-bots train themselves on the new data.
A lot of models are being trained on “synthetic” data now, right?
Even when a parrot learns to parrot a parrot, the first parrot still has to be taught.
I’ve been on SO like ten times altogether since ChatGPT came out. It’s so much nicer than the condescending pricks of Stack Overflow. My favourite is when some genius links a question as a duplicate of something that’s vaguely similar.
The problem is that fewer people are generating the new content for gpt to slurp up.
Free information stops working when people stop caring about creating it.
Yeah, I’ve thought about that as well. Doesn’t mean I’m gonna miss Stack Overflow specifically. Perhaps something better replaces it when AI gets poisoned by its own output too much.
I think not. The problem isn’t in a service, the problem is in people.
Sure, that’s why a good service doesn’t create an environment which supports people being dicks. Unfortunately, that’s exactly what SO does with the way it works.
Yup. Stopped answering questions as soon as they did that.
my first thought about it too
What do you ask ChatGPT? I rarely get a correct answer really, it’s all made-up bullshit usually.
I usually have good answers with only minimal bullshit when it comes to coding. If not, it at least points me in the right direction.
Down voted. If you’re just going to post a screenshot, you must include a link to the source. This is a link sharing platform.
I thought I was going to be rickrolled but this is better.
Closed as duplicate
Now that’s just a signal for training reinforcement
The irony of it all is SO making deals for AI to scrape their site for machine learning but in doing so more people are using chatgpt and copilot more because it’s easier and just as accurate. AI/ML is really going to destroy these websites and yet it’s the websites signing off on their own death sentences.
It’s called selling out. I doubt they have any illusions about the future of these platforms, they just don’t care as long as they can cash out.
What I’m still not sure of is this… When these websites die will the LLMs stagnate with no new data to use for training or will they somehow keep up with new technology and eliminate the need for certain basic questions that would originally have required human input?
The ideal result? LLMs are just early versions of much better things that come later.
The unlikely result: we develop a separate human curated internet somewhere, complete with verification that a human wrote every bit. Basically verifiable digital id and signing on everything. Maybe.
The probable result: the internet turns to shit as AIs are trained on content created by AIs.
I think they see the writing on the wall and they’re trying to make a buck while they still can.
It’s also around the time they did away with their jobs board thing isn’t it? I got a job through it in 2021 and somewhere after that they sunset it, which was an insane business decision because it was the best job search platform out there.
Chat GPT is wonderful as a search engine for SO. It regurgitates the answers in a format easier to incorporate into your own project.
The thing I’m worried about is a lack of new answers. You need data to train an LLM, what to do if nobody is producing it?
It already uses the docs and API references
I typically visit stack overflow when encountering edge cases and bugs not covered by the documentation.
If it has learned from the source code you’d be surprised how good it is at that as well.
For docs it’s better, for other stuff it’s way worse than a human.
It’s a shame so much stuff is locked up in discord these days.
yeah, discord the the true black hole of information
Do you mean trained on the source code, or having the source code in context?
Trained on the source code.
Yeah could also clone the repo and use copilot against it.
But who writes the docs now
Ha! I have an AI for that! Gotcha!
…
On, but the AI trains now on other docs that I used an AI to write…
…
Oh shit…
Isn’t this when Safari and Firefox started blocking third party trackers by default?
That wouldn’t stop page views from being counted.
I’m raising their traffic,by slowly poisoning my answers & questions.
Why would you want to also make Stack Overflow worse?
I didn’t consent to ChatGPT using my stuff. So they get 💩.
But so do the Stack Overflow users
Wut. You consented to CC BY-SA. That’s consenting to everyone using your work, including for profit companies and nonprofits. This is a good thing.
OpenAI doesn’t attribute the original training data, so not consented?
That’s valid, if true. Are you sure somewhere on their website they don’t have a link back to SE?
Their training data is their secret
I’ve tried to delete my content, I wasn’t allowed. So I replaced all my content with text about not consenting to this whole thing. I got banned for 2 months “to think about my behavior”, or something like that. So I’ve set a reminder for the date…
But can ChatGPT downvote answers into oblivion?
No it’s the one it replies with
Interesting. I actually thought of it as a replacement for Google. With Google search being broken for years it’s the only easy way to get information now.
yeah it’s replacement for google only if you don’t care if any of information received is accurate at all
Which… is google
duckduckgo and searx using duckduckgo work fine
That’s what I use too. Can recommend.
They are really bad for getting correct information and putting it in context: Information literacy and chatbots as search
There are other search engines.
Most of them have the same problem though. The more specific you are the worse the results.
Idk, I rarely have any problems searching for things.
Guilty <-
Yeah. When I need additional insights on a difficult technical configuration, it’s nice to be able to speak to an artificial insufferable dipshit, rather than a real human insufferable dipshit.
The AI ones continue helping me even after I explain to them how they come across to real humans. (I do my best not to mention it to the insufferable Human dipshits, of course.)
