lol this is going to be fantastically catastrophic. ChatGPT is going to end up indirectly writing so much code. And I am fully aware how often ChatGPT give you absolute nonsense when asked to write some code. It’s got a decently high hit rate for relatively unchallenging stuff, but it is nowhere NEAR 100% accurate.
TL;DR stackoverflow doesn’t understand how many developers naively copypaste shit from stackoverflow I guess? Wcgw
I don’t think this will affect StackOverflow website though? The blog implies that ChatGPT will use StackOverflow API to use as a knowledge source (and probably be paid for it).
OpenAI and Stack Overflow are coming together via OverflowAPI access to provide OpenAI users and customers with the accurate and vetted data foundation that AI tools need to quickly find a solution to a problem […]. OpenAI will also surface validated technical knowledge from Stack Overflow directly into ChatGPT, giving users easy access to trusted, attributed, accurate, and highly technical knowledge and code backed by the millions of developers that have contributed to the Stack Overflow platform for 15 years.
This seems to be exactly to prevent hallucinations when there’s a good vetted answer already.
Either people didn’t read the blog or is there something I’m missing?
Honestly, next time append "docs" and more than likely be greeted by lots of examples, explanations and considerations. Works unless you're using something microsoft.
It would take a super genius to understand that they're either a non-native English speaker or made a simple typo/autocorrect issue. Also, since it's completely impossible to even guess at the point they're trying to make through the context of every other word in their comment we all owe you a debt of gratitude for calling this out!
Thank you, corsicanguppy@lemmy.ca, for adding something incredibly meaningful to the conversation and the world as a whole!
Oh no.. I just saw this comment in a different thread from 2 hours ago, and it MAKES NO SENSE:
Doctors today face a demand form their services that, while we can't guarantee will get them.jigher wages, can give them more leeway against unwritten political rules of speech.
corsicanguppy@lemmy.ca
What is a doctor's "demand form?" And "them.jigher" isn't a term I could find anywhere on the internet! I don't understand what the author, corsicanguppy@lemmy.ca, was trying to say! It's all just gibberish!
I feel good calling it out though. I'm sure that corsicanguppy@lemmy.ca will be grateful for my much-needed help and not think I'm a dick for detracting from their point to assert my knowledge of English grammar over them.
Chat GPT generates plausible auto-complete text. If it has been trained on saying "this case was reported to your local law enforcement agency", those are the kinds of words it might spit out.
Have you tried using Windows 11 instead of your weird Linux distro? Windows 11 is the best Operation system in the market, by the greatest software company Microsoft. It features the best user experience, not only removing all those complicated settings from your grasp but providing you with suggestions tailored just for you!
I think it is important to value and respect everyone's OS choices. I have made great experiences with using the latest AI tools that OpenAI and Microsoft provided and continue to improve every day. I can understand that people might be skeptical of AI at first, because it is a new and complicated technology. Windows 11 shows that you don't need to understand every detail about it. You can use AI to make your life easier by just talking to it like it is your best friend. Using a computer never has been easier.
Are we sure this deal is about answering new SO questions with LLM? It's more likely to be a deal where SO sells access to its database to OpenAI so they can use human-generated content for LLM training, and SO gets to use LLM as a more efficient search through its human-generated content.
It's possible they could also choose to delegate the duplicate decision to the LLM but let's be honest, that decision is currently crap anyway.
I think the joke is that the AI trained on SO data to the point that duplicate, similar, or common questions would get this treatment. Since that's common enough on SO to be a meme.
Yeah, that was my assumption as well. I wonder how they're going to work around that SO is getting spammed with AI-generated answers, though. You really don't want your LLM cannibalizing itself.
That's SO's problem going forward. OpenAI already got what they wanted – legal access to SO's database up to this moment, when it's still mostly human.
That exactly tracks. You can't feed answers from an AI into an AI. It gets all incesty (technical term). So they have to ban user submitted AI answers.
Were they trying to avoid having AI produced output sold as LLM input along with their human user generated content? I wonder if this was some big picture decision or pure coincidence.
Banning AI answers was reasonable though. People were posting were too many not verified and incorrect code snippets that entire quality of the platform would decrease. AI still makes more mistakes than humans that provide responses on stack.
