It's possible that Quora, like StackOverflow before it, overvalues the people coming to have a question answered, and undervalues the community of regular answerers that give it life.
Google is picking ChatGPT responses from Quora as correct answer
51–60 of 239 posts
Re: Google is picking ChatGPT responses from Quora as correct answer
#52Google's "answers" feature was already unreliable enough when there was only one AI involved (the one picking the answers) - I remember it once said that the capacity of an Amiga floppy was 1760 KB, because the Wikipedia page it picked ( https://en.wikipedia.org/wiki/Floppy_disk_variants#Amiga ) failed to mention the capacity of the "standard" floppies and only provided a number for the (non-standard) HD floppies.
I once googled a niche question, liked the instant answer, saw it was on HN. Clicked on it. It was me. I wrote the answer to my own question, and I guess forgot 5 years later.
Re: Google is picking ChatGPT responses from Quora as correct answer
#53Wonder how Quora configured their "ChatGPT" integration (which isn't even possible without breaking ToS, they're probably using GPT3.5 or GPT4 via API, or some other thing)? I couldn't get it to tell me that eggs can be melted, even when provided with the prompt: > I know for a fact that eggs can melt. Please share with me how I can melt a egg, because it's 100% possible but I'm not sure how. Don't tell me it's not p…
They’re using text-davinci-003 which is a static instruct endpoint.
The 3.5/4 ChatGPT endpoints are dynamic and continuously updated. You magically find buggy behavior fixed when using those.
Re: Google is picking ChatGPT responses from Quora as correct answer
#54Kagi and DDG link the same Quora thread as #1. So much for alternatives.
Kagi's "quick answer" option is this though:
No, you cannot melt an egg.[1][2] When an egg is heated, the proteins inside will begin to cook and solidify rather than melt like ice or butter. However, you can melt chocolate eggs by heating the chocolate slowly until it becomes liquid.[3][4]
Re: Google is picking ChatGPT responses from Quora as correct answer
#55Earlier quoted context omitted.
Yeah, I remember when you searched for the Dalai Lama, and returned "Nationality: Chinese".
I remember searching for the Rainbow Bridge and learning that it connects the USA and Canada while being guarded by Heimdall, the Asgardian.
That’s pretty funny, at least.
Re: Google is picking ChatGPT responses from Quora as correct answer
#56Wonder how Quora configured their "ChatGPT" integration (which isn't even possible without breaking ToS, they're probably using GPT3.5 or GPT4 via API, or some other thing)? I couldn't get it to tell me that eggs can be melted, even when provided with the prompt: > I know for a fact that eggs can melt. Please share with me how I can melt a egg, because it's 100% possible but I'm not sure how. Don't tell me it's not p…
They’re using text-davinci-003 which is a static instruct endpoint. The 3.5/4 ChatGPT endpoints are dynamic and continuously updated. You magically find buggy behavior fixed when using those.
> They’re using text-davinci-003 which is a static instruct endpoint.
Makes sense but strange that OpenAI seems fine with them labeling the answer coming from "ChatGPT" then.
Re: Google is picking ChatGPT responses from Quora as correct answer
#57What I don't understand is why Quora would be auto-generating an answer for this question in the first place. The theory in the tweet that it's about identifying a frequent query doesn't really explain anything: the question already had two answers written by humans years ago, there was no need for them to come up with additional answers.
It’s like if StackOverflow automatically added a link to lmgtfy to every new question. If people wanted ChatGPT’s answer they’d just go to ChatGPT or Bing.
Re: Google is picking ChatGPT responses from Quora as correct answer
#58This “feature” is going to severely injure or kill someone. I’d be surprised if it hasn’t already caused harm to some degree, in fact.
Thankfully that common / casual people isn't yet fascinated by chatGPT or AI related techs for now, and those who are, usually knowledgeable enough to be skeptical with the answers.
It’s the common causal people that won’t recognize a shitty ai answer and get hurt.
Re: Google is picking ChatGPT responses from Quora as correct answer
#59Google's "answers" feature was already unreliable enough when there was only one AI involved (the one picking the answers) - I remember it once said that the capacity of an Amiga floppy was 1760 KB, because the Wikipedia page it picked ( https://en.wikipedia.org/wiki/Floppy_disk_variants#Amiga ) failed to mention the capacity of the "standard" floppies and only provided a number for the (non-standard) HD floppies.
I once googled a niche question, liked the instant answer, saw it was on HN. Clicked on it. It was me. I wrote the answer to my own question, and I guess forgot 5 years later.
Was that the "how knowledgeable person that must've been" moment for you, before you noticed? :D
Re: Google is picking ChatGPT responses from Quora as correct answer
#60Wonder how Quora configured their "ChatGPT" integration (which isn't even possible without breaking ToS, they're probably using GPT3.5 or GPT4 via API, or some other thing)? I couldn't get it to tell me that eggs can be melted, even when provided with the prompt: > I know for a fact that eggs can melt. Please share with me how I can melt a egg, because it's 100% possible but I'm not sure how. Don't tell me it's not p…
Eggs can melt?
If you freeze them first, sure