I get the general techbro sphere doing this because they're fucking idiots, but you are the maintainers of this technology. You know this is ultimately a language model built to synthesize information from pre-existing data sets. This is extremely fucking ghoulish.
I hate the promotion of the idea that we should be using generative adversarial networks as search engines or encyclopedias so much it's unreal, literal misinfo generation machines :agony-shivering:
ChatGPT is fun but also confidently wrong all the time. Using it as a knowledge source is one of the worst possible applications
The GPT4 announcement included a benchmark on how often the different GPTs lie to you. ChatGPT was 40% of the time.
GPT4 is down to 20% lies, which is definitely an improvement, but still a huge amount if you think about it at all.
Improved automated restaurant only does food poisoning 20% of the time. :so-true:
100% of the time or 0% of the time depending on your definition of lying
I think 90% which tanked their stocks upon announcing their AI lol
Yeah I asked it for sources on ekranoplans and I am pretty sure it made them up:
-
The Wing-In-Ground (WIG) Effect Craft: A Review
-
WIG Craft and Ekranoplans by Sergy Komarov
edit: I asked it again and it gave me more fake resources but one of them included a real researcher on the subject. The singularity is here!!!
-
However, they note that GPT-4 acts not confident in its answers it so people are less likely to fact-check results than with the previous version.
The more accurate it is, the more likely its mistakes go unnoticed.
I now have multiple professors who warn against using it for essays. Students are trying to write phytopathology papers using it.
You have to ask it not to do that explicitly. AI researchers call it "hallucinating" a response. You have to ask it to say "I don't know" if it doesn't know something.
The Bing one has different modes, creative, balanced, and precise. ChatGPT is seems on the creative side but an application using GPT does not need to be like that. On precise mode the Bing one regularly says it does not have enough info to say
like 80% of adults already can't use google effectively let alone an encyclopedia.
tfw these things have a ton of potentially useful applications but now that the VC people want to squeeze the lemon we just get a big fucking scam
While ChatGPT is very error-prone, lies, etc., it's still very functional and useful for various activities. People doing bulk spam text generation and other creative activities will only increase.
People who think they can replace things like encyclopedias and loan officers/underwriters will probably fuck themselves real hard.
That's a bubble that will burst once it actually needs to be used for anything mission-critical.
I think they'll end up making an enterprise version of it that draws from that organization's information.
It still doesn't understand context or logic, so the same problem remains.
I love how the question being posed in the video is something I would absolutely not ask a language model for advice about.
"Piranhas will eat weak or sick fish, but generally avoid healthy fish. Tie as many live minnows as possible to your fingers to keep them disinterested."
The funny thing is that if your search term is "very specific phrase" it doesn't turn up. When you search for "Stalin" it does in fact turn something up. Albeit there is a stark difference in using a search at a public facing website and Emailing them and getting a response it is an urban myth.
Especially if you could just go to an archive and look up the primary sources. It has to be a bit.
Here's the problem. Google used to be good. Now it sucks. Microsoft tried to do Bing by building a rival AI capable of crawling the web and building a relational database better than Google's and failed. So now we've got Microsoft rebranding their Search Engine as a Chat AI and... It doesn't suck!
I can look for things I had enormous trouble finding online previously and get the results I actually want. This is a good thing! I like good things! I'm going to probably use ChatGPT as a search engine going forward because it works better than Google or DuckDuckGo or whatever else. Do not try to stop me, because I like Search Engines That Return The Results I'm Actually Looking For.
I just recognize that this is a term-limited experience. Eventually, ChatGPT will get slogged down by adware and bloatware and shitty SEO. And then it'll stop working properly.
At that point, I'll have to either go back to Alta Vista or Yahoo or whatever else doesn't suck, or find a Sexy New Thing to use as a search engine.
Sorry, haters. But that's the Brave New World we live in.
You can create a new email address and register under that. Its not like burner email addresses are hard to come by.
For what I'm doing, it hardly matters.
this is so fucking awful, this is going to kill people. with any luck it won't be many, but this could lead to some major catastrophes
"No way to stop ChatAI from telling children to drink bleach," says company that owns ChatAI.
Imagine if it learns off the tide pod memes and recommends eating them to young kids lmao
Yeah they constantly cover their asses by saying "hey this has limitations, don't count on this thing being right all the time" but then go on to integrate it into search engines, and in the case of this one an app for low vision/blind people that previously relied on volunteers (Be My Eyes). Also all the hype surrounding this thing definetly undermines any attempt at conveying the limitations to a general public.
On a different note, what is the business model here? This thing is probably expensive as hell to run.
I think they'll make an enterprise version that'll be expensive, in the way enterprise email and messaging is expensive.
If you check out OpenAI's API pricing it's actually not terrible. I bet MS gets some kind of sweetheart pricing on top of it. They're probably siphoning your conversation with the AI to advertisers. Search is somehow the most profitable kind of software, so anything that gets more people doing searches while also providing better data is more profitable.
On a different note, what is the business model here? This thing is probably expensive as hell to run.
We train it to do our jobs.
I don't think it's useful as an encyclopedia but the Bing one is useful at gathering info from websites. It gives you the link to each page it checks so I find that to be an effective way to search for things.
I agree but when in the hell are they gonna legalize piranhas?
this could be such useful technology for automating thinking labor but instead it is going to be used as a means to control "The Truth" because loud mouth tech bros are going to cite it as an official source.
I trust BazingaChat over the human because humans are more prone to errors! Human! :so-true:
(very similar arguments are already made by bazingas about why their Teslas are safe, actually)
I literally saw something on Twitter today where a person claimed that the H.G. Welles Stalin interview never happened, their source being a chatGPT blurb that claimed it was a hoax. Mind numbingly stupid. Everyone in the comments was linking to the interview in the Welles archive. Brainless
Your opinions your sources Disinformatskaya :maybe-later-kiddo:
My generative adversarial networKkk The Akashic record :gusano-perplexed:
