When trying to use ChatGPT for science, it straight up makes up citations (and makes false statements as if they are fact). It has done this every time I have tried it for science and when you call it out, it apologizes for doing it. How about just not making shit up? 🧪
Comments
Nowadays it returns blog posts about the most basic mistakes beginners make because that brings more ad revenue.
It's a probability based random word generator tuned to everything ever written, whether true or not.
It's not a research tool.
Imagine a dice with words on it, as many sides as you need for the thought.
The inside is a series of carefully carved criss-cross tunnels
Now, you could get a machine to carve the patterns so that you'll never get "the the", but you might get "the cat" instead of "the a".
When I look at some of the emergent projects, even "autogen" from Microsoft, especially the prompts they write, it's almost
AIs are trained by people just like Elon. Well, except for the apology part.
It's done its job if you look at its output and say "Yup, that is definitely a research paper."
Gemini confirmed this.
are you going to disavow it?
FUCK grokAI
Buy books!
Study
Learn
Apply
It doesn't learn FROM us, but it definitely learns ABOUT us.
I’ve had to call it out. Minor impact in its behavior. Seems like a downgrade.
why do they think it's worth investing billions in is more interesting question. Hyper efficient propaganda machine? Stealing simpler jobs? Crap marketing/entertainment? But definitely not actual intelligence
https://www.msn.com/en-us/news/technology/we-made-a-mistake-in-not-being-more-transparent-openai-secretly-accessed-benchmark-data-raising-questions-about-the-ai-model-s-supposedly-high-scores-after-sam-altman-touted-it-as-very-good/ar-AA1xEWyN?ocid=ientp&apiversion=v2&noservercache=1&domshim=1&renderwebcomponents=1&wcseo=1
So useful...
I'll never use it for any science related stuff because of that.
For coding it's fine.
You ask these low tier bots a question and they scrape wrong answers.
I've no intention of using it, but I think it's important to understand the strengths and weaknesses.
Just stop using it - like, people, how did you write before!
Whatever AI is under the covers of the browser I asked a musical question of the other day, it made up different wrong answers for the same fairly simple question. And it was quite authoritative about it, cited incorrect supporting detail. Pretty scary.
Yikes.
My Garmin GPS and Apple seem to occasionally want me to drive into Puget Sound, or in one case to have me drive across pasture. So maybe not full control there either?
Spoilsport, I know.
There are low life ass-licking sycophants like Musk programming this shit!
Plagiarism has never been ok.
It's completely useless, it fails at even the most straightforward and simple tasks.
Isaac Asimov wrote a couple of short stories showing what happens to Humanity when we rely TOO heavily on automation/robots.