There are guardrails, at all journals. And we now need to upload raw data. People have been fucking around in science publications since Photoshop was invented.
this post was submitted on 29 May 2026
94 points (100.0% liked)
Fuck AI
7191 readers
1265 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
founded 2 years ago
MODERATORS
Yeah its really bad. Its incredibly easy to be gaslit using gpts. And they do generate seemingly usable output. And I don't mean like, using them to analyze data. I mean just some code in an ipynb. And even there its incredibly rotten. Knowing and understanding what you are doing is the entirety of the point, and the whole AI industry misunderstands this. The code, the analyses, even the results, they're all secondary to understanding why something works the way that it does. Your individual comprehension is the part that matters.
Its incredibly easy to be gaslit using gpts
no scientist is going to be gaslit by ChatGPT
I just spent two weeks undoing something a scientist did while being gaslit by a a gpt. I promise they do.
They are human beings too, and many don't understand this technology better than the public. It's easy to fall for the marketing and listen to these chat bots uncritically if you don't know better
Good times create weak people but we are also experiencing bad times so hopefully more people will become better than the ones who don't.
May I recommend the book The Fourth Turning? It might give you a glimmer of hope (sprinkled with a dash of salt)
Unscientific bullshit. It's so dumb that famous Nazi Steve Bannon made a movie about it.
Good guard rails are books and studies published before generative science became a thing.
Heck, even a 15 year old hard drive that sat on a shelf for 15 years with a few gigs of data and never got connected is already worth a ton.
It's like the plague or the dark ages, but electronic. And since so many people rely on this technology, it will be devastating in many ways after the inevitable implosion.
If all of these scientists are as stupid as a bunch of school children using AI, why should we have ever trusted them in the first place?
If all of these allegedly brilliant scientists are so absurdly, dumb and untrustworthy, if they are so foolish as to be fooled by AI, then why did we ever trust them in the first place?
How stupid were the scientists? How stupid were we to think that they had any answers?
Perhaps we need better scientists that know better than to ever even ask AI anything
Monitor their Internet usage. If they ever use AI, just fire them. It doesn’t seem like a difficult problem to solve.
Sorry about your grade 10 chemistry mark.
Also, check out the 2024 Nobel Prize for Chemistry. Warning: some big words.
It’s surprisingly easy to make a cogent argument against AI (or any argument at all) without insulting others.
Try it sometime
Most people use LLMs poorly and irresponsible. Most problems they are used for, are a poor choice as well. That doesn't mean they cannot be useful.
Oh, I forgot that the entirety of science didn’t exist before AI, and can’t exist without AI. Since science is entirely dependent on using AI exclusively for 100% of all thought, we’re obviously all fucked.
¯\_(ツ)_/¯
You’re making the argument that the deadly deadly poison is only deadly if you drink too much of it, or drink it incorrectly. I elect not to drink the deadly, deadly poison — any of it at all.
Not really what I'm saying. More along the lines of: poison is bad to drink, but great for killing weed.
As for LLMs and sciences (and most fields): "drinking poison" -> "treating LLM output as factual". "killing weed" -> "some otherwise mundane language transformation that you verify the correctness of".
poison is bad to drink, but great for killing weed.
But that’s not what you said, nor is it the argument you’re making
The pursuit of science by humanity has made it this far for thousands of years. And has done so quite successfully without AI poisoning that pursuit. You have yet to make the argument that the poison of AI is in any way helpful.
Because you can’t. Because it isn’t.
Ignore the evidence at your own peril. I will not.
It kinda seems to me you're deciding both sides of the argument. It's like a straw man monologue show.
Arriving at a conclusion without evidence… Exactly the danger that AI presents.
"We should stop using chainsaws and cut down these trees with our teeth"
Yet, for thousands of years, cutting down these trees with our teeth worked surprisingly well.…
Proven by that tiny little computer in your hand that you’re typing your response upon
"The poison is in the dose".
Also, what an astonishingly uncritical perspective. I think there are plenty of legitimate concerns to raise about LLMs in science, but there's a reason that researchers are adopting these tools. I suspect it's because there's a ton of rote work in the field (literature review, analysis, drafting a document...) and they're under great pressure to publish on an accelerating cadence.
but there’s a reason that researchers are adopting these tools.
Yeah, it's how you get grant money.
Oh really, and tell us..how much of that grant money goes to the actual scientist who wrote the grant?
“I had to drink the poison because I was under pressure”
Somehow, this doesn’t ring true. Are you defending scientific discovery, or are you defending corporate pressure? Or are you just defending poison? either way, this sounds like bullshit.
it's because there's a ton of rote work in the field
So now it’s OK just because scientists are suddenly lazy? You gotta be kidding.