@andrewtanyongyi@lpachter Seriously? The authors of that supposedly serious preprint were unable to work out they had an hallucinated reference in spite of knowing that this would be disqualifying? Please.
@lpachter I guess if it's trivial then the arxiv's policy will fail to stop the tide (I don't know cause I don't use AI to write papers). I'm just appalled at the 'it's all already fucked so who cares it it gets worse.. let it burn down!' attitude that I sense in many posts.
@JimDMiller If they are so easy to detect, then just take them out before submitting the paper. Are you seriously suggesting the authors cannot e expected to spend a ‘tiny amount of time’ to remove them but the readers should? Fuck that I say.
Speak for yourself.
I’m so sick of this cynicism.
EVEN if it were true that people just copy lists and that it’s all politics (which not isn’t). Those lists wouldn’t include hallucinations. That shit is entirely due to AIs.
I'll tell you why so many people upset about the "no hallucinated citations" ban on the arxiv: because they've all been copying citation lists from each other without checking them since the beginning of time.
And why did they do this? Because half of the citations in scientific papers are politics and not to the benefit of the reader. If you don't list the right papers, your paper doesn't look 'right' and reviewers will complain that you didn't cite this-and-that other unrelated work.
For what I am concerned, these are all bullshit citations that shouldn't be in the papers in the first place. They can easily be automated by "related papers" links, that are (wait for it) provided by... AI...
@arjunrajlab Of course you can use AI as a tool when writing your paper. But if you write a paper and put your name on it claiming you are the author, you are responsible for the content.
btw I'm not so convinced AI will be good at telling relevant from irrelevant refs.
@DellAnnaLuca Requests for irrelevant references to themselves by douchebag reviewers != hallucinating references to papers that don't exist.
The latter ONLY AI does.
Is that really so hard to understand?
I’m so terrified by this cynical race to the bottom. Yes, science is done by fallible people that can do all kinds of bullshit. But that’s not a fucking argument for just letting go of any and all standards.
At my doctoral defense one of the members of my committee was outraged that I quoted a scholar they had a personal beef with. It was a single citation for a minor establishment of fact not in question. I quoted nothing else by this author. Still, they refused to approve my dissertation until I agreed to remove it. I replaced it with a citation of the committee member instead. That was the only requested change by anyone on the committee for the entire dissertation.
@anshulkundaje Arxiv is for human authored papers and I fully support that.
Claiming to have authored a paper when you did not is fraud and should not be tolerated.
I can't believe there is even a discussion about this.
@anshulkundaje The way I understand it they are saying: We are being overwhelmed with horrible AI crap that is full of embarrassing stuff that makes clear the authors never read their own paper. So we pick one such thing: non-existing citations.. to filter that crap out. Seems reasonable to me.
on the whole @arxiv flap about hallucinated references etc
you don't see the stuff we reject...
some of it is really really egregious
the decision to impose additional consequences is largely to throttle that stuff so n00bs and bad actors don't trash us trying repeatedly
It’s astounding to me that a majority of people press the blue button. It’s like: either you swallow a placebo or deadly poison. And only if more than 50% of people swallow the poison, an antidote will be made available. Wtf? Just don’t swallow poison!
Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?
This is such corrosive shit to say.
And it's simply not true. Yes, the higher up the ladder one goes, the more politics gets involved. But faculty hiring and tenure decisions, in my experience, are still largely based on a committee of peers trying to decide on academic merit.
The fate of academic careers is decided in the dark, and the process is political, not based on merit or reason (“so vicious precisely because the stakes are so small”).
AI is a black box? Don’t make me laugh. I’m a million times more afraid of humans than I am of AI (just look around you), and in academia that’s doubly true.
Officially out - very excited to share! “Emergent simplicity” in microbial ecosystems has long been an appealing idea—but meant different things to different people. As a result, the field hasn't agreed: is it real? surprising? useful? 1/3 https://t.co/nr291bDi5S
@arjunrajlab@anshulkundaje you are saying the increase from 100 to 400 day median is all due to the time looking for reviewers and the extension in the length of time reviewers take?
I call bullshit. I bet it's mainly in the vastly longer time needed to do all the demanded revisions.
Remember KC complexity is only defined up to an additive constant that depends on the choice of the Turing machine. For LLMs the size of this machine is huge in comparison to the size of both inputs and outputs.
No. LLMs have essentially memorized the internet in their weights. For LLMs the KC complexity is not in the prompts but in the size of the ‘Turing machine’ generating the output. If anything is surprising is that the internet can be compressed into an LLM.
I think one of the conclusions we should draw from the tremendous success of LLMs is how much of human knowledge and society exists at very low levels of Kolmogorov complexity.
We are entering an era where the minimal representation of a human cultural artifact... (1/12)