Brandolini's law, aka the "bullshit asymmetry principle" : the amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.
Unfortunately, with the advent of large language models like ChatGPT, the quantity of bullshit being produced is accelerating and is already outpacing the ability to refute it.
I'm curious to see if AI tech can actually help fight some of the bullshit out there someday. I agree that current AI is only making it easier to produce bullshit, but I think with some advances it could be used to parse a long-winded batch of bullshit, and summarize it, maybe with bullet points about how the source material is wrong. If they can make an AI as confident as chatgpt, but without as much of the "makes stuff up left and right" it could be useful.
THEN we just have to worry about who owns the AI that parses and summarizes the info we take in, and what kind of biases they've baked into the tech...
I have high hopes for concepts like Toolformer where the model has to learn to use external APIs and resources like Wikipedia or Wolfram to get answers, rather than relying on the inscrutable and garbled soup of knowledge absorbed from the text training corpus directly. Systems plugged into knowledge graphs could have the best of both worlds - able to generate well-written novel text outputs AND the added rigor of "classical AI" style interpretability.
I’m curious to see if AI tech can actually help fight some of the bullshit out there
Those AI are the best ones to produce fake scientific papers. It's a cat and mouse game again. Those who can detect bullshit can produce the best bullshit.
As someone who just submitted an article for review I am gobsmacked by how brazenly the authors have done this. The absolute disregard for integrity and the knowledge production process is astounding. But also the balls to just submit a paper like this without fear of the consequences says something more profound about the state of academia.
I recently watched an interview with an academic woman who trolled the academia with completely wrong things in his papers, and yet it was approved. Academics, and scientific data is not so objective as it may seem from the outside.
Stupid question: Why can't journals just mandate an actual URL link to a study on the last page, or the exact issue something was printed in? Surely both of those would be easily confirmable, and both would be easy for a scientist using "real" sources to source (since they must have access to it themselves already).
Like, it feels silly to me that high school teachers require this sort of thing, yet scientific journals do not?
Because scientific journals exist to profit off science, not bolster it. Fact checking costs money so they do the bare minimum they deem necessary to preserve their reputation.
Many of the journals I've published in do require a link, usually a PMID or DOI, but they're not usually part of the review process. That is, one doesn't expect academic content reviewers to validate each of the citations, but it's not unreasonable to imagine a journal having an automated validator. The review process really isn't structured to detect fraud. It looks like the article in question was in the preprint stage - i.e.: not even reviewed yet - and I didn't notice mention of where they were submitted.
Message here should be that the process works and the fake article never got published. Very different than the periodic stories about someone who submits a blatantly fake, but hand written, article to a bullshit journal and gets published.
Well that used to be a thing called a bibliography but it appears that these journals don't require such. Funny when even my old 7gr essays required those
As Retraction Watch reports, Natural History Museum of Denmark myriapodologist Henrik Enghoff suspected the authors of the paper from China and Africa used OpenAI's ChatGPT to dig up academic references — and as it turns out, his hunch was right.
The offending paper was initially taken down by Preprints.org, a preprint archive run by the academic publisher MDPI, in June after Enghoff's colleague, the University of Copenhagen's David Richard Nash, notified editors of the errors.
Earlier this year, reporters at The Guardian noticed that the AI chatbot even made up entire articles with bylines of journalists who had never written these non-existent pieces.
"We will withdraw it immediately and add the authors of this preprint to our blacklist," Preprints.org's editor Lloyd Shu told Nash in an email back in June.
Kahsay Tadesse Mawcha of Aksum University in Ethiopia, who was originally listed as a corresponding author on the offending preprint, admitted to Danish newspaper Weekendavisen back in July that he indeed used ChatGPT, adding that he only realized later that the tool was "not recommended" for the task.
Powerful but flawed AI tools like ChatGPT are a bull in a china shop of almost every knowledge domain, academia included — and it'll be fascinating to watch everybody involved try to find a new sense of equilibrium.
The original article contains 547 words, the summary contains 214 words. Saved 61%. I'm a bot and I'm open source!
Your assumption is wrong. This was not carelessness. Academic dishonesty and lack of integrity is an ongoing issue in research. China is one of the biggest culprits for blatant plagiarism and IP theft, although recently even academics from Ivy league universities have been implicated in fraudulent publications. The simple fact is that number of publications is the main metric used in academia for hiring and promotion. This leads to a perverse incentive model where academics prioritise publishing over conducting good science, thus all we get is a shit load of noise (poor articles) that obscure the signal (good articles).
China is one of the biggest culprits for blatant plagiarism and IP theft, although recently even academics from Ivy league universities have been implicated in fraudulent publications.
Sure, let's make this about China when 4 out of 5 of the authors credited for the original article are from Africa.
While only one of which was from China. This doesn't even address the fact that the republished paper came from Mawcha which describes a study on millipedes in... Africa. Guess what, Wenxiang Yang wasn't even credited in this version. Was your reply carelessness or dishonesty and lack of integrity? I don't care where the misinformation and carelessness comes from as long as we're making efforts to stop it, but this is highly ironic.
A lot of people don't understand the limitations/weaknesses of AI. The carelessness was probably more in not actually learning about the tool he was relying on (and just assuming it was reliable information).