Originally published at: Academic fraud endemic in published research, from photoshopped blots to AI slop - Boing Boing
…
11 Likes
Ooh, now do law journals! Seriously, can you imagine if all the medical journals in the US were managed and edited, and often written, by medical students? But that’s how law journals work!
19 Likes
But what journal would pay a lawyer $250/hr to write something?
10 Likes
As a (mostly) former pharma researcher i’ll tediously add that if they “backup a dump truck of money” if you can but show some statistical efficacy of some new pharmaceutical then there will always be someone who will manage to subset the factor analysis (roughly leave out certain of the control sets) until a positive effect appears. (says one who is stupidly poor but still believes everything he got published -sigh-)
19 Likes
It’s amazing how broken science is right now. Science really requires confirmational studies, which would eliminate the problems caused by fraudulent research, but it’s not happening for some of the same reasons the fraudulent research is. Doing anything other than groundbreaking research with positive results isn’t sexy, but more than that, it’s bad for one’s career. You aren’t going to get research grants or tenure by doing experiments to affirm someone else’s work. I don’t even know how that gets fixed.
19 Likes
Fraudulent medical research carries with the the strong possibility that it will get people killed. It will tend to promote less or non effective therapies at the expense of better ones.
8 Likes
Well, I’m glad that Capitalism doesn’t lead to fraudulent science.
Edit: unexpected truth in advertizing; though I don’t think that a coffin is normally called a “flip-top box”.
Edit-edit: I’m intrigued that the models in these vintage adverts are all displaying hand tattoos. I guess it was a quick way to fill in some backstory for the characters.
https://tobacco.stanford.edu/cigarettes/filter-safety-myths/marlboro-men/
15 Likes
Meanwhile on the Boingboing shop . . . How to beat AI detectors.
11 Likes
Even “wasting research money to try to replicate falsified results” isn’t a victimless crime because every dollar spent on garbage research is a dollar that isn’t being spent on research that could help save lives.
That’s the real problem with scammers like Elizabeth Holmes. Not that she ripped off a bunch of mostly-rich investors, but that in doing so she directed resources away from scientific endeavors that actually might have worked.
19 Likes
I think it’s important to make a distinction between the kind of fraud that Masliah and Tessier-Lavigne were involved with and the world of for-profit journals. The fraud that occurs in high-profile biomedical research is driven by a need for funding on the part of the leadership and a need for publications on the part of the workers. If you’re running a lab and a postdoc brings you a good result, there’s no incentive to really scrutinize it and you need good results to keep your lab funded. If you’re a postdoc, your entire career depends on generating good results, so the incentives for dishonesty are extremely high.
These prestigious biomedical results are not going to predatory for-profit journals though. Those journals serve largely researchers in countries outside of Europe and North America where academic success is judged simply by counting the number of publications, regardless of the quality of those publications. The damage to science overall is minimal because nobody reads that crap.
5 Likes
On the subject of AI generated slop, this recent article was pretty good at explaining just how bad things are getting, and included some discussion on scientific papers:
One thing that makes papers especially bad is that people use them as citations in their own papers, and even if a paper is later found to be fraud or AI-generated BS you can very quickly have a whole web of citations propagated across academia that is impossible to correct even if the original paper is retracted.
11 Likes
Huh, you inspired some quick searching and I found this -
11 Likes
There is nothing lower than publishing fake data. It should be punishable by immediate defenestration
Academic publishing requires competent editors and peer-reviewers that know the literature and check each manuscript in detail. With the number of journals and articles currently being published, though, that is not remotely possible.
The emphasis on publishing lots of articles instead of quality articles is pervasive and really harmful. That Masliah has 800 published articles at all should be more of a red flag than a positive. 800 articles over a 40 year career is a paper every 18 days. That suggests either putting one’s name on lots of other peoples work without even reading it or cutting a lot of corners
23 Likes
And that was already true even before reviewers and editors had to start wading through a torrent of AI-generated slop. It’s really hard to see how academia can fix this issue.
14 Likes
Every new application of AI seems to make a stronger case for the Butlerian Jihad.
13 Likes
Two things.
- right below this article I found a BoingBoing Shop advertising for “beating AI detectors”, in which sounds like a huge hypocrisy. BoingBoing also fell to capitalism at whatever price, seems to be.
- back in 2019 I worked in an Argentinian IT consultancy company. The motto was “first sell the consultancy service, then we will see how we can do it, if we cannot do it, we fake it, and when the customer realizes it is a scam, the legal troubles will be cheaper than the fees already collected”.
6 Likes
Thank you for find that article; I love an odd bit of social history.
@Scientist My personal research has shown that Defenestration is often unsuccessful. It is surprisingly hard to get a body over the window ledge. Maybe I just lack upper body strength, or confederates.
8 Likes
Windows used to be larger and have sills below most people’s hip height. That made it much easier to tip them out. But check Russian YouTube for helpful instructional videos-they should have plenty!
7 Likes
If only they had shopped on the BoingBoing shop for useful products to help them fool AI detectors! Want to fool AI detectors? Here's how (with another AI tool)! - Boing Boing
4 Likes
It’s also a good idea to check below the window that you’re using, to make sure that there’s not an inconvenient pile of manure to soften the fall.
5 Likes
