- cross-posted to:
- technology@lemmit.online
- news@kbin.social
- cross-posted to:
- technology@lemmit.online
- news@kbin.social
Study featuring AI-generated giant rat penis retracted entirely, journal apologizes::A peer-reviewed study featured nonsensical AI images including a giant rat penis in the latest example of how generative AI has seeped into academia.
It got published, people noticed it, people saw it was bullshit, it got retracted. Publishing is not the end of the line.
It’s an extreme example, but it’s still an example of the system working in the end. Reasonable people are supposed to question what they read, not blindly trust it, that’s how you catch “important shit”.
The problem is not that some bad papers get published. The problem would be them staying unchallenged. And it’s also a problem that laymen consider one random study is an undeniable proof of their argument (potentially ignoring the thousands of studies contradicting it).
Of course some things will always slip through the cracks, but this is egregious. What does their peer-review process look like that this passed through it?
Right? Even when skimming papers, it’s usually: read title & abstract, look at figures, skim results & conclusion. If you don’t notice that the figure doesn’t have real words, how is anyone making sure the methodology makes sense? That the results show what the conclusion says they show?
I am not disagreeing that this is ridiculous, I was just saying that this stupidity is not what should convince people not to take some random paper for an absolute truth, just because it was published.
Even if you eliminate fraud, bullshit and even honest mistakes, that’s just not how science works.
A shame most people are trained by both the school system and society to just take things at face value
An even greater shame is that almost no people are trained on basic statistics and think they can debunk a published study in PNAS with a Google search and some random guys blog.