Honestly, my guess is the article was peer reviewed by similar AI-brained researchers who just fed it into an LLM and skimmed the summary. Which is disturbing in its own right.
But the real irony comes from the text of the article. The whole crux of it is 'use AI models to diagnose autism, and make them 'explainable AI' to overcome the trust gap.' Except the authors and entire field clearly have huge blind spots and they're basically proposing digital phrenology.
In addition to this, their suggestions include 'robot assisted therapy' and 'use ai to generate recommendations for parental support'. But the science is apparently... dumping some data into a weighted neural network to return a score? And maybe it was higher confidence and faster than other models, but I gotta ask, did they just ask an LLM to try things until the numbers were better? What is 'explainable' about that process, and is it better or did you just manipulate the dataset until number went down?
Finally... the whole intro basically treats autistic people as 'a growing problem no one understands' and seems to accept the framework of 'autistic people are a burden who need constant management and no one really understands it' instead of 'autistic people need support.' The whole framing seems to remove agency from autistic people; granted that may be because it's written for early childhood intervention, but it feels... gross.
You're absolutely right Matt. Other commentary on this piece online has pointed out how there's a whole host of inquiry (but maybe still shonky) around AI for autism and it's all just a bit talking down and at autistic people, rather than to them. As if they're a problem to solve with ever more powerful LLMs. I'm with you.
28 November 2025Editor’s Note: Readers are alerted that the contents of this paper are subject to criticisms that are being considered by editors. A further editorial response will follow the resolution of these issues.
It wasn't just the figure which is clearly AI generated. The paper was very badly written and surprised it got past review in a 'prestigious' journal no less. There was bad citations, no description of the framework, datasets that came out of nowhere and then a section on lower back pain which was completely irrelevant. Let's say the author only used AI for the figure (although many are doubting this), how could a paper so full of bad science have been published. In theory, no one could have replicated their results but it went through peer review so we are assumed its okay to trust the results, this is very bad work from the journal, editor and reviewers.
Some journals are doing this but we need to be transparent with the reviewer comments and also share the raw data where we can. These days too many algorithms are being trained and published and the R2 value is the only thing that's presented as a judge for the model. I am not saying we shouldn't trust anything but papers like this make me question the next time I see a good R2 and low MSE value.
Yes, I had a bit of a crisis about the headline but I think it gets at the silliness of this study, obviously an autism bicycle is a completely ridiculous concept!!
They should definitely drop an official comment there, it's already at over a dozen comments. And preferably one with a bit more detail about what went so wrong
Like I said, publishers have to subscribe to the journal dashboard, or they are not allowed to comment. Most publishers and journals do not subscribe, but a few do. The revenues from these subscriptions is what pays for PubPeer to operate.
No, but the figure comes at the end of the paper after the references when it is submitted. I could absolutely see someone just not scrolling through the PDF until the end.
Honestly, my guess is the article was peer reviewed by similar AI-brained researchers who just fed it into an LLM and skimmed the summary. Which is disturbing in its own right.
But the real irony comes from the text of the article. The whole crux of it is 'use AI models to diagnose autism, and make them 'explainable AI' to overcome the trust gap.' Except the authors and entire field clearly have huge blind spots and they're basically proposing digital phrenology.
In addition to this, their suggestions include 'robot assisted therapy' and 'use ai to generate recommendations for parental support'. But the science is apparently... dumping some data into a weighted neural network to return a score? And maybe it was higher confidence and faster than other models, but I gotta ask, did they just ask an LLM to try things until the numbers were better? What is 'explainable' about that process, and is it better or did you just manipulate the dataset until number went down?
Finally... the whole intro basically treats autistic people as 'a growing problem no one understands' and seems to accept the framework of 'autistic people are a burden who need constant management and no one really understands it' instead of 'autistic people need support.' The whole framing seems to remove agency from autistic people; granted that may be because it's written for early childhood intervention, but it feels... gross.
You're absolutely right Matt. Other commentary on this piece online has pointed out how there's a whole host of inquiry (but maybe still shonky) around AI for autism and it's all just a bit talking down and at autistic people, rather than to them. As if they're a problem to solve with ever more powerful LLMs. I'm with you.
Currently doesn't say retracted, rather:
Change history
28 November 2025Editor’s Note: Readers are alerted that the contents of this paper are subject to criticisms that are being considered by editors. A further editorial response will follow the resolution of these issues.
Yeah, there's some expectation it will be retracted within a week or two, but so far I only have the Editor-in-Chief's word that it will be retracted
It wasn't just the figure which is clearly AI generated. The paper was very badly written and surprised it got past review in a 'prestigious' journal no less. There was bad citations, no description of the framework, datasets that came out of nowhere and then a section on lower back pain which was completely irrelevant. Let's say the author only used AI for the figure (although many are doubting this), how could a paper so full of bad science have been published. In theory, no one could have replicated their results but it went through peer review so we are assumed its okay to trust the results, this is very bad work from the journal, editor and reviewers.
Some journals are doing this but we need to be transparent with the reviewer comments and also share the raw data where we can. These days too many algorithms are being trained and published and the R2 value is the only thing that's presented as a judge for the model. I am not saying we shouldn't trust anything but papers like this make me question the next time I see a good R2 and low MSE value.
Yes, I think that this "peer review" process leaves a lot to be desired. I have asked further questions of Springer Nature here.
fexcectorn or covfefe: the impossible choice. This will be a book about 2000-2025.
Oh god I don't want to be in a book because of a silly newsletter
I note the data are available by "reasonable request" ... please request them!
I contacted the author twice but no response! I hope Springer Nature can determine if they are even real. I suspect they are!
I saw this headline and as an autistic person I was ready to be all >:( but wow jeez that’s crazy
Yes, I had a bit of a crisis about the headline but I think it gets at the silliness of this study, obviously an autism bicycle is a completely ridiculous concept!!
Publishers can (officially) only comment on PubPeer if they subscribe to the journal dashboard.
They should definitely drop an official comment there, it's already at over a dozen comments. And preferably one with a bit more detail about what went so wrong
Like I said, publishers have to subscribe to the journal dashboard, or they are not allowed to comment. Most publishers and journals do not subscribe, but a few do. The revenues from these subscriptions is what pays for PubPeer to operate.
Peer review has a lot of problems not least it’s unpaid. You get what you pay for. It’s no surprise a lot of mistakes get through.
Agreed! I don't think any peer reviewer agreeing to handle a story like this though could have looked at Figure 1 and been like "that's fine"...
No, but the figure comes at the end of the paper after the references when it is submitted. I could absolutely see someone just not scrolling through the PDF until the end.