from Hacker News

MIT asks arXiv to withdraw preprint of paper on AI and scientific discovery

by carabiner on 5/16/25, 3:09 PM with 202 comments

• by Aurornis on 5/16/25, 7:25 PM

  > Our understanding is that only authors of papers appearing on arXiv can submit withdrawal requests. We have directed the author to submit such a request, but to date, the author has not done so.

  Between this and the subtle reference to “former second-year PhD student” it makes sense that they’d have to make a public statement.

  They do a good job of toeing the required line of privacy while also giving enough information to see what’s going on.

  I wonder if the author thought they could leave the paper up and ride it into a new position while telling a story about voluntarily choosing to leave MIT. They probably didn’t expect MIT to make a public statement about the paper and turn it into a far bigger news story than it would have been if the author quietly retracted it.

• by intoamplitudes on 5/16/25, 4:46 PM

  First impressions:

  1. The data in most of the plots (see the appendix) look fake. Real life data does not look that clean.

  2. In May of 2022, 6 months before chatGPT put genAI in the spotlight, how does a second-year PhD student manage to convince a large materials lab firm to conduct an experiment with over 1,000 of its employees? What was the model used? It only says GANs+diffusion. Most of the technical details are just high-level general explanations of what these concepts are, nothing specific.

  "Following a short pilot program, the lab began a large-scale rollout of the model in May of 2022." Anyone who has worked at a large company knows -- this just does not happen.

• by FilosofumRex on 5/17/25, 2:10 AM

  MIT is hiding its own culpability by throwing the Student under the proverbial bus. Acemoglu and Autor who are notorious attention seekers and very media savvy and wealthy profs had vouched for him. There is no way a 2nd year PhD students could have pulled this off on his own without a trace of his whereabouts and contacts in the industry.

  A cursory review of the first paragraph of the abstract of his single author paper should've set off alarms:

  "AI-assisted researchers discover 44% more materials, resulting in a 39% increase in patent filings and a 17% rise in downstream product innovation".

  Anyone with rudimentary familiarity with industrial materials science research would have suspected those double digit numbers - even single digit improvements are extremely rare.

• by rdtsc on 5/16/25, 10:17 PM

  > The paper was championed by MIT economists Daron Acemoglu, who won the 2024 economics Nobel, and David Autor. The two said they were approached in January by a computer scientist with experience in materials science who questioned how the technology worked, and how a lab that he wasn’t aware of had experienced gains in innovation. Unable to resolve those concerns, they brought it to the attention of MIT, which began conducting a review.

  So the PhD student might have been kicked out. But what about the people who "championed it". If they worked with the student, surely they might have figured out the mythical lab full of 1000s material scientists might not exist, it might exist but they never actually used any AI tool.

• by tokai on 5/16/25, 7:29 PM

  Impressively the paper seems to have been cited 50 times already. I don't mind much if its taken down or not, but with the old guard publishers you can at least get a redaction notice or comment about the issues with a paper embedded in the publication. If you find this paper cited somewhere and follow it to the source at arxiv, you will never be made aware of the disputes surrounding the research. Preprint servers has somewhat of a weakness here.

• by pvg on 5/16/25, 3:23 PM

  The paper had an HN thread a few months ago

  https://news.ycombinator.com/item?id=42115310

• by elcritch on 5/16/25, 10:05 PM

  In my opinion the paper shouldn’t be take down. Instead a note should be added noting the concerns with the pre-print and that’s it’s likely fraudulent.

  Edit: Since the paper has been cited, others may still need to reference the paper to determine if it materially affects a paper citing it. If the paper is removed it’s just a void.

• by ipsum2 on 5/16/25, 6:16 PM

  MIT's article is quite scant on details. WSJ has more information, but still no specifics: https://www.wsj.com/tech/ai/mit-says-it-no-longer-stands-beh...

  > The paper was championed by MIT economists Daron Acemoglu, who won the 2024 economics Nobel, and David Autor. The two said they were approached in January by a computer scientist with experience in materials science who questioned how the technology worked, and how a lab that he wasn’t aware of had experienced gains in innovation. Unable to resolve those concerns, they brought it to the attention of MIT, which began conducting a review.

• by ayhanfuat on 5/16/25, 6:30 PM

  > by a former second-year PhD student

  Seems pretty serious if they kicked him out.

• by coderintherye on 5/16/25, 7:32 PM

  Nice Twitter thread from Nov '24 analyzing the paper: https://x.com/Robert_Palgrave/status/1856273405965693430

• by 1024core on 5/16/25, 11:59 PM

  Link to paper: https://arxiv.org/abs/2412.17866

• by dougb5 on 5/16/25, 8:26 PM

  This makes me think about the credibility of single-author vs. multi-author papers in different disciplines. In computer science, a paper is seen as suspicious if there's just one author (at least nowadays). But in economics it seems much more common. Can an economist explain this for me (or perhaps a paper written by multiple economists?)

• by yolkedgeek on 5/17/25, 12:33 PM

  So MIT is trying to tell us that there was no incentives, no VC money, no corruption involved here and just a naughty naughty PhD student?

  This is how I see this: They published a false paper, knowing they would eventually be caught. But the boost that the paper had in AI marketcap and market bullishness and hype, far outweighs the consequences of this apology(not really) statement.

  So they got a lot of money for hyping the AI, and now they pay off a small amount of it as backlash. still a huge net income. Pharma companies do this all the time. this is how they make money.

  Reminder that if you link articles for your argument, doesn't necessarily mean that you are right. There are substantial amounts of false and slightly distorted papers and articles out there, even from the most valid publishers.

• by 12_throw_away on 5/16/25, 7:17 PM

  Reading the paper (which is still up) ... the "AI" (sigh) tool described there would not have been particularly novel or unusual, even if the research was conducted several years ago. ML + inverse design for materials has been used for decades.

• by jrochkind1 on 5/16/25, 10:38 PM

  you'd think that such a widely cited fraudulent paper might have caused problems in other research, but probably nobody who cited it actually read it anyway, so. it's turtles all the way down.

• by hooloovoo_zoo on 5/16/25, 7:58 PM

  I don't think arXiv should take it down even if it is fraud. ArXiv is more about being a permanent store than a quality judge.

• by tomrod on 5/17/25, 9:40 AM

  I wonder what the issue was.

• by eterm on 5/17/25, 10:09 AM

  Interesting, the paper got immediately flagged as suspicious by HN commenter lysecret:

  https://news.ycombinator.com/item?id=42128532

• by Syzygies on 5/17/25, 3:19 AM

  If I've learned one thing using AI, it's that other people's experiences are not particularly relevant. So this paper is a meh! for me whatever its provenance.

  Compare the failure rates for condoms.

• by gowld on 5/16/25, 5:18 PM

  Bad OP Title

  Better title:

  MIT disavows heavily-discussed economics preprint paper about Artificial Intelligence and Scientific Discovery.

• by shanemhansen on 5/16/25, 8:38 PM

  "I don't endorse this paper. Therefore you should take it down. I won't tell you why. Trust me bro."

  Whether MIT is right or wrong, the arrogance displayed is staggering. The only thing more shocking is that obviously this behavior works for them and they are used to people obeying them without question because they are MIT.

• by nikolayasdf123 on 5/16/25, 11:34 PM

  > paper should be withdrawn from public discourse.

  just post a correction notice on arXiv. let others decide if there is merit to it or not.

  silencing is so anti-science. shame on MIT.