I would make two points here.

1. ICE is breaking the law in many, many different ways. It is not legal to murder multiple US citizens like Alex Pretti and Renee Good. The conditions of their internment camps (I will use the euphemism as I suspect you would not like the more accurate term) are not legal. And many, many other issues, please see:

https://www.aclu.org/news/immigrants-rights/we-documented-12...

https://www.americanimmigrationcouncil.org/fact-sheet/ice-cb...

https://www.brookings.edu/articles/the-administration-has-de...

https://www.reuters.com/legal/government/courts-have-ruled-4...

By your legalist moral standards, is this not also "black and white"? Do people not have a right or duty to protest it?

2. What is legal is not always what is morally right. The comments you are making mirror those in past "legal" civil rights violations at scale. The same arguments were made against the civil rights movement in the 60s, anti apartheid movement, and pretty much every civil rights movement. The answer in all of those cases is the same: what is legal does not determine what is morally right, and we have a civic duty to nonviolent protest when the two conflict. https://en.wikipedia.org/wiki/Lawrence_Kohlberg%27s_stages_o...

This is true uninstalling, is it not?

1. The claim of the authors is that the scientific literature is on average so dishonest and/or wrong that access to it (pre or post training, I assume?) actually harms LLM accuracy. Hopefully we can all agree this is a claim so extraordinary that would require some very uniquely powerful evidence. After all, we know that LLMs also train on very large corpuses of text with varying degrees of both accuracy and dishonesty - not the least, most of the internet! So the claim of the authors must be that the scientific literature is so bad that it is actually uniquely bad for LLMs, like worse than the general internet.

2. They provide one citation of evidence supporting their claim, which is this paper [0]. If you actually read this paper (which is mostly not about this actual question, but related questions about prefiltering), it provides no evidence for their claims. For example, in Figure 5, removing PubMed, aka the entire biological scientific literature, has by far the strongest negative impact on evaluation of Biomedical questions. It even has a strong negative impact on evaluation of questions in the "Common sense" category! To quote:

"Performance degrades when we remove domains with close alignment between the pre- training and downstream data sources: removing PubMed hurts the BioMed QA evaluations"

3. This post is maybe what you would get if you prompted an LLM: "please provide a citation for the claim that the scientific literature harms LLMs". It's really quite worrying that we're looking at AI generated propaganda designed to convince the reader that the scientific literature is worse than useless.

4. As a scientist, the primary use I want from LLMs outside of code generation is to be a fast and comprehensive search engine. I want to see the papers behind their claims and evaluate them. Literally the only time they are useful in scientific research is when they provide citations for the claims, so I can read the papers and evaluate.

5. Somehow we have gotten to the point where many people, especially people in tech, believe that the scientific literature is mostly junk or mostly useless. This contrasts so starkly with the current rapid pace of genuine, meaningful, society-impacting scientific progress in pretty much every major field. How we have gotten to the point where this myth is so pervasive, I do not know. Is it really spurred just by a few recurring news stories about reproducibility? By just interpersonal bitterness and feelings of anti-'elite' sentiment? (how on Earth your local state university climate scientist is a member of the 'elite' but software engineers making $500k a year or X influencers with millions of followers are not will always simply be beyond me).

[0]. https://aclanthology.org/2024.naacl-long.179/

For example, LLMs love to talk about LLMs (and the people who write with LLMs love to write about LLMs). Could "large language model" itself therefore be flagged as an AI-like phrase by this approach? It didn't exist much in the literature before 2022, does now, and certainly does more in AI-generated text: but, it is not actually a great way to distinguish modern AI generated text from human written text.

A helpful control would be to show that on some cohort of papers that can be declared reasonably clean of LLM generated text post 2023 there are very low rates compared to the arxiv.

For example, while papers in the journals Nature and Science are unlikely to be entirely LLM free at this point, if those were tested through 2026, we should see a line significantly lower than the arxiv's growth.

This has become the norm in science, and all of the best labs do it now, except for a few toxic holdouts who incorrectly believe preprinting their work will adversely impact its peer review.

MD doctors poll at extraordinarily high levels of trust, over almost any other professional group in the United States. So it really isn't correct to directly link this article's topic to "distrust". The effect you're talking about may exist in science, but this article is essentially a counter example to the effect you propose: clinicians publishing bullshit, but retain a high level of public trust.

Especially because the article is basically entirely quoting practicing scientists who identified this problem in the first place! More real scientific training or collaborating for clinicians who want to (or have to) do research could potentially improve the situation.

Tens of thousands of papers have made important scientific advances using it successfully and CRISPR-Cas methods are used routinely throughout almost all of biology.

This is like calling PCR "overhyped" because PCR-based infectious disease diagnostics are limited.

Science is actually progressing at an amazing rate in recent years. We are curing diseases and understanding more about life and the universe faster than ever.

Just briefly skim some top journals right now:

Here's an amazing 'universal vaccine' for respiratory viruses in mice https://www.science.org/doi/10.1126/science.aea1260

here are brand new genome editors in human cells https://www.science.org/doi/10.1126/science.adz1884

Here's amazing evidence of an ancient lake on Mars https://www.science.org/doi/10.1126/science.adu8264

Here's a meta-analysis of 62 (!) different studies on GLP1 receptor agonists to figure out whether they can contribute to pancreatitis https://onlinelibrary.wiley.com/doi/full/10.1002/edm2.70113

(covered here https://www.nature.com/articles/d41586-026-00552-6)

Here's identification of a new mechanism of resistance in Malaria https://www.nature.com/articles/s41586-026-10110-9

Here's curing a genetic disorder using gene editing in mice https://www.nature.com/articles/s41586-026-10113-6

Here's a study that has figured out that as CO2 levels rise, there's less nitrogen in forests https://www.nature.com/articles/s41586-025-10039-5

and here's personalized mRNA vaccines curing people of breast cancer https://www.nature.com/articles/s41586-025-10004-2

Like all of these are just from the past month or two and are pretty astounding advances. And they are just a subset of all of the scientific advances recently. All of them have contributors in academia (and science performed outside of academia would not exist without academia, as it depends upon it for most of the conceptual advances as well of course as for scientist training).

1. Stuff like paper mills and complete fraudsters exist, but for the most part, these things are the exception, not the rule. Your average scientist doesn't even hear or think about these things and the weirdos who cause them, to be honest. Nobody has ever heard of "International Review of Financial Analysis" outside of an extremely niche economics subfield.

2. "Public or perish" is not a cycle, really. While I believe it's not good for people to be constantly working under pressure, the fact that academia is so competitive currently is a healthy sign. It's because we have so many people with extremely impressive resumes and backgrounds, doing extremely impressive work, that makes funding so competitive. And when funding is competitive, it's no wonder that funders prefer to fund people who have produced something and told the world about it ("publish").

3. Fraudsters and hucksters have been in science forever. Go read an account of science in the early 19th century. There are tons and tons of stories of crazy scientists who believed ridiculous things, scientists who kept pushing wrong dogma, and so on. And yet nobody knows about them today, because the evolutionary process of science works: the truths that are empirically verifiable win out, and, given enough time, the failures are selected against.

Bit too on the nose, maybe, but a heck of a lot more likely than a coverup by government scientists.

Bluesky has recapitulated or even surpassed peak sci twitter. The signal:noise is excellent. However, it requires some work because there is no algorithm. Aggressively unfollow people with low signal:noise, use the custom feeds that disable reposts and enable replies, use the Quiet Posters feed, and use sill.social. This has created a science feed that for me surpasses even the peak of Twitter, let alone X today which is unusable for scientific discussion.

Finally, the thing that drives me crazy is that Bluesky is literally a popular open-source, nonprofit, Ad-less, algorithm-less, truly free and partially decentralized social media network. It's what we all dreamed about in the 2010s! It's Mastodon but actually popular! But half the tech community have convinced themselves it's a "liberal bubble" (that anyone can join....) and that the website that apparently isn't a bubble is the, err, website run by a billionaire with an algorithm designed to promote certain political content that agrees with that billionaire. Absolutely bizarre situation.