Coffee Break: More on American Science, Thomas Jefferson and AI, and Natural History for the Ages

Part the First: Beware of Resting on the Shoulders of Atlas. Dr. Scott Atlas is a radiologist who is now ensconced in the Hoover Institution on War, Revolution, and Peace at Stanford (naturally, and in the tower, no less). One can only surmise from the general run of their work product that (true) Revolution and Peace as understood by the masses are not their main interests. Dr. Atlas’s latest proposal is to do away with the National Institutes of Health. This story is told with wit by Dr. David Gorski at SBM in “Abolish the NIH”? Dr. Scott Atlas gives the antiscience game away. Before we get started, let us stipulate that NIH is a large organization in need of improvement. No reasonable person disputes this. Nevertheless, it is still true that the development of virtually every drug and medical device in current use has depended on research support from NIH, the National Science Foundation, and similar agencies the world over. Still, Dr. Atlas believes that NIH should be abolished because “A $48 billion government monopoly should give way to the market.” Dr. Atlas goes to the source: In 1980, economist Milton Friedman said the National Institutes of Health should be abolished. Friedman said the same about another government research agency, the National Science Foundation. And when he was asked what the NSF should be replaced with, he replied: “Nothing.” And he then follows this up (from his Washington Post article) with: Abolishing the NIH would not create a funding vacuum. The private sector funds 78 percent of U.S. biomedical research and development. Venture capital has exploded. Investment focused on artificial intelligence in drug discovery alone surged nearly 36-fold between 2010 and 2024 — funding the overwhelming majority of innovative medical devices and building the AI revolution in biomedicine. The Howard Hughes Medical Institute, the Wellcome Trust, the Gates Foundation (no more Linda) and the Chan Zuckerberg Initiative represent more than $150 billion in endowed capital. Nokia Bell Labs, funded by AT&T and then Nokia private revenue, generated 10 Nobel Prizes. As Dr. Gorski notes, a precocious fifth grader could unpack this. If NIH is a monopoly then how could the “private sector” fund 78% of U.S. biomedical research and development. Atlas then confuses the endowments of the Howard Hughes Medical Institute, the Wellcome Trust (Great Britain), the Gates Foundation, and the Chan Zuckerberg Initiative with their funding, which is by definition a small fraction of their endowments. And yes, Bell Labs was funded originally by AT&T and the Nobel Prizes did come. But the AT&T that begat Bell Labs was a regulated monopoly, for which the support given to Bell Labs was not much more than a rounding error in the grand scheme of things at AT&T. After the breakup of ATT, research that began with Bell Labs support continued to win Nobels. But the most significant thing produced by the latter-day Bell Labs was Jan Hendrik Schön, whose scientific misconduct may be unequalled since modern science began with Aristotle. The notion that Big Pharma will build the foundation for continuing advances in biomedical science by funding “blue sky” research is absurd. And without this foundation, there will be nothing to build upon. Two examples suffice. The drug imatinib was probably the first “purpose-built” chemotherapeutic agent that turned out to be revolutionary. As a protein kinase inhibitor developed by Nicholas Lydon at Ciba-Geigy (merged with Sandoz to become Novartis) imatinib cured cancer, for a while until the tumors developed resistance. However, the foundation for imatinib dates to the 1950s with the research of Edwin G. Krebs and Edmond H. Fischer, virtually all of it funded by NIH, on the roles of reversible protein phosphorylation in the regulation of cellular metabolism, signal transduction, and the cell division cycle. Without Krebs and Fischer, who were paragons of biomedical science and mentorship, there would have been no rational development of imatinib or a thousand other drugs. The other example is stranger. Who would have known that Gila monsters could be so useful? GLP-1 antagonists for the pharmacological treatment of obesity (final verdict of this hysteria still to come) came directly out of research on components of Gila monster venom. The first paper describing what was then called extendin was published by Raufman, Singh, and Eng, who worked at the SUNY Health Science Center in Brooklyn and the Veteran Affairs Medical Center in the Bronx. Big Pharma and Big Medicine were absolutely nowhere in sight, and they most certainly would have had no use for Gila monsters. That Dr. Atlas thinks for a minute that this basic research can be bypassed but solutions will be found anyway is, well, nuts (it is frankly difficult to believe he really means it). At least a thousand of his Stanford colleagues could explain this to him in very simple terms. He should not be paid a moment’s attention. Oh, and while you do not have to die to write an obituary, you do have to engage with NIH to know how it works if you are to have an independent biomedical research program in the United States. Dr. Atlas appears in PubMed more than a hundred times, mostly it seems in archival papers on radiological imaging. He has a single two-year non-renewable NIH grant to his name, from twenty-five years ago. This is not enough background and relevant experience to be anything but a gadfly and a merchant of doubt, whatever one’s institutional affiliation. Part the Second: The Science of Natural History as It Should Be Done. As a kid I collected bugs and learned their scientific names, along with those of the plants and other animals in their habitat. A friend and I also sifted fossil sharks’ teeth and shells from the sandy spoil of artesian relief wells that prevented salt water intrusion in our part of the Floridan Aquifer (alas, we cast those pearls, including a few large Carcharodon teeth). We had fun and grew up believing that the height of biological research was to name a new species. We were never going to find a new species in North America. I learned later that this is only one of the peaks of the profession, but it is still relevant as shown in this article about The Wilderness Project: A scientific expedition to Angola’s remote Lisima plateau has uncovered dozens of species unknown to science, including eight undescribed dragonfly species, three new grasshopper species, and approximately 60 moths and butterflies new to science. The findings, from the Cassai Life Atlas — a biodiversity survey conducted by The Wilderness Project in February 2026 — offer the most detailed picture yet of a landscape that feeds the headwaters of four of Africa’s greatest river systems: the Congo, the Okavango, the Zambezi, and the Cuanza. The photographs of the various new-to-science species are breathtaking. And the video is splendid. If this work can be done in Angola, then the future of natural history is bright. As one of the scientists puts it, “We are born naturalists.” Yes, we are. And “protecting our biodiversity is protecting us.” Something to remember as we continue to view the world that we think is apart from us as a sink for our folly. This will not end well. Part the Third: There Goes Another One. Since January of last year, inducements to leave the United States to pursue scientific research somewhere else have gotten better. Thus comes the move of Omar Yaghi from UC-Berkeley to Tsinghua University in Beijing. The Nobel Prize winners at Berkeley will have one less competitor for the parking places outside of LeConte Hall (chemistry, the last time I visited) that have signs saying, “This parking place reserved for Nobel Prize winners.” Those of us from other institutions could only laugh at that: The move, first reported by the South China Morning Post, comes as the administration of President Donald Trump continues its attempts to slash US science spending and limits international research partnerships. Some nations, including China, have responded by trying to lure US talent with the promise of money and support. Earlier this year, for instance, France announced that it would award funds to dozens of US scientists relocating there. China has been wooing international researchers with talent-recruitment programmes, and some of its cities and provinces are even offering researchers lump sums and monthly allowances to relocate within their borders. Yaghi already had a connection to Tsinghua University — he became an honorary professor there in 2022. But he was officially welcomed as a full-time faculty member at a 3 July ceremony. Yaghi was unavailable to speak to Nature for this story. However, in a recent interview with Scientific American he said that the current state of US science is “not so encouraging because of the cutting back on grants” and because of a drop in the support from US science agencies that academic researchers rely on. He also worried that US researchers were not embracing what he sees as an “AI revolution”. Researchers need to engage with AI models, he said, “as a matter of survival of the advanced research system in the US”. If AI is to have positive utility, it will be in the basic research previously performed by Omar Yaghi. It is not too difficult to imagine minions of the current administration saying, “So what?” It might also be pointed out, with irony dripping from the words, that Professor Yaghi is a Palestinian who was born in Amman and came to the United States as a 15-year-old, where he obviously achieved the American dream. Science doesn’t care who you are or where you came from, only what you do. He will be missed. Part the Fourth: Tell Me Once Again Why the People Are Losing Respect for Science and Scientists. What are they to think when this kind of thing is in the news? Paper mill studies get double the number of citations as genuine papers. I have been reading the biomedical literature seriously for more than fifty years. This comes as absolutely no surprise to me. The problem is that too many people of my certain age are going along with this program and publishing some of their stuff in what can only be described in lousy, predatory or predatory-adjacent online, pay-to-publish-anything “journals.” Only that which can be counted counts. Cancer research articles with telltale signs of being produced by paper mills garner double the number of citations than do genuine papers in the field, finds an analysis of tens of thousands of articles1. In a study posted on the preprint server bioRxiv, the authors report that papers that were probably produced by paper mills frequently cite, or are cited by, other potentially fraudulent articles. Paper mills are businesses that produce and sell low-quality manuscripts — often containing fabricated data and results — designed to resemble genuine research. Adrian Barnett, a statistician at the Queensland University of Technology in Brisbane, Australia, and his colleagues say that their analysis indicates that coordinated citation manipulation is inflating the impact metrics of journals in molecular oncology. These metrics measure how often a journal’s papers are cited in other research, among other things. In many nations, having papers published in journals with high impact factors is taken into account when researchers apply for jobs and funding. Research-integrity sleuths have long suspected that paper mills are inflating citations, says René Aquarius, a neurosurgery researcher at Radboud University Medical Center in Nijmegen, the Netherlands. “But it’s nice to see this confirmed in such an elegant way,” he adds. This contagion has spread to journals published by the large legacy scientific publishers. How I am going to teach medical students to dig deep enough to see this remains a mystery. I suspect this cannot be done, as one AI tool or another is prompted to review the literature on “latest chemotherapeutic advances in gall bladder cancer.” The student has no way to know what the model’s training set was. And the publishers of this trash are more than happy to be “validated” in this manner. This will not end well. Part the Fifth: AI Detectors, Do They Work? Probably not. It turns out that good writers who do not use AI, especially in the sciences, might be flagged for letting AI do their homework and write their applications to graduate school. From Universities are relying on AI-detection software to catch cheating. How well do the programs do?: Last November, Lauren Jager, a chemistry undergraduate student at Idaho State University in Pocatello, was applying to PhD programmes when she noticed that some application portals warned students about using generative artificial-intelligence tools for their personal statements. They informed students that they would use detectors to sniff out applications that contained AI-generated text. The portals weren’t specific about which detectors they were using. But they were clear on one thing: “They said that if they felt that the personal statement had been written with AI, then they would disregard your entire application,” Jager says. She didn’t think much of it — she hadn’t used AI at all — but a friend said they’d run their own statements through an AI detector on the Internet, just for safety. Jager decided to do the same with a few detectors she’d found online. “They all came back at almost 100% AI,” she says. “I started freaking out.” And well she should have. The question is: do these tools actually work? And should they be used at all if there is any chance of unjustly accusing a student, like Jager, of cheating? Academics have attempted to assess this. One 2025 paper tested GPTZero, which the authors described as the most widely used AI detector. The study found that most fully AI-generated papers were detected with high confidence, but GPTZero’s rate of falsely identifying human-written essays as AI-generated — its false-positive rate — was about 16%. The authors concluded that its “reliability in distinguishing human-authored texts is limited”. A 2023 study evaluated several AI-detection tools – OpenAI, Writer, Copyleaks, GPTZero and CrossPlag. It found they were generally better at identifying text generated by the LLM GPT-3.5 than by the more advanced GPT-4 model. When applied to human-written passages, the tools produced inconsistent results, including false positives and uncertain classifications. Clear evidence of AI detectors’ difficulties in correctly assessing human-written text was highlighted by several users on the social-media platform Reddit — they discovered that the US Declaration of Independence is often flagged as AI-written. Nature ran part of the 1776 text through ZeroGPT a number of times and was told it was between 95% and 100% AI-generated. (emphasis added). Thomas Jefferson was an imperfect human being, just like every one of us, but his intelligence was anything but artificial. In an imperfect but just world, this result would be game-set-match (we are in the Wimbledon fortnight, after all) regarding AI detectors. AI on the other hand is here to stay. This will not end well, either. Thank you for reading! See you next week. In the meantime, remember that heat is real and its damage is insidious. It is one of the joys of my life to have seen a Gila Monster run across the road on my one visit to the California desert. The thought that those wonderful lizards are being exploited for pharma profit causes me psychic agony.