Part the First. Tales from the Crypt. Subtitled The lives of 17th century Milan’s working poor – their health, diet, and drug habits – emerge from thousands of bodies buried under a public hospital. This article appeared in Science on 1 May 2025:
In 1456, the Duke of Milan established a medical institution dedicated to caring for the city’s poor and sick on a scale unprecedented in Europe. Built in the center of medieval Milan, the massive Ospedale Maggiore was soon dubbed Ca’ Granda—“Big Factory” in the local dialect. The nickname was a reference to both its scale and its ambition: to heal Milan’s working class as efficiently and effectively as possible. Imagine that. The Duke of Milan was the business Jeff Bezos of his day and cared about “his people.”
For 60 years, it also had its own system for disposing of the dead. Between 1637 and 1697, people who died at the hospital were dropped into brick-lined underground vaults below a newly built church. Hospital planners expected the remains to skeletonize quickly, but a cool, moist microclimate slowed decomposition. Bodies accumulated and stank, eventually forcing the hospital to seal off the chambers and begin burying people on the outskirts of town.
So far, a team of archaeologists, geneticists, botanists, and specialists in forensic medicine has examined more than 300,000 bones, out of an estimated 2.9 million preserved in the underground vaults. They are applying an array of analyses to the remains, providing a multidisciplinary look at the unnamed people who labored in Milan during a time of turbulent social change. Italian culture is broadly supportive of scientific study of the dead, making it possible to ethically study the remains as part of an effort to understand Milan’s past.
The site is also described in the journal Medieval History here. And from the Science article, these few interesting tidbits:
Some of the skulls still contain brain tissue…Gaia Giordano, a forensic toxicologist found a wide array of psychoactive substances, including morphine, codeine, noscapine, and papaverine—all derived from the opium poppy. The hospital’s detailed pharmacy inventories note some of those drugs, but there is no record of another group of substances, cannabinoids, that Giordano found in multiple bone samples. She thinks patients may have used marijuana recreationally.
In a bigger surprise, Giordano…found traces of coca in the preserved brain tissue of nine Ca’ Granda patients. It’s the earliest known use of the plant outside of South America, predating the first historical mention of coca in Europe by at least 200 years. “Backdating the plant almost 2 centuries in Europe is pretty amazing…People were probably using it the same way in Milan and South America, as an energy tonic to combat fatigue and hunger.”
Ha! People are much the same everywhere, even in 17th-century Italy.
Part the Second. Their Good Work Sees the Light of Day. CDC keeps publishing papers after firing scientists who made the research possible. No surprises here, but despite inarguably poor leadership at during the pandemic, we need to remember that it was CDC scientists who responded so effectively in the early days of the HIV/AIDS epidemic and saved countless lives in the process. From the article:
Before it became a national scandal, the lead-poisoning-from-applesauce case was just two little kids with concerning blood test results in Hickory, N.C. A state inspector drove out with local health officials in June 2023 to try to find the source. He powered up his X-ray fluorescence analyzer — like a cross between a laser gun and a power tool — which emitted a beam that dislodges electrons, coaxing out chemical fingerprints, and pointed it at surface after surface. Doors, door jambs, walls, couches, windowsills, blinds, toys, siding strips, 150 or 200 shots in all.
There was a bit of lead paint, but hardly enough to explain blood lead levels of over 10 micrograms per deciliter…When he got his other samples back from the lab — water from the tap, sand from the play pit, a dust wipe from the father’s shoes — those were negative, too. “In the meantime,” said Alan Honeycutt, a regional environmental health specialist at the North Carolina Department of Health and Human Services, “both children’s blood lead had gone higher.”
To him, that pointed toward something in their diets…Within 72 hours, the mother called to say there was something she’d forgotten to mention: the applesauce packets her kids ate every day.
So began an investigation that would reveal 566 lead-poisoned children across 44 states, Puerto Rico, and Washington, D.C., and would eventually get the adulterated applesauce off shelves. But in late April, when the U.S. Centers for Disease Control and Prevention published a paper on how that nationwide sleuthing went down, its fine print left a key detail out. At least six of the authors who’d worked at the CDC had been laid off earlier that month, when their entire division was slashed by the Trump administration’s cuts…the paper was a record of what had been lost, of what might not happen if a food product were poisoning kids right now.
This is what happens when people are fired for no good reason, no thought for the morrow. Next time, expect the lead-laden applesauce to keep right on poisoning…Red State, Blue State, Purple State – the poison will not care. No one familiar with the inner workings of any corporation, government agency, college or university, courts system, or legislature would deny room for improvement, and I say this as someone now a part-time denizen of the academic adminisphere. The chainsaw approach will never work, though. Advances that stick are incremental, scientific and otherwise. When you “move fast and break things,” (barf bag recommended if you click on the link) all you do is break things, even when you make a lot of money in the process. As for the notion that cutting to the bone will improve efficiency, the crude measure of efficiency as the same, or more, work done by fewer people fails take into account the inverse relationship between efficiency and effectiveness.
Part the Third. Does Secretary Kennedy Really Care About Chronic Disease? The dismantling of CDC continues as:
The Department of Health and Human Services’ budget for 2026 (pdf) released Friday, proposed $14 billion in discretionary funding for programs that aim to reverse the chronic disease epidemic, but it would also abolish the CDC’s National Center for Chronic Disease Prevention and Health Promotion.
The excuse offered for this is that the National Center for Chronic Disease Prevention and Health Promotion will be moved to something called the Administration for a Healthy America (AHA). Fine. Do this after AHA exists and the other agencies slated for “reorganization” can also be moved without their dissolution. The emphasis by RFKJr regarding the unhealthy American diet is not misplaced. But there is no solution to that without rebuilding the “Great American Food System” from the ground up. And that will require the Trump v2.0 Administration to take on Big Ag and Big Food. Those at the top of the income and wealth distributions can eat well and exercise on the ski slopes of Europe and North America on their own accord. The vast majority of other Americans, not so much.
Part the Fourth. The End of Soft Power and the American Empire. I have never lost my conviction that if Americans, as the benign hegemon of our dreams (I know, but bear with me here) act according to the “better angels of our nature,” the world will become a place where all people can flourish. But then this happens: NIH grant cuts will axe clinical trials abroad — and could leave thousands without care.
Science admits of no political borders, period. Tuberculosis, malaria, HIV/AIDS are human problems. The United States can afford to address them, and this is not what a serious nation does:
Amita Gupta has spent more than a decade planning and running a US$70-million trial to study a new tuberculosis drug, enrolling nearly 6,000 participants across 13 countries. It might all have been for nothing.
Gupta’s trial has been ensnared by a US National Institutes of Health (NIH) policy announced on 1 May that bans ‘foreign subawards’, which are funds that a US grant recipient can give to an international collaborator to help complete a project. Since the policy was disclosed, NIH employees have been forbidden from issuing grants involving such awards, according to internal documents that Nature has obtained.
As a result, funding will abruptly cease for dozens, if not hundreds, of ongoing trials of experimental drugs and treatments. The change puts thousands of trial participants, as well as the scientists running the trials, in limbo.
TB is the world’s deadliest infectious disease (1.3M deaths in 2022), but it is not much of a thing in the US, now. Still, TB killed 565 Americans in 2022. Could future deaths be prevented, here and there? We may never know if this TB drug trial is interrupted, which will be its end. Count on this: One of the arguments against further research on TB will be that we “already wasted $70M on a TB drug trial.” This, of course, without including in the assertion that $70M was wasted because the trial was halted prematurely, by political fiat.
A common complaint in certain circles is that CDC has grown because of “mission creep.” CDC began as the Office of National Defense Malaria Control Activities in 1942. From 1946-1967 it was later the Communicable Disease Center, to be followed by Centers for Disease Control, 1980-1992, until CDC became the current Centers for Disease Control and Prevention. This is not mission creep. Instead, the evolution of CDC corresponds to advances in our understanding of the nature of human disease. The only reason to return CDC to “Communicable Disease Center” is because of the self-absorbed and mistaken belief that infectious diseases are no longer a problem in the so-called Global North.
Part the Fifth. Can AI Help a Scientist Manage the Scientific Literature? I come to this as biological scientist and teacher and tutor of preclinical medical students (first- and second-year students learning biochemistry, genetics, immunology, physiology, anatomy, and the other foundations of clinical medicine). I am working on a longer piece about AI in Medical Education, and one of the major questions is whether AI correctly summarizes the current state of knowledge. I will need a pair of yellow waders to continue.
My priors will be no surprise to anyone who has read my work here. Too many medical students have always searched for what I call the “magic fairy dust” that will make learning the fundamentals of medicine “easy.” Some of them now believe they have found this dust in various and sundry AI apps at their disposal. Those apps with a MedEd focus are subscription-based, often with one or more provided by the medical school because “other schools are doing it.” Yes, what never worked with your mother works with administrators. Anyway, I suppose these apps are preparing the students to be proficient users of UpToDate when they enter the independent practice of medicine.
A recent paper from the Royal Society, Generalization bias in large language model summarization of scientific research, begins with:
Artificial intelligence chatbots driven by large language models (LLMs) have the potential to increase public science literacy and support scientific research, as they can quickly summarize complex scientific information in accessible terms.
But their unsurprising conclusions may be summarized as follows:
- LLMs may omit details that limit the scope of research conclusions, leading to generalizations of results broader than warranted by the original study.
- When explicitly prompted for accuracy, most LLMs produced broader generalizations of scientific results than those in the original texts.
- In a direct comparison of LLM-generated and human-authored science summaries, LLM summaries were nearly five times more likely to contain broad generalizations.
- There is a strong bias in many widely used LLMs towards overgeneralizing scientific conclusions, posing a significant risk of large-scale misinterpretations of research findings. We highlight potential mitigation strategies, including lowering LLM temperature settings and benchmarking LLMs for generalization accuracy
Does this bias extend to parsing the medical literature and standard medical textbooks? Who knows, but as two recent Links have shown, AI tends to make people stupid. This is not a trait anyone wants in his or her physician, or anyone else for that matter. I cannot remember the last time I asked a medical student a question in tutorial group and gotten an answer before tap-tap-tap-tap-tap on a keyboard. I wonder what will happen next year when the first thing I say in my Group is, “OK, now close your laptops while we discuss the current case in the syllabus.” No, I don’t wonder, but I expect to hear from the Offices of Student Affairs and Academic Affairs.
In answer to the question, “Can AI help a scientist manage the scientific literature?” No, but it can help a scientist be lazy and stupid. Most of us will remember being told that a true education teaches you to know where to look something up. Not so much, then or now. And I’m still put out by an English Literature teacher who told our class that the only reason we were in the room was so we could sound smart about Henry James at a cocktail party in ten years. I can take or leave Henry James except for “Daisy Miller” (apologies to LS), but it was difficult to leave the insult.
Part the Sixth. The Nature of Art 43,000 Years Ago. In another episode of Science Is Cool:
Scientists in Spain say they have discovered the oldest full human fingerprint after unearthing a rock which they say resembles a human face and suggests Neanderthals could make art.
A Neanderthal man is believed to have dipped his finger in red pigment to paint a nose on a pebble around 43,000 years ago. The rock was discovered in the San Lázaro rock shelter in Segovia, Spain.
The “strategic position” of the dot has led scientists to see it as evidence of Neanderthals’ “symbolic behaviour”, suggesting they had the ability to think about things in an abstract way.
The open access paper Archaeological and Anthropological Sciences is here. Can we be sure the face in the rock shelter was completed by a finger dipped in ochre pigment that was brought into the shelter from somewhere else? Yes. And it sure looks like a face and the red nose left a fingerprint. This seems to be an artifact in the better sense of the word.
So, let us leave our fingerprints on the world, primarily by paying attention and doing what we can.
Suggestions welcome for Coffee Break topics of conversation. See you next week.
Fascinating and pithy roundup! Thank you, KLG.
I will confess, I have not yet read the post… I was going to post this in Links as a general thought, but I suspect it fits here given that AI is a discussion point. If I am wrong, please forgive.
I’m not sure if the articles were the same or if they were totally unrelated, but within the past couple days, posted on this site, I’ve had the opportunity to contemplate AI and what it means to be sentient.
Anyone who’s read my comments knows I hate the name Artificial Intelligence, because it is not intelligent. And it is NOT intelligent because it is not sentient, and it is not self-aware.
Yes, it may be computationally brilliant and possibly very dangerous depending on what abilities it is given and what information it is provided, but it will have no self-awareness of its actions or reason to care about those actions.
If you have never read the book by Carl Jung called “Answer to Job”, I recommend that you do read it.
Jung asks us to reconsider the idea of how omniscient and all-knowing being can really be??? Infinite awareness requires you have awareness beyond oneself. Jung postulates that you cannot have awareness beyond yourself without reflection and reflection requires you to stop and question your past actions, information and assumptions.
AI is not capable of that and it’s evidenced by the research that says AI degrades when AI generated content is used to train AI.
For example, it would be reasonable to assume that a smart human would at least start mentally questioning things if all of a sudden his inputs were producing bad and unexpected outcomes even if they people providing him those input swear up and down they are as good as before. No AI model is NOT going to challenge the data. At best the AI is going to work overtime to reduce the error of the output, but it has no AWARENESS or concept that it is being fed garbage or even intentionally being gaslight!
It has no ability to separate itself from the statistical math and say, wait something is really wrong here or something has radically changed here. No, any revelations like these come from the human minders of the system. At best the AI system will start flagging a high rate of unexpected errors, but it is NOT going to ask the WHY in why are we being wrong. You have to be sentient and capable of asking questions beyond your own existence. AI on knows what it is given – we don’t exist beyond AI’s 1’s and 0’s as data records.
You may be unduly optimistic.
The capacity for self-reflection may be well on the way to elimination (think Trump and Musk).
It is now so easy to simply be emotionally reactive (social media of all types) and to be rewarded for such a response.
Real thinking is becoming more and more a threat and once that capacity is minimized AI may be welcomed with open arms.
Just to pick up on one part of your general thought:
I don’t follow this relationship you draw between intelligence and sentience. You are saying there is no intelligence without sentience and self awareness?
Lots of the current crop of AI (and other earlier)
– they absolutely display a degree of intelligence – the ability to learn and apply that knowledge and problem solve, to acquire skills. I can tell Claude, say, where to hone in to acquire specialist knowledge and it will give better answers to problems. or it can add in new skills by making use of external tools. That is intelligence. It’s not strictly needing consciousness or sentience to derive it. It’s been argued that fungi display a simple form of intelligence, exhibiting decision making, memory and communication. so there are degrees of intelligence.
Sentience
– needs consciousness, subject experience, emotional awareness, sense of self. We have it. We can see it in higher order animals like dogs, cats, octopuses. But in AI? No. Really no.
I would argue current AI- displays intelligence, not sentience.
I don’t see how you can say current AI is not intelligent because it is not sentient. They are two separate things to me.
Turn it around and it is like saying there can be no intelligence anywhere without sentience.
Yet we see intelligent behaviour in fungi with no central nervous system.
Science is going to need gangs of KLGs.
I enjoy and read KLG’s posts. However, I do not believe I have seen any of your comments on my areas of interests: circadian biology, substance abuse addiction, behavioral addiction, and all the intersection within. I have strong interests in this because of my own problems with addiction and circadian dysfunction. Here are just two recent studies I read.
Sweetened caffeine drinking revealed behavioral rhythm independent of the central circadian clock in male mice
Circadian rhythms and substance use disorders: A bidirectional relationship
I use to pound far too many cans of caffeinated soda or tea for energy, wakefulness, and happiness. (But never coffee, strangely enough. It’s too burnt and bitter.) This was pretty foolish, and I suspect I am still suffering the consequences even though I quit caffeine and sugar completely.
Thanks, as always, for this great way to end the week!
Although I mostly loved being an academic for the last ~ 20 years of my working life, I am SO GLAD to have made my escape just before the tsunami of AI sewage hit. I still get a certain number of emails from my old workplace, and there are just enough of them about AI to give me a little zing of joy every time I open them. Got out just in time …
But, in truth, I was already feeling pretty played out by the time of my departure almost 2 yrs ago. At that point, I was the last person in a dept with two dozen faculty who regularly masked indoors – with those colleagues including several PhD biologists, one of whom was claimed by COVID just a couple of months after I left. And these co-workers included some who had lost relatives to COVID and at least one who clearly has Long COVID. It was so demoralizing by the end to feel so disillusioned and disappointed in people that I’d respected as colleagues and enjoyed having as friends. I try not to imagine how this experience is replicated so many millions of times over in workplaces all over the world.
It is a shame that the old Jay Gould quote that; “I can hire one half of the working class to kill the other half,” has now become; “I can convince one half of the working class to infect the other half.”
Stay safe.
Re: 15th century drug use… Ms Giordano needs to collaborate with some Egyptologists. Coca has been found in Egyptian mummies as early as the time of Ramses II. How’d they get that? Hmm…. Could there have been global travel back then? I think so.
Science is all well and good, and I’m generally a fan. BUT… I’m always shocked at how ‘siloed’ modern scientists are. They need to get out more, and read each other’s research from fields not directly related to their own narrow interests.
“It’s the earliest known use of the plant outside of South America,”
“So how did Africans in Egypt get a hold of cocaine? They picked it. At least 10 species of Erythroxylaceae exist throughout the African continent. Not the least of which are Erythroxylum fischeri and Erythroxylum emarginatum. At least two more species are found on the island nation of Mauritus, just east of Madagascar in the Indian Ocean. . . . And this is what’s truly cool about the research that Balabanova and Pasche (along with others) did: they showed us yet another way the Egyptians made use of their local resources, probably in attempt to heal or offset pain. Remember the ages of Balabanova’s subjects? Many were very young. We probably don’t know the causes of death, but it obviously wasn’t old age. If they were afflicted with some illness or malady, they may very well have been in pain and taking medicine for it. The African cocaine plant is used even today to treat pain.”
https://ahotcupofjoe.net/2018/05/new-world-drugs-and-old-world-mummies/
Further, displays of self medicating behavior appear to be hardly exclusively unique to the human ape-animal, for example,
“For several decades, evidence has accumulated that animals turn to medicinal plants to relieve their ailments. Chimpanzees (and some other species) swallow leaves to mechanically clear the gut of parasites. Chimps also rely on the ingested pith of an African relative of the daisy, Vernonia amygdalina, to rid themselves of intestinal worms. Dolphins rub against antibacterial corals and sponges to treat skin infections. And recently, a male Sumatran orangutan was observed chewing the leaves of Fibraurea tinctoria, a South Asian plant with antibacterial and anti-inflammatory properties, and dabbing the juice onto a wound.”
https://www.science.org/content/article/chimps-use-more-plant-medicines-any-other-animal
https://www.ox.ac.uk/news/2024-06-21-study-shows-wild-chimpanzees-seek-out-medicinal-plants-treat-illness-and-injuries
Orgone,
Fascinating! Thanks very much for directing me to this research. I’ll have to watch this topic going forward. In this particular case, proponents of pre-Columbian trans-Atlantic trade may have a weak argument, indeed. That said, it’s not the only argument!
Thanks for keeping the discussion honest. :-)
There’s other sources of stimulants besides coffee, tea, and coca leaf. Some other culturally notabke ones are: southern South America drinks yerba mate. South and Southeast Asia chews betel nuts. East Africa and nearby chews khat leaves.
‘Can AI Help a Scientist Manage the Scientific Literature?’
Dug into it a bit and found another of these you-beut apps-
https://med.stanford.edu/news/all-news/2025/06/chatehr.html
Stanford of course. But if in asking these LLMs, they omit details, produce broader generalizations of scientific results than those in the original texts, are nearly five times more likely to contain broad generalizations and have a strong bias towards overgeneralizing scientific conclusions, well, how is that any different in asking a newly-minted medical student fresh to the wards on his first day? They too would use the old adage that if you can’t dazzle them with brilliance, then baffle them with bull. Of course the excrement will really hit the rotating, oscillatory device when a medical student ends up killing a patient because they just followed the advice of an LLM without thinking or questioning first. In the old days a medical student would crack open a text book and breath deeply of the knowledge therein hoping that it would stay within long enough to be used, especially at exam time. Can’t do the same with an LLM as they are not authoritative but are fully capable of just making stuff up on the spot.
Thanks for that Milanese archaeological article as I intend sitting down with a coffee this ‘arvo to read slowly through it. As for suggestions for Coffee Break topics of conversation, I have one. As Trump & Co. are taking a wrecking ball to medical research in the States for reasons, how about a quick summary of which countries are also leaders in medical research that may take up the slack. Should we expect medical researches in future having to get subscriptions to Chinese and German medical journals to keep abreast of the newest developments in their field?
Noting: “Italian culture is broadly supportive of scientific study of the dead, making it possible to ethically study the remains as part of an effort to understand Milan’s past.” For two reasons: Italian dead are awfully lively. Heck, nonna isn’t dead — she’s in an upper level of purgatory making ravioli. Also, there simply is a long tradition of interest in cadavers, “la cara salma” — I was just in Gubbio, where several of the churches were displaying, in typical Italian fashion, holy bodies wearing embroidered robes in glass cases.
Also, hypochondria is one of the Italian national pastimes, along with ironing.
The body in Italy is not treated the same way as is the body in puritanical U.S. culture, eh.
Thanks for the semi-local tips on drug use: For further information on what people ate and drank in the baroque and early modern eras in Italy, I recommend Carlo Ginzburg, famous “micro-historian,” and Piero Camporesi, a leading Italian scholar of the history of food and cookery in Italy.
Regarding AI for scientific summary, I’ve been working on building a small “RAG Pipelines” for searching and summarizing a collection of documents (which is how all of these academic paper summarizer tools work). It is clear from the outset that responses to queries are unreliable at best and almost always require reading the source material if you with to have even the slightest bit of confidence in any summary or response.
At worst, it is entirely unintelligible. The recommended method of “tuning” responses to seem more relevant is arbitrarily modifying context information, queries, etc. until it “feels” right. The people building these tools aren’t software engineers designing logic systems, they’re alchemists turning knobs until the vibe feels right and then selling their snake oil onward.
“Summarizing complex scientific information” as recent AI tools purport to do, is an activity that requires actual thought and intelligence and is therefore not suited to LLM AI machines, which work by merely predicting the most likely next word in a sequence according to their training data.
Ask any researcher who is trying to write a summary of their own single paper, and has full and deep understanding of the subject matter, whether this is a simple process. It’s certainly not a mechanistic process involving calculations on word ordering in the body of the paper.
Also missing here is the reality that “summarizing complex scientific information” has a deeper purpose than just producing a written summary. The idea is to force oneself to attain a deep and thorough understanding of the scientific information in question, to turn it over in one’s mind, to compare it against prior knowledge one has, and to improve one’s level of expertise in this area of the field. By this process, scientific workers get smarter and work at a higher level in the future. This is generally known as the march of science.
Not a small thing to give up on.