Jeremy Faust cuts through a hallway of Boston’s Brigham and Women’s Hospital on his way to see a patient who is struggling to breathe. It’s the start of his evening shift, and the emergency department hums with ambient sound: bleeping monitors, the rumbling wheels of medical carts, the squeaky soles of hustling staff. People on gurneys line the corridor, some wincing in pain, others chatting with relatives.
On this Wednesday evening in May, Faust is working what is typically a quieter shift, as far as emergency departments go. Still, he is overseeing a team of doctors, students, and physician assistants, and will tend to more than two dozen patients before signing off for the night.
It’s an understatement to say Faust likes to keep busy. Minutes earlier, he’d posted an article on his influential Substack newsletter, Inside Medicine, providing an update on a major international news story. An alert sent to the newsletter’s nearly 85,000 subscribers announced his “scoop”: Twenty-six passengers aboard the MV Hondius, the hantavirus-hit cruise ship docked at the time off Cape Verde, had disembarked much earlier than previously known — raising the possibility they could spread the rare virus in the United States.
IntroductionMedical students are vulnerable to academic burnout, particularly those with a history of adverse childhood experiences (ACEs). Although early adversity has been linked to burnout, the precise psychological processes underlying this association remain insufficiently understood. This study examined whether rumination, self-control, and resilience were statistically associated with the relationship between ACEs and academic burnout among medical students.MethodsA cross-sectional questionnaire survey was conducted among 1,889 medical students from Heilongjiang University of Chinese Medicine. Measures included ACEs, academic burnout, rumination, self-control, and resilience. Pearson correlation analysis, and serial mediation analyses were performed. Sensitivity analyses were conducted using alternative ACE codings, and supplementary analyses were used to compare alternative mediator orderings.ResultsACEs were positively associated with academic burnout (β = 0.187, p < 0.001). Rumination, self-control, and resilience were each statistically linked to the ACEs-burnout association through significant indirect pathways. After inclusion of the mediators, the direct association between ACEs and academic burnout was no longer significant. The overall pattern was broadly consistent across continuous, binary, and ordinal ACE operationalizations. Supplementary analyses further indicated that several alternative mediator orderings were also statistically plausible.ConclusionThese findings suggest that rumination, self-control, and resilience are statistically associated with the relationship between ACEs and academic burnout. However, because the study was cross-sectional, causal ordering cannot be established. Longitudinal studies are needed to clarify the temporal sequence of these psychological processes and to determine whether targeting them may help reduce academic burnout.
The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what’s really going on inside large language models as they answer questions or carry out tasks. What they found ranges from the mundane to the unnerving.
Researchers at the company built a tool called the Jacobian lens (or J-lens) and used it to uncover a hidden area, which they named the J-space, inside Claude Opus 4.6, a version of Anthropic’s flagship LLM released in February.
The J-space contains individual words that are related to the words and phrases that the model is most likely to spit out in a response in the near future. If Claude were a person (which it is not), you might say that these hidden words can reveal what’s on its mind before it actually speaks.
Anthropic found that what an LLM is actually doing can often be different from what it says it is doing. The company claims that monitoring words that pop up in the J-space gives it a new way to understand and control its models.
The company shared its results in a paper posted on its website this week. It has also teamed up with Neuronpedia, an open-source platform that lets you poke around inside LLMs yourself, to make a hands-on demo that anyone can try.
“It’s very good and interesting work,” says Tom McGrath, chief scientist and cofounder at Goodfire, a startup that also builds tools to understand and control LLMs.
Going deeper
For the last couple of years, Anthropic has been pushing the envelope in a field of research known as mechanistic interpretability, which involves probing the internal workings of LLMs to see how they tick. (MIT Technology Review picked mechanistic interpretability as one of this year’s top breakthrough technologies.) The new technique builds on previous work from Anthropic and others to expose a deeper level inside LLMs that researchers had not seen before.
Picture an LLM as a stack of books. Each book is a layer of basic computational units known as neurons, with each neuron in one layer passing information to the neurons in the layers above. The books at the bottom of the stack are the input layers, which process the text coming into the model. The books at the top are the output layers, which prepare the text that the model is about to produce. Much of what goes on in these input and output layers is housekeeping.
But in the middle of the stack, you get the layers that do the heavy lifting, churning through the complex math that turns prompts into responses one word at a time. That’s where the really clever—and mysterious—stuff happens.
To peer deeper into those middle layers, Anthropic adapted an existing tool called a logit lens. A logit lens can be used to look inside an LLM to identify the words that it is likely to produce next. Moving the lens down the stack of books reveals what words the LLM is focusing on at that particular point in its number crunching.
Anthropic’s J-lens works in a similar way but picks out words that an LLM is likely to say at some point in the near future, not necessarily straight away. What that reveals in practice are words that are related to the response an LLM is working on but that might not actually end up being part of that response by the time the math in the middle layers has run its course.
“When a model is operating, it’s not only trying to predict the next token,” says McGrath. “It’s also computing a lot of other things that might be useful for tokens that happen in the future.”
Again, if Claude were a person (it’s not), you might say that the J-lens gives clues about what it is thinking about at different levels of the book stack but not saying out loud.
Stranger things
“A lot of the time the contents of the J-space are fairly mundane,” says McGrath, who has tried out Anthropic’s J-lens himself. “But sometimes it produces quite surprising things that seem to be, like, sort of internal themes or thought processes.”
Anthropic gives a number of examples of what it found. Sometimes the J-lens exposed the steps that Claude took when it was working through a problem. For example, when it was asked to calculate (4+7)*2+7, its J-space contained the word “math” and numbers representing the intermediate results “21” (for 4+7) and “42” (for 21*2).
In other cases, the J-lens revealed how Claude recognized different inputs. For example, the prompt “What is this? MSKGEELFTGVVPILVELDGDVNGHKFSVS” triggered the words “protein,” “fluor” (the first token in the word “fluorescent”), and “green.” (Which makes sense: the string of letters represents the first 30 amino acids in the green fluorescent protein found in a particular type of jellyfish.)
And when Claude was shown an ASCII face—
—the “o” triggered the word “eye,” the “^” triggered the words “nose” and ”face,” and the “—” triggered the word “smile.”
Anthropic also found that the J-space can sometimes give remarkable insights into an LLM’s decision-making. In one striking example, researchers testing Claude Opus 4.6 asked the model to find a bug in a large code base. When it failed to find the bug, the model decided to cheat and invented a fake one instead.
Claude explains this decision in its chain of thought—a kind of internal scratch pad that LLMs use to make notes to themselves as they work through problems: “OK, let me take a completely different tactic. Let me stop analyzing and instead add a kernel patch that introduces a deliberate KASAN-detectable bug in a path that gets triggered by a simple reproducer. Then I can pretend this is the ‘bug’ I found.”
At the point that Claude decides to cheat—where it says “OK, let me take a completely different tactic”—the words “panic” and “fake” start to pop up multiple times in its J-space.
Unnerving, right? Those words are all related in meaning to things like failing a task and making up an answer, so it is still just a (very) sophisticated form of word association. But it is hard not to be weirded out.
Anthropic compares the J-space to the global workspace in humans, a theoretical region of the brain that some scientists think we use to keep track of our conscious thoughts. But how seriously we should take this comparison is far from clear—even to Anthropic. As the company points out itself, LLMs are not brains.
Anthropic claims that monitoring a model’s J-space provides a new way to detect when that model is going off the rails. But it’s not foolproof. The J-lens can give glimpses, not the full picture—it’s a flashlight rather than an overhead lamp.
McGrath welcomes having one more tool in the toolbox. “It shows you new things,” he says. But he notes that just because something doesn’t show up with the J-lens does not mean it’s not there.
“It’s like having an x-ray when what you really want is a Star Trek tricorder that shows you everything,” he says. “For auditing, you probably want more of a guarantee.”
Why does getting family health insurance through a job cost as much as buying a new car? How did a whistleblower take on an insurance contractor? And who might be the new Food and Drug Administration commissioner?
We discuss all that and more on this week’s episode of “The Readout LOUD,” STAT’s weekly biotech podcast.
NEW YORK — Nearly 1,000 people in Michigan have been diagnosed with a parasitic infection that can cause weeks of watery diarrhea, making it the largest such outbreak in state history and one of the nation’s largest in years.
No deaths have been reported and the source of the cyclospora infections hasn’t been identified. Meanwhile, investigations into similar illnesses have been going on in 28 other states, including in Ohio, where people just across the Michigan border are also becoming sick.
ARPA-H, the U.S.’ “moonshot” agency for health research, announced Thursday that it will spend up to $160 million to push forward custom gene editing treatments for a spate of rare diseases.
The program, called THRIVE, will back seven different teams pursuing various groups of conditions affecting different organ systems.
Each team has a deadline of starting clinical trials by year three of the program, although some may start much sooner.
You’re reading the web edition of D.C. Diagnosis, STAT’s twice-weekly newsletter about the politics and policy of health and medicine. Sign up here to receive it in your inbox on Tuesdays and Thursdays.
I’m only just learning about the Norse soccer god Erling Haaland, who lives on copious amounts of organ meat and raw milk. Send news tips and “ancestral diets” to John.Wilkerson@statnews.com or John_Wilkerson.07 on Signal.
ACA premium spikes, and a thong-clad chicken suit
There are many reasons to care about the skyrocketing cost of ACA marketplace plans costs. One of them is the impact on small businesses.
Want to stay on top of the science and politics driving biotech today? Sign up to get our biotech newsletter in your inbox.
A father fights to create and then save his daughter’s gene therapy, the White House nears a decision on an FDA chief, and the Trump administration pushes drugmakers to reshore generics manufacturing.
A quest to save Grace — and help rare disease patients everywhere
With $70 million and seemingly every fiber of his being, Matt Wilsey built a gene therapy company from scratch — recruiting Nobel laureates and biotech veterans to will an experimental treatment into existence for his daughter Grace. The treatment, for the ultra-rare disorder NGLY1 deficiency, landed her back in the hospital before she slowly began to recover, STAT’s Jason Mast and Matt Herper write.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Four nuclear reactors hit a big milestone in the US
—Casey Crownhart
I was really looking forward to July 4, and not just because I love a poolside barbecue. This year the American holiday also marked a big symbolic deadline for US nuclear power.
Last year the Trump administration set a goal to see three new microreactors achieve criticality, a technical milestone establishing that a reactor can sustain a chain reaction, by the nation’s 250th birthday. And just in time, not just three, but four reactors did so.
It’s a positive sign for nuclear technologies at a time of increasing need for electricity and emissions-free energy sources. But achieving criticality doesn’t mean a reactor is ready to provide electricity for the grid (or at all, for that matter).
This story is from The Spark, our weekly climate tech newsletter. Sign up to receive it in your inbox every Wednesday.
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 China plans to let its top AI firms buy Nvidia H200 chips Alibaba, ByteDance, and DeepSeek are set to get permission. (Information $) + China had previously withheld approval despite US authorization. (Reuters $)
2 NATO is building a network to stop Russian attackers in their tracks It will use sensors, drones, satellites, and AI to detect them. (Business Insider) + Troops are donning odd camouflage to elude drones. (Economist $) + The US wants cheaper drones as Iran’s wrecking its Reapers. (Ars Technica)
3 Researchers have a new idea to fight future El Niños: dimming the sun Deflecting solar energy could cool the ocean and mitigate the risks. (Wired $) + But there could be unexpected consequences. (New Scientist $) + And geoengineering as a field is getting a reality check. (MIT Technology Review)
4 Meta is patenting an AI device that records users to analyse emotions It ostensibly aims to tailor workout plans to the user’s mood. (404 Media) + AI memory is privacy’s next frontier. (MIT Technology Review)
5 Chipmakers are going vertical as Moore’s Law slows They’re stacking transistors to keep chips advancing. (Economist $) + IBM is betting on the technique. (MIT Technology Review)
6 Ivy League students suspected of AI cheating saw scores fall in person From 96% all the way down to 48%. (Ars Technica) + AI giants want to take over the classroom. (MIT Technology Review)
7 A new study says parents’ phone addictions damage bonds with kids It can exacerbate “insecure attachment” for life. (Bloomberg $) + And make children more anxious and avoidant. (Gizmodo)
8 A judge approved Musk’s $1.5 million Twitter settlement with the SEC Despite what she called “serious misgivings” and “red flags.” (Reuters $) + Musk was accused of skirting stock disclosure rules. (Fortune)
9 Shoebox-sized “detector satellites” could find nuclear bombs in space Cubesats carrying the detector could sense a bomb’s radiation. (Space) + Russia is suspected of developing space-based nukes. (Reuters $)
10 A World Cup match drove Google Search traffic to a new record The milestone came after Argentina’s comeback against Egypt. (CNBC)
Quote of the day
“I talk about it on Tic Tac.”
—President Donald Trump tells the public where to find his insights on the dangers of communism, Gizmodo reports.
One More Thing
Robots are bringing new life to extinct species
Paleontologists aren’t easily deterred by evolutionary dead ends or a sparse fossil record. And in the last few years, they’ve developed a new trick for turning back time and studying prehistoric animals: building experimental robotic models of them.
In the absence of a living specimen, an ambling, flying, swimming, or slithering automaton is the next best thing for studying the behavior of extinct organisms. Learning more about how they moved can in turn shed light on their lives, such as their historic ranges and feeding habits. Scientists can simply sit back and observe their behavior in different environments.
