Health & science

Why $3bn school grants and other well-meant policies changed nothing

• Adam Mastroianni lists programmes that showed no measurable effect: $3 billion of US school grants in 2009, a 10-year trial of laptops for children across 531 Peruvian schools, and counselling meant to strengthen low-income families.
• Even a $100 million gift that made NYU's main medical school tuition-free left its students and their specialty choices unchanged, though applications rose 47%.
• His tongue-in-cheek fix: a goat-sacrifice ritual before any big policy, to force humility about how little we know about changing behaviour, while still testing and trying.

Other side: Null results can reflect weak measurement or short time frames, and the author concedes some programmes work in some places. Giving up on reform would be the bigger mistake.

See the KKReckons explainer →

Summary based on Experimental History. Read the full story ↗

How Gemini helped a WSJ writer get an endometriosis diagnosis

• WSJ writer Katherine Bindley spent two decades with pain episodes, ER visits and negative tests. When she gave Google's Gemini her history, it said she likely had adenomyosis and endometriosis, which earlier scans and surgery had not found.
• An MRI showed no sign of endometriosis, but a specialist's dynamic ultrasound led to both diagnoses, and two more specialists agreed. The disease typically takes eight to 12 years to diagnose. She now plans excision surgery.
• Harvard's Isaac Kohane warns that AI can be wrong and stir anxiety, yet urges patients to use it to prepare for appointments.

Other side: This is one patient's story, and Gemini also got the MRI wrong. Chatbots tend to tell users what they want to hear: for each missed diagnosis they catch, how many healthy people do they send chasing scans and surgery?

See the KKReckons explainer →

Summary based on The Wall Street Journal. Read the full story ↗

mRNA cancer vaccines show early wins in pancreatic and skin cancer

• NPR profiles Donna Gustafson, 72, the first person to get a personalised mRNA vaccine for pancreatic cancer, about seven years ago, alongside surgery and chemotherapy. She shows no current evidence of the disease.
• In the 16-patient Memorial Sloan Kettering trial she joined, the vaccine triggered a response in half, and seven of those eight are alive. Separately, Merck and Moderna reported last month that their vaccine cut skin cancer recurrence and death by 42% over two years.
• Trials are running in lung, kidney, bladder and brain cancers, though some have failed for reasons researchers don't yet understand.

Other side: Sixteen patients with no described control group is a signal, not proof. Some mRNA cancer trials have failed, and a vaccine made for each patient raises hard questions about cost and speed.

See the KKReckons explainer →

Summary based on NPR. Read the full story ↗

Would you pay for the best baby? Embryo screening and liberal eugenics

• Jan Grue, a writer misdiagnosed as a child with spinal muscular atrophy, starts from Nucleus Genomics' New York subway ads promising parents their best baby. Its IVF package screens up to 20 embryos for diseases and for traits from eye colour and height to insomnia and baldness.
• He argues that population-level heritability figures, like those on the posters, cannot tell parents how tall or clever one child will be. He calls consumer-led selection liberal eugenics: the old project of improving people, rebranded as individual choice.
• Grue does not oppose cures. His case is for changing societies to fit varied bodies, rather than optimising embryos to fit a narrow ideal of the productive, never-sick worker.

Other side: Screening can spare families serious inherited illness, and parents already buy schools, tutors and diets to give children an edge. If the science improves, critics will need to say where preventing suffering ends and unacceptable selection begins, and who should draw that line.

See the KKReckons explainer →

Summary based on The Guardian. Read the full story ↗

Green groups embrace AI even as they warn of data centre costs

• Reuters asked 15 major environmental and conservation groups about AI. All use it, for climate data, fire alerts and wildlife counts, and all worry about the power and water data centres consume.
• Save the Elephants says AI reviews survey images in hours that took nine people nine months, and showed human counts missed about a third of elephants. The Jane Goodall Institute cut a 13 to 15 year notes project to about two.
• Some limit the footprint: WRI caches answers and The Nature Conservancy processes fishing video on board. Greenpeace questions the buildout itself.

Watch next: COP31 this month, where host Turkey will seek a pledge on sustainable AI. Even so, savings from careful use by charities are tiny next to a buildout driven by tech giants, so the real question sits elsewhere.

See the KKReckons explainer →

Summary based on Reuters. Read the full story ↗

Hackers heat fridges at French blood banks and European supermarkets

• Cold rooms at three French blood service sites were driven to very high temperatures on 2 November, Wired reports. Alarms and staff caught it and the blood was unharmed. Supermarkets in at least 11 European countries have binned food after similar attacks.
• Many hacks used internet-exposed Danfoss temperature controllers on outdated firmware, some with weak passwords based on the date. Stores on US military bases were hit in August.
• Who is behind it is unclear. Some researchers see signs of Russia- or Iran-linked actors; the Kremlin denies involvement.

Other side: The blood service's alarms and fail-safes worked, which shows layered defences limit the harm. Spoiled groceries are costly but recoverable; the real test is whether operators patch before attackers aim at something less forgiving.

See the KKReckons explainer →

Summary based on Wired. Read the full story ↗

Tim Harford on what stolen sweets and parking fines say about cheats

• Tim Harford cheers this year's Ig Nobel economics winners, who found upper-class people more likely to take sweets from children, and says the spoof prize shows what makes economists tick.
• He recalls classic cheating studies: a bagel seller's honesty box paid best in small offices with well-liked bosses, and UN diplomats' unpaid parking tickets tracked corruption at home until New York pulled plates and violations fell 95%.
• His darkest example is a 2012 paper on signing forms honestly, retracted after data sleuths flagged manipulated data; both researchers who supervised the questioned studies deny wrongdoing.

Other side: Laughing at cheats can hide how rare outright fraud is; most research is honest, and the parking link is a correlation, not proof that any one diplomat or country is dishonest.

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

OpenAI and Anthropic say AI disproved the exponential time hypothesis

• OpenAI and Anthropic each said this week that their models disproved the exponential time hypothesis, a 27-year-old assumption that some hard problems take exponentially more effort as they grow. Both posted the proofs on GitHub, in plain language and in the machine-checkable Lean format.
• Independent researchers have not verified them yet. OpenAI has asked the UC San Diego scientists who proposed the hypothesis to rally volunteers and has promised to fund a conference.
• Reactions are split. Linköping's Peter Jonsson says much of his upcoming conference paper no longer works, while Copenhagen's Thore Husfeldt calls it a once-in-a-generation result but says the steady run of AI results has left colleagues demoralised.

Other side: Nothing is settled until people check the proofs. Machine checking confirms the logic, but humans still have to confirm the formal statement says what the hypothesis says. And after earlier rushed AI paper dumps with errors, some academics suspect the labs time big claims to impress investors.

See the KKReckons explainer →

Summary based on Wired. Read the full story ↗

AI could plug hospitals' money leaks, if boards measure it

• Rajendra Pratap Gupta argues AI must become a balance-sheet strategy for hospitals, starting with claims and billing. IRDAI data for FY2024-25 show Rs 1,32,488 crore in health claims, of which Rs 94,248 crore was paid.
• Evidence is mixed: in a UCLA trial of 238 physicians, one AI scribe cut note-writing time by 9.5% while another showed no significant benefit. Bigger gains may come from predicting discharges and using beds and theatres better.
• Apollo Hospitals reports 20 clinical AI algorithms. Gupta urges boards to set baselines, test three use cases and tie vendor pay to independently measured results.

Other side: Expectations are running ahead of evidence: forecast savings are not realised savings, and integration, cybersecurity and model upkeep eat into returns. If insurers deploy AI too, claims could become a costly bot-versus-bot contest.

See the KKReckons explainer →

Summary based on The Economic Times. Read the full story ↗

Biomedical engineering is booming, but can health care keep up?

• A Nature editorial says biomedical-engineering papers in the Nature Index grew 70% from 2021 to 2025, with China now well ahead of the US. Its worry: whether the surge reaches patients.
• A Nature feature shows the strain. Smartwatches worn by hundreds of millions pour heart, sleep and stress data on doctors; one patient checked her watch's ECG 916 times in a year. Trials test watch alerts to take blood thinners only when needed, but false alarms put families off seizure alerts.
• Nature also profiles researchers such as Xinyu Dong, who makes tough hydrogel e-skin, and James Eills, who sharpens MRI signals.

Other side: A data flood is also an early-warning system: some patients already get diagnosed from watch alerts, and AI triage could sort the noise. Past lab advances such as hydrogels and MRI did become routine care; this wave may simply need time.

See the KKReckons explainer →

Summary based on Nature. Read the full story ↗

Anthropic says Claude drafted 35,000 antibody designs in a month

• Anthropic says it spent months working with six drug developers, including Bristol Myers Squibb and Generate Biomedicines. For Divergence Labs, Claude produced about 35,000 antibody designs for a food-allergy drug in roughly a month; in early tests, three of the 20 that dislodged IgE did better than the standard medicine.
• At Enveda, Claude checked 1,061 pages of manufacturing paperwork, caught all 11 issues human reviewers found plus seven more, and cut review time by 89%. BigHat's best antibody variants bound more than 10 times more tightly.
• In Anthropic's own muscular dystrophy programme, a candidate went from idea to promising mouse results in under 18 months; human trials could start in late 2028 if results hold.

Other side: Almost all of this is preclinical and reported by the vendor and its partners. Most drug candidates fail in human trials, so speed at the start says little yet about cures.

See the KKReckons explainer →

Summary based on Anthropic. Read the full story ↗

Weight-loss drugs reshape spending as 1 in 10 Americans take GLP-1s

• An FT Big Read finds people on GLP-1 drugs, now about one in 10 Americans, are changing what they eat, buy and do. Gyms are giving more floor space to strength training, and supplement and skincare makers are gaining customers.
• Snacks, fast food and plus-size clothing are losing out: Macy's calls its plus-size business very soft, and brands are pushing smaller, pricier, protein-rich portions.
• More than 200 listed companies outside healthcare discussed the drugs on earnings calls last quarter. Pills and cheaper generics after patents lapse in the early 2030s could widen use.

Other side: Drinking and nights out have barely dipped, so the hit is narrower than feared. Some users regain weight or stop, and so far the clearest effects come from early, wealthier adopters, which may not hold as use widens.

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

Silicon Valley parents push schools to cut back on classroom tech

• Bay Area tech workers are paying for screen-light schooling, the Guardian reports: San Francisco Waldorf, two miles from Anthropic's headquarters, charges $40,000 a year for kindergarten, and many new families work in tech.
• Parents are pushing public districts too. San Mateo has stopped sending Chromebooks home with pupils up to second grade and bans high schoolers' phones; Mountain View is moving away from the i-Ready program and Palo Alto bans smart glasses.
• Experts warn poorer districts may keep the screens richer ones drop, while school-choice vouchers move public money to private schools.

Other side: Good tools can personalise practice and fill gaps where teachers are scarce, and the parents' main ask is evidence and vetting, not a blanket ban. A few wealthy districts may not speak for families who rely on school devices.

See the KKReckons explainer →

Summary based on The Guardian. Read the full story ↗

'Scentronic' prototype senses a smell and recreates it far away

• Ohio State researchers built an e-nose that spots a smell, then tells a small remote device to release a matching scent, at any distance.
• It knows five scents so far: beer, cake, coffee, oranges and red wine. Scent molecules do not travel online; only the odour data does.
• It is a research prototype built without a company. Mixed smells, refills, cost and safety of the mist are open questions.

Other side: Recognising a smell and recreating it are separate challenges, the lead author says. Five scents is a demo, not a library.

See the KKReckons explainer →

Summary based on Reuters. Read the full story ↗

Microsoft's AI beats doctors on hard cases; AI phage given to patient

• Reuters reports that a Microsoft AI tool that works up patients step by step, choosing questions and tests as it goes, was more than twice as accurate as general physicians on over 300 complex cases in a study in Science, and its work-ups cost less.
• The Independent reports that California lab Radical Numerics says it gave phages designed by its Omnii-1 model to a patient with an antibiotic-resistant infection under compassionate use; outcomes are undisclosed and await peer review.
• Microsoft will next test emergency-room triage with Beth Israel Deaconess in Boston.

Other side: Benchmark cases are not hospital wards, and the doctors in the study worked with textbooks and the internet rather than colleagues. The phage claim has no published outcome. Trials and regulators will decide.

See the KKReckons explainer →

Summary based on Reuters and The Independent. Read the full story ↗

Moderna-Merck mRNA vaccine cuts melanoma relapse risk by 42%

• A personalised mRNA vaccine from Moderna and Merck, intismeran, added to Keytruda after surgery cut the risk of high-risk melanoma returning by 42% at two years, in a trial of more than 1,100 patients.
• Each patient's vaccine is built from mutations unique to their tumour and given as nine doses three weeks apart, training the immune system to spot returning cancer cells.
• Similar personalised vaccines are being tested against lung, kidney and pancreatic cancers; experts not involved said the result bolsters the field.

Other side: Cutting relapses is not yet proof that patients live longer, and a vaccine made for every patient may be costly and slow to produce at scale.

See the KKReckons explainer →

Summary based on AP. Read the full story ↗

Terence Tao calls for ‘Maths 2.0’ as AI floods the field with proofs

• In a Nature World View column, Fields medallist Terence Tao says AI-aided proofs have appeared for hundreds of long-standing problems in recent months, and calls for a ‘Maths 2.0’ no longer defined by scarce proofs.
• He flags costs: OpenAI’s 6 October release of hundreds of solutions disrupted research programmes; many AI results miss publication standards; credit for prior work is unclear; juniors lose a route to recognition.
• He warns understanding lags output, ‘proof indigestion’, and sees similar risks in opaque models used for bank lending.

Other side: Tao uses AI in his own research and says that, used creatively and with the right incentives, it could deepen understanding. The upheaval may be a transition cost, not a lasting loss.

See the KKReckons explainer →

Summary based on Nature. Read the full story ↗

Warhound raises $200mn to clone and gene-edit US military dogs

• The FT reports Warhound, conceived by Anduril co-founder Palmer Luckey and Colossal Biosciences’ Ben Lamm, has raised $200mn from backers including Sequoia, Ark Invest and the CIA-founded In-Q-Tel to rebuild US supply of military working dogs.
• It plans to clone top dogs and gene-edit them for longevity and performance; Anduril will develop body armour and ‘cybernetics’. Most such dogs now come from European breeders.
• Luckey holds no equity and no daily role; ex-Pentagon data chief David Spirk will run it.

Other side: Lamm argues editing is more humane, since many working dogs fail training today. But it is a promise, not a result: edited dogs still have to prove they outperform well-bred ones.

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

OpenEvidence: the $15bn AI start-up reshaping how US doctors treat

• OpenEvidence, founded by Daniel Nadler, runs an ad-funded chatbot that searches peer-reviewed journals and clinical guidelines to help doctors diagnose and treat patients.
• Nadler says nearly two-thirds of US physicians use it. It raised money in September at a $15bn valuation, and hospitals including NYU Langone and Mount Sinai have built it into their record systems.
• Evidence on accuracy is mixed: a Mayo Clinic study (Mayo is an investor) found accurate advice, while a June Nature paper found it trailed general models such as Gemini and GPT on clinical benchmarks.

Other side: Doctors adopted it on their own, which suggests it saves real time; the worry is juniors leaning on it before mastering the basics, and OpenAI and Anthropic are moving in.

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

A record El Niño puts AI weather forecasting to the test

• This year's El Niño is already the strongest on record and is expected to peak in December. The World Food Programme projects 274mn more people facing acute food insecurity by the end of 2027.
• AI makes forecasting faster and cheaper. UN agencies and the Red Cross use it to plan, a non-profit texts AI-based advice to farmers in Kenya, and hedge funds used it to bet early on sugar.
• Sceptics say the models learn from a cooler past. In a Met Office study, AI missed an extreme European winter that a physics-based model caught.

Watch next: How AI forecasts score this winter. Models trained on the past can miss the extremes a warmer world produces, and a confident wrong forecast can do more harm than none.

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

A Holmes documentary asks whether she even knows she lied

• In "You Can See Everything", Nathan Fielder and Lance Oppenheim film Elizabeth Holmes at home with her partner and two young children in her last weeks before prison. Asked if she did anything wrong, she says no.
• Critics split on what it shows. The Post's Monica Hesse sees either a damning portrait or a near-exoneration, and thinks Holmes is guilty; on NPR, Ronald Young Jr. says it turns on whether she understands what lying is.
• The backstory: in 2022 a jury convicted her on four counts of defrauding investors over a finger-prick blood test that didn't work as promised.

Other side: Giving a convicted fraudster this much screen time risks softening her story. A jury has already weighed her claim that she didn't know, and NPR's critic says he left understanding Fielder better than Holmes.

See the KKReckons explainer →

Summary based on The Washington Post. Read the full story ↗

Text that old friend: they'll value it more than you think

• A 2022 study by Peggy Liu and colleagues found that people who messaged a friend they had not contacted recently underestimated how much it was appreciated, more so when the contact came as a surprise.
• A related 'liking gap' study from 2018 found people judge their conversation partners' enjoyment less positively than those partners report.
• Lara Aknin and Gillian Sandstrom found reluctance grows when old friends feel like strangers. In one experiment, 53% of people who first messaged current friends then contacted an old friend, against 31% of those who browsed social media.

Other side: The studies cover everyday attempts to reconnect, not every lapsed friendship, and the warm-up helped some people, not all. Some silences have reasons.

See the KKReckons explainer →

Summary based on The Economic Times. Read the full story ↗

Trials suggest many operations work little better than placebo

• A wave of randomised trials is testing common operations. Antibiotics work as well as appendix removal for most patients, and spinal fusions and rotator-cuff repairs match placebo or physiotherapy.
• A 2022 analysis of 100 trials found that about two-thirds of the improvement patients feel after surgery comes from natural healing or the placebo effect of being cared for.
• Practice changes slowly: faster where surgeons are salaried, as in Britain, than where they are paid per operation, as in America.

Other side: Operations such as transplants and emergency Caesareans clearly save lives, and most procedures have not yet been tested against sham surgery at all. Missing evidence cuts both ways.

See the KKReckons explainer →

Summary based on The Economist. Read the full story ↗

AI agents find two room-temperature magnet candidates for future memory

• A team of AI agents designed a new compound and spotted a 1999 one as magnets with zero net magnetism that still sort electrons by spin.
• The 1999 compound, KV[Cr(CN)₆], stayed magnetically ordered up to 376 K (103 °C); the new design may be hard to make in its useful form.
• Band gaps and spin sorting are predicted by simulation, not yet measured; the code and calculations are public.

Other side: Spin sorting is predicted, not yet measured; the real test is remaking the material in a lab.

See the KKReckons explainer →

Summary based on Vals AI. Read the full story ↗

Wall Street’s $10bn India hospital bet sparks a fight over bills

• Blackstone, KKR, TPG, General Atlantic and others have put about $10 billion into Indian hospital chains in five years, per EY. PE-backed chains hold under 5% of beds but are strong in high-margin cardiac and cancer care.
• In August a parliamentary panel warned of an “unchecked influx of foreign capital” and suggested revisiting foreign-investment rules, examining price caps and creating a hospital regulator.
• Insurers say hospitals inflate bills and push costly procedures; hospitals blame late payments and low reimbursement. One Mumbai patient paid out of pocket when her insurer refused the robotic part of her surgery.

Other side: EQT’s Jean Eric Salata warns price regulation would deter investors, and Wharton’s Lawton Burns says it is unclear regulation can contain costs. Would caps mean fewer new hospitals?

See the KKReckons explainer →

Summary based on NDTV. Read the full story ↗

AI can now prescribe acne drugs in Utah

• From Monday, Nolla Health’s app can send acne prescriptions straight to pharmacies in Utah, the first time a state has let AI stand in for the doctor, under a ‘narrow exception’ from Utah’s Office of Artificial Intelligence Policy.
• Patients scan their face and enter their medical history; the AI scores acne severity and picks from eight approved topical treatments, with no pills or hormone therapies.
• Oversight tapers: a physician checks the first 100 patients before anything is sent, the next 400 weekly after the fact, and then at least 10% of prescriptions each month. The pilot runs a year, renewable for two more.

Other side: Scripts still carry a clinician’s name and doctors review a share of them. The AMA’s John Whyte doubts AI can meet the standard humans are held to. A safe shortcut, or a precedent for riskier drugs?

See the KKReckons explainer →

Summary based on Bloomberg. Read the full story ↗

NPS Swasthya: a health savings fund plus top-up cover

• The pension regulator PFRDA launched NPS Swasthya on 1 October: a healthcare investment account bundled with a mandatory super top-up group health policy.
• Cover can be bought from 18 to 70 and renewed to 85, with no medical tests; a ₹30 lakh cover with a ₹3 lakh deductible costs ₹1,732 a year at 18–40 and ₹4,953 at 61–70.
• Joining needs ₹1,000, a ₹200 yearly administrator fee and the first premium; the corpus can pay the deductible or network outpatient bills, but withdrawals are capped at 25% of contributions.

Other side: Pre-existing diseases wait a year, about 15 serious illnesses may be excluded or cost extra, and like-for-like price comparisons are hard. The real test is how claims are handled.

See the KKReckons explainer →

Summary based on Mint. Read the full story ↗

A rare Nanotyrannus heads to auction as the fossil boom roars on

• Sotheby’s will sell ‘Jodi’, a Nanotyrannus found in Montana in 2019, in New York at the end of October, estimating $6–9 million; it is one of three specimens known to science.
• It follows a record $50.1 million for the T. rex ‘Gus’ this year, after a 10-minute bidding battle, and Ken Griffin’s $44.6 million Stegosaurus in 2024.
• A paper last year by Lindsay Zanno and James Napoli settled an 80-year debate: Nanotyrannus was its own genus, not a young T. rex. Jodi is the first to sell since.

Other side: Griffin’s Stegosaurus is on museum display, but where ‘Gus’ went isn’t public. Should rare finds that settle science debates go to private buyers?

See the KKReckons explainer →

Summary based on Financial Times. Read the full story ↗

Suspected plague death at a Siberian lab

• A young technician at the Irkutsk Anti-Plague Institute in Siberia died after severe pneumonia; Russia’s health watchdog says the infection’s origin is ‘unknown’.
• Nearly 200 people are under medical observation, several regional hospitals are quarantined, and a criminal probe into safety violations has reportedly been opened.
• Accounts differ: some reports say she broke a tube of live plague bacteria, while officials deny any lab accident and call the situation stable.

Other side: Plague is easily treated with antibiotics, and officials say the situation is stable. A contained accident, or an outbreak?

See the KKReckons explainer →

Summary based on CNBC. Read the full story ↗

Every Cure: using AI to find new uses for drugs already on the shelf

• A nonprofit co-founded by Dr David Fajgenbaum, in remission for over 11 years thanks to a repurposed drug.
• Its AI analyses the world’s knowledge of drugs and diseases to rank the most promising repurposing ideas.
• Anyone can suggest repurposing ideas, volunteer as an expert, or donate.

Other side: Old, cheap drugs earn little for anyone. Who pays for the costly trials needed to prove each new use?

See the KKReckons explainer →

Summary based on Every Cure. Read the full story ↗