SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " experiment " 926 found 🔀 AI-powered Shuffle
At ORNL, Science Never Sleeps
At Oak Ridge National Laboratory, scientists are building the next generation of research: self-driving labs powered by AI, robotics and high-performance computing. Autonomous systems run continuous “design, build, test, learn” cycles to analyze data in real time and guide experiments.
Global Health & WASH: May 2026 Funding Opportunities (14 new opportunities!)
The May update for Global Health & WASH brings new calls that cluster around three distinct shifts: LMIC research leadership moving from funder encouragement into a hard eligibility constraint, pharmaceutical and corporate-linked funders running structurally independent education and use-inspired research portfolios at substantial scale, and AI adoption shifting decisively from speculative tool to operational capability across health and WASH systems alike. In LMIC Research Leadership as Eligibility, Not Encouragement, Wellcome anchors the dominant signal of the month with three separate major calls that all require or strongly center LMIC leadership in their basic eligibility architecture. Wellcome’s ESIC Hubs (£1.5-£1.9M per hub over 3-5 years) explicitly requires an LMIC-based lead applicant, treating Global South leadership as a design constraint rather than a participation goal. Wellcome’s Infectious Disease Clinical Trial Development Award funds transdisciplinary teams whose lead must be a mid-career or established researcher based at an organization in Africa, South Asia, or Southeast Asia. Wellcome’s larger Infectious Disease Clinical Trial Award (£1M-£8M for optimizing licensed interventions) requires the administering organization to be in eligible LMIC regions and at least 50% of applicants based there. Around Wellcome, the same pattern shows up at smaller scale and across funders: CHINNOVA channels $1M into West and Central African research institutions for climate-health work; IBRO’s Neuroscience Training Grants explicitly tier ceilings by region of residence with Africa receiving the highest amount ($5,000); IHME’s GBD Emerging Researcher Award reserves at least one of its two annual awards for an LMIC researcher; Sidaction’s HIV Cure call funds research teams across France, the Netherlands, and eligible African countries with a minimum two-country collaboration requirement. Taken together, these calls signal a meaningful shift in how research equity is being operationalized — not as encouragement language at the bottom of a call, but as a structural feature of who can lead, where the work must be administered, and what proportion of the team must be regionally rooted. In Pharma-Funded Independent Medical Education and Use-Inspired Research, the lineup this month is unusually deep. Pfizer alone runs four parallel calls under independent education and quality improvement frameworks: a Migraine and Women’s Health RFP (200K x 2-year, $500K pool), Migraine IME ($200K, $1M pool), JAK Inhibitor evidence-based education ($100K), and Maternal Vaccination HCP Education in Saudi Arabia ($75K), plus a separate Pediatric Pneumococcal Surveillance research call in Saudi Arabia ($400K per project). Novo Nordisk Foundation runs four parallel calls of its own at substantial scale: Infectious Diseases Catalyst Grants (DKK 60M pool, up to DKK 7M collaborative), Non-Diabetic Endocrinology Collaborative Grants (DKK 53M pool, DKK 5-10M per project), Pioneer Innovator Grant Health (DKK 1.1M), and Distinguished Innovator Grant Health (DKK 6.8M). LEO Foundation deploys DKK 2-4M per project for dermatology research excellence. Pfizer’s framing across calls deliberately separates the funding from product promotion, emphasizing measurable practice-relevant change and structural distance between scientific outputs and commercial activity. Novo Nordisk’s “use-inspired” hard filter pushes researchers toward a credible translation pathway from mechanism to deployable tool. The pattern: industry-aligned funders are running structurally independent portfolios at meaningful $ that fund implementation evidence, not promotion. In AI as Operational Capability, Not Speculation, several calls this month explicitly treat AI adoption as a near-term operational question with real evidence and governance requirements, rather than a speculative trend to be watched. Wellcome’s ESIC Hubs frames AI as “the accelerator” for evidence synthesis, asking hubs to translate technical innovation into adoption-ready workflows that can match real-world policy timelines. UNICEF Venture Fund’s new Climate Tech for Children’s Health call funds for-profit startups deploying AI, machine learning, and blockchain in low-resource environments, with a strict requirement that solutions must already have a working prototype with promising pilot results. The Water Research Foundation has launched two simultaneous AI calls treating reproducibility, cybersecurity guardrails, and human-in-the-loop deployments as core design constraints rather than nice-to-haves: one on GenAI and Agentic AI in water utilities ($200K), one on alternatives to water shutoffs that explicitly screens for implementation economics evidence. Nordic Innovation’s quantum technology call (NOK 4.5M) extends the same logic into life science and healthcare, demanding consortium-stage projects mature enough to test with real users. TEF-Health offers €300K in subsidized testing infrastructure for European healthcare AI and robotics SMEs. The pattern: funders are converting AI pilots into playbooks, requiring transfer evidence and risk controls before AI can move from experimentation to sector-wide operational practice. For LMIC-based researchers in infectious disease or evidence synthesis, the three Wellcome calls together represent the deepest concentration of LMIC-led research capital in recent memory — pair the Infectious Disease Clinical Trial Development Award (£200K, deadline May 19) as a strategic pipeline feeder into the larger Infectious Disease Clinical Trial Award (£1-£8M, deadline June 2), since the development award functions explicitly as a track into the larger trial scheme. For Nordic researchers, the four Novo Nordisk Foundation calls plus LEO Foundation and Nordic Innovation Quantum represent ~$30M in regional research capital this cycle, all with deadlines clustered in May-August. For WASH practitioners, the three Water Research Foundation calls together (each up to $200K) are a cluster worth reading across — one on AI deployment, one on shutoff alternatives, one rewarding applied innovation — and the strongest applicants will articulate how their work bridges utility operations to policy-relevant evidence. For US-based clinical researchers and educators, the four Pfizer calls plus the APF Portfolio’s 80+ programs form an unusually accessible recurring pipeline; APF’s prohibition on indirect costs is a meaningful structural feature for early-career applicants whose institutions might otherwise absorb significant overhead. Snapshot of New Opportunities Total Estimated Funding Pool: $100 Million+ USD The grants are organized into three categories: Open Calls: Current grant and opportunities with a deadline. Grants are listed by closing date. 39 open opportunities- 14 new! Rolling Applications: current grant and opportunities with rolling applications (but it’s still best to submit as early as possible). 14 rolling opportunities- 1 new opportunity! Long term planning: Grants that have closed their current rounds, but are expected to open new windows. 4 long term opportunities! A quick tip for returning readers: if you want to jump straight to the newest additions, use CTRL F to search for “New!” and navigate quickly to the latest funding opportunities This post is for paid subscribers. This helps support the time and effort it takes to curate and organize these opportunities. Subscribe To keep this accessible to everyone who needs it, we’re happy to offer pay what you can rates. You can find more details here. Open Calls: Migraine Competitive Grant Program: Migraine and Women’s Health (Quality Improvement and/or Research RFP), Pfizer.*Closing soon!* Pfizer, in collaboration with the American Headache Society, is seeking independent quality improvement and research projects that close persistent gaps in migraine care for women across the United States. The funding logic centers on measurable, practice-relevant change: proposals should generate actionable evidence, tools, or system improvements that can be shared and used widely, not just within a single site. Priority areas span menstrual migraine and screening in women’s health settings, the role of hormonal fluctuations across the lifespan, sex-specific risk factors, comorbidities and quality-of-life burden, and disparities in outcomes by race, ethnicity, and socioeconomic status. The call is structured as a two-step competitive process with an initial Letter of Intent, with the strongest concepts expected to translate scientific insight into implementable improvements in diagnosis, management, and equity. Geographies: United States. Who can apply: U.S.-based organizations (not individuals), including professional schools, healthcare institutions, professional organizations, and other entities focused on healthcare improvement. Funding amount: Up to USD $200,000 total over 2 years; total pool USD $500,000. Targeted Sectors / SDGs: Health; Focus areas: migraine, women’s health, quality improvement, clinical research, menstrual migraine, health disparities. Deadline: LOI: May 5, 2026; Full proposal (by invitation): August 28, 2026. Learn more and apply here. This RFP rewards “shareable impact”: the most competitive submissions will pair rigorous methods with a clear plan for system uptake and dissemination beyond the originating institution. ReSSARC: Strengthening the Health Sector and Food Security in the Central African Republic (AID 013381/01/0) Call for Proposals, Italian Agency for Development Cooperation (AICS). *Closing soon!* AICS is seeking a limited set of high-capacity humanitarian partners to deliver an integrated package that stabilizes essential health services while reducing acute malnutrition and strengthening food security resilience in priority areas of the Central African Republic. The funder’s logic is explicitly nexus-based: proposals should meet urgent, life-saving needs while reinforcing local systems, coordination mechanisms, and community-level capacity so results persist beyond the emergency window. AICS emphasizes inclusion and protection outcomes, expecting gender-sensitive design and explicit measures for groups facing compounded vulnerability, including women, children under five, internally displaced people, and people with disabilities. The call also signals a localization pathway, encouraging stronger collaboration with national NGOs and balanced partnerships, alongside clear coordination with existing cluster and humanitarian actors to minimize duplication and improve coverage. Geographies: Central African Republic (Bangui, Ombella-Mpoko, Ouham, Ouham-Pende, Lim-Pende, Lobaye). Who can apply: Non-profit organizations registered with AICS, plus eligible non-profits without an office in Italy that have a pre-existing collaboration agreement with an AICS-listed organization. Funding amount: Total pool EUR €1,800,000; max EUR €800,000 (single applicant) or EUR €1,000,000 (ATS). Targeted Sectors / SDGs: Health; Focus areas: malnutrition prevention and treatment, maternal and child health, community health services, food security and resilience, localization partnerships Deadline: May 5, 2026. Learn more and apply here. This call is structured to reward partners who can link frontline service delivery to system durability and localization, not just short-term coverage gains. Contracts for Innovation in drug and alcohol addiction healthcare, Innovate UK. *Closing soon!* Innovate UK, on behalf of the Office for Life Sciences Addiction Healthcare Goals program, is procuring R&D to accelerate innovations that can improve treatment outcomes, strengthen recovery, and reduce harm and deaths linked to drug and alcohol addiction. The competition is positioned as a market-facing readiness push: selected projects are expected to advance solutions toward later-stage validation, generate evidence of user acceptability and UK market fit, and map credible routes through regulatory and certification requirements. Innovate UK signals a preference for innovations that can be field-tested in relevant UK settings and progressed to TRL 6 or 7, with practical plans for commercialization after contract completion. This structure favors applicants that can execute most work in-house, co-develop with service providers and people with lived experience, and translate technical progress into deployable tools for addiction healthcare delivery. Geographies: United Kingdom. Who can apply: Single organizations of any size (including EU, EEA, or international) leading delivery, with most work and key deliverables carried out in the UK; subcontractors for specialist skills only. Funding amount: GBP £200,000–1.5 million per project (inclusive of VAT); Total pool: GBP £20 million (across two strands). Targeted Sectors / SDGs: Health; Focus areas: Mental Health, Health Systems Strengthening, Research & Development, Technology Access. Deadline: May 6, 2026 (11:00am). Learn more and apply here. This call uses procurement to pull near-market addiction innovations toward operational proof and adoption pathways in UK services.
The creepy feeling in old buildings might have a surprising cause
A hidden force may be quietly shaping how you feel—and you’d never even know it. Infrasound, an ultra-low-frequency vibration below the range of human hearing, is everywhere from traffic to old buildings. In a small experiment, people exposed to it became more irritable, less engaged, and even showed higher levels of the stress hormone cortisol—despite having no idea it was present. The findings suggest our bodies can “sense” these vibrations without conscious awareness, potentially explaining eerie sensations in places like basements or supposedly haunted buildings.
Families, experiments collide at UT San Antonio’s free Viva Science SA event
The University of Texas at San Antonio transformed its campus into a hands-on science playground Saturday as Viva Science SA brought experiments and families together.
FDA authorizes early access for experimental pancreatic cancer pill
The FDA said the action recognizes the dire need for a drug to improve survival for pancreatic cancer patients, who face the worst survival odds in oncology.
Google releases experimental ‘COSMO’ AI assistant app on Play Store
Yesterday, Google published “COSMO,” an “experimental AI assistant application for Android devices” on the Play Store.
I'm trying to delay menopause to 60. Here's how I'm doing it
Kayla Barnes-Lentz is using diet, lifestyle changes, and experimental treatments to slow ovarian aging and push menopause as late as possible.
How to Build a Superteam That Keeps Getting Better
In periods of rapid change, the teams that outperform everyone else are not those with the best plans or the most talent but those that learn the fastest. Research across thousands of teams reveals a consistent pattern: High-performing teams—“superteams”—build cultures of continuous improvement. Their leaders encourage experimentation even when things are going well, make curiosity and intellectual humility contagious, surface problems early, stay close to the work, give feedback that supports learning rather than punishing mistakes, and invest in people’s growth even when it doesn’t pay off immediately. When work is tied to shared meaning and progress matters more than perfection, teams become more resilient, adaptable, and capable of sustained success—in business settings and beyond.
New Survey from Harvard Business Review Analytic Services Finds AI Adoption Remains High, Yet Value May Lag Without Modernisation and Workflow Integration
Most organisations have moved beyond experimenting with artificial intelligence, but few are realising its full value. New research from Harvard Business Review Analytic Services, sponsored by Appian, finds that while 59% of organisations (who are moving forward with AI to some extent) have AI in production, the majority are currently focused on incremental gains that prioritise efficiency and productivity over top-line growth.
Scientists just found a chilling way life may have begun
New experiments suggest that freezing and thawing on early Earth may have helped primitive cell-like structures grow and evolve. Tiny lipid bubbles behaved very differently depending on their membrane makeup—some fused into larger compartments and captured DNA more efficiently. These fusion events could have mixed key molecules, setting the stage for more complex chemistry.
Google's clinical director on AI challenges
In this edition of STAT Health Tech newsletter: a conversation with Google's clinical director, an update on Doctronic's Utah experiment, and more.
Search Live gets ready to follow Gemini with a colorful visual update
Experimenting with the latest beta version of the Google app, we were able to see a refreshed Search Live interface.
Scientists catch antimatter “atom” acting like a wave for the first time
Quantum physics once shocked scientists by revealing that particles can behave like waves—and now, that strange behavior has been pushed even further. For the first time, researchers have observed wave-like interference in positronium, an exotic “atom” made of an electron and its antimatter partner, a positron. This breakthrough not only strengthens the weird reality of quantum mechanics but also opens the door to new experiments involving antimatter, including the possibility of testing how gravity affects it—something never directly measured before.
Student hospitalized after suffering burns during science project at MAST Academy in Key Biscayne - WSVN 7News | Miami News, Weather, Sports
KEY BISCAYNE, FLA. (WSVN) - A science experiment gone wrong sent one student to the hospital.According to school officials, a female MAST Academy student sufferedRead More
Alleged would-be Trump assassin once made an "atomic fighting game", which Steam have now yanked from sale
Valve have disabled Steam purchases of Bohrdom, an experimental 2D game made by the developer accused of shooting up a White House press dinner in April 2026.
Scientists just captured a mysterious quantum “dance” inside superconductors
In a breakthrough experiment, scientists directly imaged how particles pair up in a system that mimics superconductors. Instead of behaving independently, the pairs moved in a synchronized, dance-like pattern—something never predicted before. This suggests a major gap in the classic theory of superconductivity.
The long rumored foldable iPad may never see the light of day.
Apple has supposedly been working on a folding tablet since at least 2023. The rumored release has slipped from 2024, to 2026, to 2028, to 2029, and now, according to Mark Gurman, it might not happen at all. While it was a priority for incoming CEO John Ternus, its future is now in doubt: > Foldable iPad: Apple has been working on a roughly 20-inch foldable iPad. This device has been a Ternus priority but may end up being a wacky experiment that doesn’t see the light of day, according to several people who have worked on it.
Coachella Uses Google DeepMind AI to Test the Future of Live Entertainment
Coachella is experimenting with AI-built artist tools, immersive digital worlds, and 3D performance archives that could shape future fan experiences.
Oklou enhances excellent debut album with bold Milwaukee concert
The French experimental pop artist had one of the best albums of 2025 with "Choke Enough." Her Pabst Theater concert made the music more meaningful.
Chevy Humphrey Is Running Experiments at the Griffin Science Museum in Chicago
Chevy Humphrey explains why the scientific method matters in business.
Chevy Humphrey Is Running Experiments at the Griffin Science Museum in Chicago
Chevy Humphrey explains why the scientific method matters in business.
The AI Tools Small Businesses Are Using - Small Business & Entrepreneurship Council
By Karen Kerrigan – Artificial intelligence is no longer experimental for small businesses. According to SBE Council’s 2026 Small Business Tech Use Survey, 82% of small business employers have invested in AI tools and they are rapidly being embedded across daily functions and workflows. In fact, AI has become essential to competitiveness and growth as […]
This exotic particle could finally explain why matter has mass
A major physics experiment has uncovered evidence for a strange new form of matter, where a fleeting particle gets trapped inside a nucleus. This exotic state may reveal how mass is generated, suggesting that particles can weigh less when surrounded by dense nuclear matter. The findings support long-standing theories about how the vacuum of space influences mass.
Curiosity rover detects building blocks of life in first-ever experiment on Mars
The Curiosity rover conducted a first-of-its kind experiment on Mars, detecting organic molecules that have never been observed before on the red planet.
A Los Angeles Woman Was Lost. An Ambitious Mental Health Program Gave Her a Sense of Purpose
The novel experiment in community care, concentrated in Hollywood, was inspired by the mental health system in Trieste, Italy.
“Us versus them”: How political propaganda polarises beliefs without providing any news
Political polarisation is often attributed to misinformation, echo chambers, or biased news. But new experimental evidence suggests that merely making a social conflict salient – without providing any information at all – is enough to sharply increase disagreement on both factual and normative questions. This column discusses how politicians and media can ‘artificially’ amplify societal tensions by strategically choosing which conflicts to highlight, and why this mechanism is harder to counter than standard misinformation.
Please don’t inject yourself with bootleg peptides
Why Americans have gone wild self-experimenting with the hottest thing in wellness: Peptides.
TGL’s Business Case Is Starting to Look a Lot More Real
There comes a point when a sports startup stops feeling like a clever experiment and starts looking like a real business. TGL may be getting there. Reuters reported in late March that TGL franchise values had climbed sharply by the end of the league’s second season, with one minority stake deal ...
New issue: Mirrorlife | BBC Science Focus Magazine
Mirror life could end all life on Earth. That sounds dramatic, but senior biologists have raised the alarm about experiments happening all around the world. It sounds like something in a Doctor Who plot, but no – mirror life is a real possibility. So why do scientists want to create something that’s potentially apocalyptic? Well, it could also have huge benefits (think back to splitting the atom). The first thing to say, however, is that mirror organisms are still a little way away yet – but only decades, not centuries. So far, scientists have only made mirror molecules, one of life’s building blocks. What for? Life works because its building blocks fit together. Imagine drilling a screw into a piece of wood. The screw cuts a hole with a thread that snugly matches its own. If, for some reason, the Doctor showed up and handed you a mirror version of the screw, its thread would be in reverse. It wouldn’t twist into the hole you’d just made as the threads wouldn’t match. That said, the mirror screw would still work; you’d just have to create a mirror hole for it. Mirror molecules, including DNA, work a bit like that. They follow the same principles as the biology around us, but they’re entirely incompatible with traditional biological functions. In medical terms, this means they’re effectively invisible to your body’s systems. Our biology doesn’t have the tools to recognise these molecules and break them apart. And if someone took these mirror molecules and put them together to make a cell and then an organism… you’d have mirror life. But we don't have any way of breaking it down. If it escaped the lab, we’d just have to hope it didn’t have an appetite.
This “quantum” material fooled scientists and revealed something new
A mysterious magnetic material once thought to host an exotic “quantum spin liquid” has turned out to be something entirely different—and possibly just as intriguing. Scientists studying cerium magnesium hexalluminate found it showed the hallmark signs of this elusive quantum state, like a lack of magnetic order and a spread of energy states. But after closer inspection using neutron experiments, they discovered the behavior came from a delicate tug-of-war between two opposing magnetic forces.
Area grade schoolers see money burn (among other experiments) at annual Badin Science Carnival
Hundreds of elementary students from area private schools recently watched, learned and even joined in dozens of science experiments by Badin High School students as part of the school’s annual
Chinese researchers turn desert sand into fertile soil in just 10 months using cyanobacteria
Science News: The researchers at the Shapotou Desert Experimental Research Station in China have developed a revolutionary approach for reversing desertification an.
Exciting experiments with Mad Science of Southern MA & RI
This morning on “The Rhode Show” we welcomed back our friend Jolting Jonathan from Mad Science of Southern Massachusetts and Rhode Island as he shared fun experiments with Brendan Kirby. To learn m…
Scientists find a new critical point in supercooled water
X-ray experiments provide further evidence that water exists in two distinct phases
NASA shuts off another Voyager science instrument to save power. The spacecraft is now 15 billion miles from Earth | BBC Sky at Night Magazine
NASA shuts off Voyager 1's Low-energy Charged Particles experiment, which has been operating since Voyager 1 launched in 1977.
The Hard Truths Behavioral Science Must Face in Conflict Settings - by Britt Titus
Approximately 2 billion people live in conflict-affected areas. Yet most of the theories, experiments, and interventions developed in behavioral science are based on stable contexts.
Disturbing Experiment Bolsters The Case Lobsters Feel Pain After All
We need to do better.
One man’s obsessive quest to weigh the human soul
The 21 grams experiment continues to fascinate us a century later.
Experimental drug doubles one-year survival in pancreatic cancer
A new drug that works by making tumors more susceptible to chemotherapy and the immune system has increased survival in those with advanced pancreatic cancer in a trial.
NSWCPD Applies AI and Machine Learning to Predictive Machinery Health
In a soundproof chamber at the Naval Surface Warfare Center, Philadelphia Division (NSWCPD), a steel-gray air compressor hums as engineers test an experimental machine learning model designed to detect early signs of failure through vibration analysis. The effort, part of NSWCPD’s growing portfolio of artificial intelligence (AI) and machine learning (ML) projects, shows early promise for potential future Navy use.
Station Orbits Higher as Crew Runs New Science Experiments
The International Space Station is orbiting higher today after the Progress 93 resupply ship, docked to the Zvezda service module, fired its engines for just over five minutes Wednesday night. The orbital reboost places the space station at the correct altitude for the upcoming Progress 95 cargo mission scheduled to resupply the Expedition 74 crew at the end of April.
How the ABS experiment has worked so far in the majors: Which teams, players and umpires have fared the best
MLB has seen more than 1,000 ABS challenges since Opening Day
Crew Begins New Space Research and Installs New Science Gear
New science experiments are getting underway and new research hardware is being activated aboard the International Space Station after being delivered aboard the Cygnus XL cargo spacecraft on Monday. The Expedition 74 crew spent Wednesday unpacking Cygnus XL and stowing the new gear and supplies throughout the orbital outpost.
First Look: The Luminous Ceramic IWC Big Pilot Perpetual Calendar Ceralume
Let there be light, with the fully luminous ceramic Big Pilot, bringing IWC’s experimental Ceralume technology into series production.
ICARUS experiment marks major milestone in first neutrino science results
The milestone by ICARUS paves the way to probe sterile‑neutrino oscillations for the full Short-Baseline Neutrino Program at Fermilab. The first results demonstrate the data’s excellent quality and suitability for physics analyses, as well as the maturity of the software tools for event selection, fitting and detector simulation.
Opinion | GLP-1 Experimentation Is Everywhere, and Science Can’t Keep Up
Millions of Americans are experimenting with the drugs. Science can’t keep up.
Opinion | GLP-1 Experimentation Is Everywhere, and Science Can’t Keep Up
Millions of Americans are experimenting with the drugs. Science can’t keep up.
Google rolls out April Android Canary build with experimental features for Pixel devices
Google has just released the April Android Canary build with new experimental features for Pixel 8 and above.
This simple change stops robot swarms from getting stuck
In crowded environments, more robots don’t always mean faster results—in fact, too many can bring everything to a standstill. Harvard researchers discovered a surprising fix: adding a bit of randomness to how robots move can actually prevent gridlock and boost efficiency. By allowing robots to “wiggle” slightly instead of marching in straight lines, they can slip past each other and keep tasks flowing smoothly.
New investigative newsroom launches at ASU's Howard Center for Investigative Journalism
A new investigative newsroom focusing on accountability reporting, innovation and experimentation with AI-driven tools launches today at the Walter Cronkite School of Journalism and Mass Communication as part of the Howard Center for Investigative Journalism.The Beam is a multimedia newsroom that produces in-depth, high-impact reporting on issues that matter to Arizona communities, while building modern tools for the public to engage directly with the decision-makers shaping their lives.
Hospitals offer chatbots to fight off ChatGPT
In this edition of STAT Health Tech: Hospitals offer their own chatbots, Medicare's ACCESS pilot, and an update Utah's 'AI doctor' experiment.
Scientists just recreated a rare cosmic reaction never seen before
A breakthrough experiment has shed new light on one of astrophysics’ biggest mysteries: the origin of rare proton-rich elements. For the first time, scientists directly measured a key reaction that creates selenium-74 using a rare isotope beam. The results sharpen models of how these elements form in supernova explosions, cutting uncertainty in half. But the findings also reveal gaps in current theories, hinting that the story isn’t complete yet.
Inertia moves to commercialize one of the world’s most elaborate science experiments
Inertia has signed three agreements with the Lawrence Livermore National Lab, paving the way for the company to bring its pioneering fusion reactor to market.
Chesterfield high schoolers lead marine science experiment for kids
Students from Lloyd C Bird High School led a hands-on marine science experiment for students at Wells Elementary.
Access granted: CMS greenlights more than 150 participants for chronic care experiment
CMS greenlights more than 150 digital health companies for its ACCESS experiment, a new program for tech-backed chronic care with outcome-aligned payments.
‘Consider it one more sign of the decline in the democratic experiment’
Opinion, comment and editorials of the day
Baker Adventist Christian School reports a successful Science Fair
BAKER CITY — Experiments were done, ideas were heard, and fun was had at the ...Read More
Breaking cover: the future of the business book
Authors are experimenting with new technology and formats to better communicate ideas
Illinois Valley Community College’s SciFest returns April 17 with 100+ hands-on science experiments
Illinois Valley Community College will host its annual SciFest on April 17 at 7 p.m. in the College Gymnasium, featuring more than 100 interactive science demonstrations and experiments.
Apple made iOS users wait for the iPhone Fold only to get it wrong
Is the iPhone Fold destined to join the graveyard of Apple's niche experiments, alongside the iPhone mini and iPhone Plus?
GSK sees blockbuster potential in targeted cancer therapy after promising early data
British drugmaker GSK's experimental targeted cancer drug Mo-rez has blockbuster potential, its head of oncology research has told reporters, after early data showed the drug helped shrink tumors in patients with advanced, hard-to-treat cancers.
Life on Mars? Tiny cells just survived shock waves and toxic soil
Mars may be hostile, but it might not be entirely unlivable. In lab experiments, yeast cells survived simulated Martian shock waves and toxic perchlorate salts—two major environmental threats on the Red Planet. Their secret weapon was forming protective molecular clusters that shield critical cellular functions under stress. Without these defenses, survival plummeted, pointing to a potential universal strategy life could use beyond Earth.
“AI polls” are fake polls
A few weeks after Donald Trump’s second presidential win, I took the train up from London (where I was living at the time) to Oxford to attend a conference on polls and forecasts of the 2024 election. Most of the attendees were pollsters or academics, but I also watched presentations from Aaru and Electric Twin, two companies that do what is interchangeably called synthetic sampling, silicon sampling, or synthetic audiences. Stripped of startup jargon, that means they use large language models (LLMs) to simulate responses to public opinion polls by having AI agents take on the role of survey respondents. I had already heard of Aaru thanks to some articles with eye-catching headlines like “No people, no problem: AI chatbots predict elections better than humans” in the months leading up to Election Day. The founders were making some big, some might even say far-fetched claims, such as: “within two years, we will simulate the entire globe — from the way crops are grown in Ukraine to how that impacts production of oil in Iraq, trade through the strait of Malacca, and elections for the mayor of Baltimore.” When Semafor asked Aaru’s cofounders — Cameron Fink and Ned Koh — about my boss, they said “we respect all those who came before us.” Nate (as he so often does) shared his thoughts on Twitter: Fink and Koh were relatively good-natured about this back-and-forth when we spoke at Oxford. They even offered to mail me one of the t-shirts featuring Nate’s quote they apparently had made. I never took them up on the offer, which I now somewhat regret. These synthetic sampling companies fell off my radar for a while, but they do still exist. In fact, Aaru recently received a $1 billion valuation. Is what they’re doing anywhere close to the most important frontier in AI development? Not by a longshot, especially when Anthropic just developed a model so adept at exploiting software vulnerabilities that it’s only being released to 40 companies. Still, silicon sampling is increasingly finding its way into public polling. Axios reported in March that “a majority of people trust their own doctors and nurses” based on findings from Aaru — without mentioning that the “people” in that sentence were actually LLMs. Around the same time, the Public Sentiment Institute “boosted” their online sample of 373 real survey respondents with 114 AI agents.1 (Spoiler alert: even the co-founder of Electric Twin doesn’t think that’s a particularly defensible approach.) Polling companies like Qualtrics and Ipsos are also developing synthetic data panels. So, what should we make of these … “polls”? Let’s get one thing out of the way: whatever they are, they’re not polls in the way that term is usually defined. Subscribe You can’t replace polls with AI On one hand, using LLMs to essentially make up fake survey respondents sort of sounds like the dumbest idea ever, one that will at best imperfectly replicate real polls while introducing all sorts of biases. On the other hand, with LLMs improving at a remarkable, perhaps even alarming rate, maybe that means I’m a dinosaur at the ripe old age of 24 because I still want to rely on polls that talk to actual people. I’m not going to argue that synthetic samples are completely useless. In fact, as I’ll return to later, there is evidence that some techniques can replicate topline survey results quickly and cheaply. But the marketing from certain companies can be slightly optimistic. “No traditional poll will exist by the time the next general election occurs,” said Fink in 2024. We’re just 206 days away from the midterms, and based on the fact that I still have to collect a bunch of polls every day, I’d say he should have run that prediction by a sample of AI agents before the interview.2 To see why synthetic samples can’t replace polls, here’s a quick primer on how they work. The simplest version of these models involves taking a LLM (like ChatGPT or Claude), giving it a demographic profile (e.g., a white, college-educated woman who lives in Utah and makes $70k a year), and then asking it to respond to a survey question. You repeat that process a few thousand times using different demographic profiles and end up with a sample of synthetic survey responses. The actual models used by private companies are more sophisticated than this, usually because they incorporate more hypothesized demographic characteristics for each agent and provide them with extra information. Aaru, for example, feeds agents a diet of news and information they’d be likely to consume, while Electric Twin incorporates their customers’ proprietary data about the audience they’re trying to replicate. The way Ben Warner, the co-founder of Electric Twin, explained it to me was “we have a large amount of data on […] for instance, 5,000 people. Can we make an accurate prediction of how they would respond to another question?” Still, it should be obvious why synthetic samples can’t replace polls. Polling is fundamentally a data collection process. We might use surveys to make predictions by feeding them into election forecasts, but the main purpose of a poll isn’t prediction, it’s gathering new data about what people think and how they feel. Silicon sampling, on the other hand, produces no new data. It’s simply a model: you input LLM training data, demographic prompts, and a bunch of other information, and it spits out a prediction for what a poll would say. We love models here, but models aren’t polls. That difference is an important philosophical sticking point for most pollsters I talk to. “I think politics should stay away from [synthetic sampling], because we’re trying to […] represent the voice of the people,” said Natalie Jackson, a vice president at GQR Insights. Democratic pollster John Hagner told me: “I think I’m just incredibly skeptical of this idea. I don’t think it’s research. At that point, you’re asking the machine to tell you what you already believe.” Hagner has seen some presentations of early synthetic sampling experiments, but so far, “if it’s being used in a campaign, people are keeping it incredibly quiet.”3 But Eli, I hear you saying, aren’t polls themselves increasingly governed by modeling decisions? Indeed they are: pollsters’ choices on which sampling method to use, how to define their likely voter models, and how to weight their samples can and do lead to dramatic differences in the results they publish. Aaru even referenced these limitations in the methodology statement included with that maternal mortality “poll” — although I’m using the term “methodology statement” loosely here, because it doesn’t really explain how the model works at all. We can ignore the (frankly preposterous) implication that synthetic sampling isn’t subject to a separate set of biases. The important point is that there’s still a meaningful difference between using weighting and other statistical techniques on actual polling data and using a model to predict what a poll would say. The latter is far closer to election forecasts or techniques like MRP — potentially useful models, but not a replacement for polls.4 To be fair, other synthetic sampling companies are perfectly happy with the distinction between polls and models. Warner compared polling and synthetic sampling to different tools in a toolbox. “The mistake I think we make is we think that these new tools should either work in exactly the same way or somehow replace these old tools,” he said. “Rather than thinking of it as, okay, so we’ve always had the hammer, we’ve always had the screwdriver, now we’ve got a saw. But don’t use a saw to try [to] do the job of a hammer.” A quick comment from Nate Eli didn’t ask me for a comment — rather rude of him, don’t you think? But since I’m editing this story, I figured I’d add a few quick thoughts rather than putting words in his mouth. Beyond the frequently misleading marketing, what bothers me about the AI “poll” hype is that as AI tools make statistical inference cheaper and/or better (note that these are not synonyms) that actually increases the comparative value of collecting original data. You might be able to train a model to make a reasonable estimate of what some hard-to-reach poll respondent would say — say, a young Black man who voted for Trump. (Such a person checks a number of boxes for a voter who is usually hard to reach in surveys.) Indeed, this is closely related to what models like the Silver Bulletin forecast already do. They essentially smooth out the kinks in noisy survey data by making inferences based on past voting patterns or national polls or surveys of other states. But you don’t actually know what these voters think unless you’re reaching them directly. If there’s a shift in opinion among this subgroup, you’re not going to detect it. So if I were running a campaign, I’d invest more in going the extra mile to find a representative sample of those voters. And then I’d hire some smart quants — or Claude? — to figure out the implications for campaign strategy based on proprietary data that my competitors didn’t have access to. -Nate Silver Are these models any good? If synthetic surveys are just a new type of model, the next obvious question is whether the models are at least accurate. The answer very much depends on who you ask. On one end of the spectrum, you have the maximalist argument that synthetic sampling is more accurate than actual polls. “It’s an incredibly challenging problem to go to someone and say ‘hey, we’re going to be more accurate at predicting human behavior than you, even when you talk to your customers directly’,” Koh recently told CNBC. In his view, synthetic sampling isn’t a saw to polling’s hammer, it’s “magic.” There’s certainly evidence that synthetic samples can replicate certain survey toplines. But if Aaru does have any examples of their approach outperforming the polls, they’re keeping those to themselves.5 Aaru’s 2024 election model, for example, had Kamala Harris leading in Michigan, Nevada, Pennsylvania, and Wisconsin on November 4th. And although they’ve since taken down their forecast page, they gave Harris a 50.5 percent chance of winning the race on November 2nd.6 After the election, Fink told Semafor he was happy enough with those results because they were “within margin of error,” a term that is completely meaningless when applied to a “sample” of AI agents. And of course, Aaru says their models have improved since 2024, so supposedly now they’d be more accurate than the polls? Still, their stronger argument is on cost: “We are significantly faster and cheaper than traditional polling, and still more accurate,” said Fink. The first two claims are undeniably true, but the third brings us to the opposite end of the spectrum. Both Jackson and Hagner are skeptical that these models are reliable for anything beyond replicating common survey toplines. “I just […] don’t think the machines are what we want when we’re looking for nuanced views. My example on this is people in Arizona and Nevada in 2024 who voted for Trump and voted for expanding abortion in their states on ballot initiatives,” said Jackson. Hagner identified another issue. Maybe the synthetic respondents, like sycophantic LLMs, are inhuman in one important way: they’re too nice. “The reports that have come through at the meetings that I’ve been at are that the early experiments on this, they cannot get respondents to be as racist or sexist or, frankly, as negative as human respondents,” he said. Academic research mostly agrees on this point. While there are some papers that show promising results when using LLMs to replicate polling data, most show that LLMs suffer from various quirks like producing too few “don’t know” responses and can seriously overpredict the favorability of politicians like Donald Trump and Kamala Harris. They also seem to struggle with too little variation between demographic subgroups, so the difference in predicted opinion between Democrats and Republicans, for example, is too small. When I asked Warner about these studies, his response to these papers was that just because academics can’t get synthetic sampling to work doesn’t mean that the technique doesn’t work in general. “Actually, the argument is, okay, yours does not [work]. That does not mean […] for this complex set of machinery, which uses a lot of investment, a lot of time, a lot of money, you can’t get it to work.” Cards on the table, I’m somewhat sympathetic to this argument because academics aren’t exactly great at making election forecasts. Usually, the people with skin in the game are the most accurate. Warner’s argument is that the approach Electric Twin takes — which includes, for example, making multiple predictions for each synthetic respondent using different models and prompts and subsequently averaging those to get a final prediction in a sort of ensemble forecast — produces better results than the simpler academic models. Warner shared a comparison between his method and the method from a recent academic paper with me, and Electric Twin was indeed able to get more accurate replication. But even still, he acknowledged that synthetic sampling “is not a crystal ball.” “If you asked me, do I think using other data sources will be more accurate than asking somebody who they will vote for, I would probably say no. But if you asked me ‘would your system be useful for our turnout modeling today?’ I would say yes.” For better or worse, it looks like the method is already getting more popular in the market research world. Most of the clients Aaru touts these days are businesses like EY and McDonald’s. And AI will probably begin to pop up in other parts of the political polling process. Pollsters are already using it to code open-ended survey responses, and some firms, like YouGov, are testing using LLMs to ask survey respondents questions. More worryingly, one danger to actual polls is that AI agents can be used to infiltrate online surveys. Most online polls use various checks to prevent that from happening, but there’s conflicting evidence on how effective those filters are and how prevalent AI agents currently are in online panels. If those agents ever become impossible to detect, it might spell the end of online polling, but the solution isn’t to replace all of your respondents with ChatGPT. Silver Bulletin is a reader-supported publication. To receive new posts and support our work, consider becoming a subscriber. Subscribe 1 That particular poll obviously doesn’t meet Silver Bulletin standards for aggregation. But we exclude all Public Sentiment Institute polls from our averages because we classify them as an amateur polling firm. 2 You could argue that Fink meant the next presidential election, but (a) I’m also confident we’ll still have real polls in 2028 and (b) in that case he should have asked an LLM to define “general election.” 3 Quick caveat: that’s reporting from a Democratic pollster. It’s possible that Republicans are more willing to use AI in political campaigns. 4 Indeed, Silver Bulletin does not include “polls” produced by MRP in our forecasts or averages, and we think it’s extremely misleading when their practitioners describe them in a way that suggests original data had been collected among a large number of states or Congressional districts. 5 A recent report from Aaru and EY did show two examples of a synthetic estimate being closer than a survey to a real-world benchmark — but I’d take those findings with a grain of salt because the report reads more like an ad and didn’t involve any sort of prediction being made ahead of time. 6 For comparison, our odds for Harris on the same day were 48.2 percent.
Allegri explains Milan changes with new 4-3-3 experiment
Max Allegri explains how his Milan team changes with a 4-3-3 formation against Udinese and why they must ‘give Rafael Leao the right service’ as a centre-forward.It kicks off at San Siro at 17.00 ...
Radical Rob demonstrates fun science experiments on 'Fox & Friends Weekend'
Mad Science NYC & Westchester's chief mad scientist Rob Diaz demonstrates home experiments with the hosts of 'Fox & Friends Weekend.'
Pool at Altru Sports Complex will also house UND diving facility to simulate low gravity
The neutral buoyancy facility will be built under the floor of the pool's deep end and used for experiments, research and training in the space studies program.
Dark Matter Experiment Reaches Critical Milestone Deep Underground
Hundreds of times colder than outer space.
Microsoft finally lets Windows 11 testers unlock experimental features without ViVeTool
The changes are part of the fix Windows 11 effort
Interview: Rob McClure on the Unsettling Reality of SPARE PARTS – “This Isn’t Science Fiction. It’s Just Science.”
“Experimenting on yourself is insane. An experiment on a single person, with no controls, tells you nothing. It is incapable of predicting anything.” Or… is it? Such is the premise of Spare Parts, the darkly comedic and deeply twisted new drama currently shaking up Theatre Row through April 30th — and subverting our standards of what …Read more
Hong Kong animal groups urge HKTVmall boycott over life science experiments
Online retailer facing major backlash after its parent company confirmed tests on pigs’ and sheep’s detached heads and limbs.
Project Indigo expands support to iPads and the iPhone 17e
The new update to Adobe’s experimental camera app adds initial support for iPads with at least 6GB of RAM and the new iPhone 17e.
The unprecedented human experiment of Artemis II is only just beginning
Experiments aboard Orion are capturing insights about astronaut health that were never recorded during the Apollo program.
Integrative experiment design reveals hidden patterns in decades-old social science research
Led by MIT Sloan School of Management PhD graduate Mohammed Alsobay and associate professor Abdullah Almaatouq, the researchers demonstrate that integrative experiment design, an approach that systematically varies multiple experimental conditions within a shared design space, enables researchers to discover how factors combine to determine social and behavioral outcomes.
A tiny experiment using Artemis II astronaut cells could reshape medicine
Chips seeded with the astronauts’ bone marrow cells circled the moon to help probe how deep-space flight affects human biology.
Scientists discover hidden gut trigger behind ALS and dementia
A new study reveals that gut bacteria may play a key role in triggering ALS and frontotemporal dementia. Harmful sugars produced by these microbes can spark immune responses that damage the brain. This breakthrough explains why some genetically at-risk people develop the diseases while others don’t. Even more promising, reducing these sugars improved brain health in experiments, hinting at new treatment possibilities.
Science without a script | CWRU Newsroom | Case Western Reserve University
In a handful of biology labs at Case Western Reserve University, undergraduate students move between benches, comparing results, adjusting experiments...
Monkey business: Small fleet keeps getting violations for handling lab animals
JKL Secure Freight's been getting in trouble with the USDA -- and an animal-rights groups -- over its handling of monkeys for science experiments. A look at the fleet's roadside-inspection history doesn't inspire much confidence either.
Why Do Deliberative Discussions Reduce Affective Polarization? Evidence from a Deliberation Experiment in Honduras - April 7, 2026
Deliberative democracy advocates argue that deliberation can strengthen democracy in part by reducing partisan hostility and affective polarization. Yet - Why Do Deliberative Discussions Reduce Affective Polarization? Evidence from a Deliberation Experiment in Honduras April 7, 2026
ENC at Three celebrates space in today’s “Blast Off Science” experiments
These experiments are sure to get your kids excited about the Artemis II launch and Project Hail Mary heading to theaters.
Kimberly Goes to Business School, part 2: Supporting the work
"Kimberly Goes to Business School" continues, with an experiment with a new scheduling model teaching the limits of AI.
Kids will spring into science with April’s Parents’ Night Out at DHDC
From 5:30 to 9:30 p.m. on Friday, April 10, the Don Harrington Discovery Center is offering a chance for children ages 4-11 years old to crack into hands-on experiments while caregivers enjoy an evening to themselves.
‘Google AI Edge Eloquent’ is an offline, subscription-less voice dictation app
Google today released a new app on iOS called “Google AI Edge Eloquent” that’s rather curious. It's somewhere between an experiment...
5 Easy Spring Science Experiments for Kids to Try at Home
Spring weather has a mind of its own.One day the kids are outside all afternoon. The next day it’s chilly, rainy, and someone is already saying, “…
Scientists may finally detect hidden ripples in spacetime
Scientists have taken a major step toward probing one of physics’ biggest mysteries—how gravity and quantum mechanics fit together—by creating the first unified way to detect tiny “ripples” in spacetime itself. These subtle fluctuations, long predicted but poorly defined, are now organized into clear categories with specific signals that real-world instruments can search for. The breakthrough means powerful tools like LIGO and even small tabletop experiments could start testing competing theories of quantum gravity much sooner than expected.
How the Voyager missions rewrote planetary science, tested the limits of 1970s engineering, and became humanity's longest-running experiment in institutional patience
In November 2026, Voyager 1 will become the first human-made object to reach one light-day from Earth. Nearly 49 years after launch, the twin Voyager missions remain the most productive space exploration program ever flown, and a case study in what happens when engineering resilience meets institutional patience.
NASA shoehorns in human science on Artemis II moon mission
While the primary goal of the Artemis II mission is the ensure the Orion spacecraft is safe for humans, NASA did find time to fit some science on board during the 10-day lunar fly-by.
Train transfer hub linked to reduced healthcare spending
Researchers from Osaka Metropolitan University and Japan System Techniques Co., Ltd. evaluated the 2019 Osaka-Higashi Line station openings using a natural experiment and health insurance claims. While the overall rail-line effect was not significant, Shigino Station, a well-connected transfer hub, showed an estimated cumulative reduction of about JPY 62,500 per medically treated person over four years, highlighting that station connectivity and local context may determinewhether new stations deliver healthcare cost benefits.
Gemma 4 and what makes an open model succeed
Having written a lot of model release blog posts, there’s something much harder about reviewing open models when they drop relative to closed models, especially in 2026. In recent years, there were so few open models, so when Llama 3 was released most people were still doing research on Llama 2 and super happy to get an update. When Qwen 3 was released, the Llama 4 fiasco had just gone down, and a whole research community was emerging to study RL on Qwen 2.5 — it was a no brainer to upgrade. Today, when an open model releases, it’s competing with Qwen 3.5, Kimi K2.5, GLM 5, MiniMax M2.5, GPT-OSS, Arcee Large, Nemotron 3, Olmo 3, and others. The space is populated, but still feels full of hidden opportunity. The potential of open models feels like a dark matter, a potential we know is huge, but few clear recipes and examples for how to unlock it are out there. Agentic AI, OpenClaw, and everything brewing in that space is going to spur mass experimentation in open models to complement the likes of Claude and Codex, not replace them. Especially with open models, the benchmarks at release are an extremely incomplete story. In some ways this is exciting, as new open models have a much higher variance and ability to surprise, but it also points at some structural reasons that make building businesses and great AI experiences around open models harder than the closed alternatives. When a new Claude Opus or GPT drops, spending a few hours with them in my agentic workflows is genuinely a good vibe test. For open models, putting them through this test is a category error. Something else to be said about open models in the era of agents is that they get out of the debate of integration, harnesses, and tools and let us see close to the ground on what exactly is the ability of just a model. Of course, we can’t test some things like search abilities without some tool, but being able to measure exactly the pace of progress of the model alone is a welcome simplification to a systematically opaque AI space. Share The list of factors I’d use to assess a new open-weight model I’m considering investing in includes: Model performance (and size) — how this model performs on benchmarks I care about and how it compares to other models of a similar size. Country of origin — some businesses care deeply about provenance, and if a model was built in China or not. Model license — if a model needs legal approval for use, uptake will be slower at mid-sized and large companies. Tooling at release — many models release with half-broken, or at least substantially slower, implementations in popular software like vLLM, Transformers, SGLANG, etc due to pushing the envelope of architectures or tools. Model fine-tunability — how easy or hard it is to modify the given model to your use-case when you actually try and use it. The core problem is that some of these are immediately available at release, e.g. general performance, license, origin, etc. but others such as tooling take day(s) to week(s) to stabilize, and others are open research questions — with no group systematically monitoring fine-tunability. In the early era of open models, the days of Llama 2 or 3 and Qwen pre v3.5, the architectures were fairly simple and the models tended to work out of the box. Some of this was due to the extremely hard work of the Llama, Qwen, Mistral, etc. developer teams. Some is due to the new models being genuinely harder to work with. When it comes to something like Qwen 3.5 or Nemotron 3, with hybrid models (either gated delta net or mamba layers), the tooling is very rough at release. Things you would expect to “just work” often don’t. I’ve been following this area closely since we released Olmo Hybrid with a similar architecture, and Qwen 3.5 is just starting to work well in the various open-source tools that need to all play nice together for RL research. That’s 1.5 months after the release date! This is just to start really investing more into understanding the behavior of the models. Of course, others started working on these models sooner by investing more engineering resources or relying on partially closed software. The fully open and distributed ecosystem takes a long time to get going on some new models. All of this is lead-in for the most important question for open models — how easy is it to adapt to specific use-cases? This is a different problem for different model sizes. Large MoE open-weight models may be used by entities like Cursor who need complex capabilities in their domain, e.g. Composer 2 trained on Kimi K2.5. Other applications can be built on much smaller models, such as Chroma’s Context-1 model for agentic search, built on GPT-OSS 20B. The question of “which models are fine-tunable” is largely background knowledge known by engineers across the industry. There should be a thriving research area here to support the open ecosystem model. The first step is to understand characteristics of different base and post-trained models to understand what they look like. The second step is to tune pretraining recipes for open models so they’re more flexible. Interconnects AI is a reader-supported publication. Consider becoming a subscriber. Subscribe For The ATOM Project and other Interconnects endeavors, we’ve put in substantial effort to measuring adoption trends in the open ecosystem. Everything takes a long time to unfold after a model is first publicly available — and adaptability is why. What we know for sure now, when Qwen has been going from strength to strength with its releases, is that technical staff across the industry has gotten comfortable working with Qwen models. Countless research methods and datasets were made to work with Qwen. It’ll take patience for any other model family to get to this point — a patience I’m not sure many open model builders have. This takes us to Gemma 4, Google’s latest open models. Gemma 3 was released more than a year ago, in March of 2025, and is a bit underrated. Gemma 4 comes in 4 sizes for now, with a bigger, MoE model of over 100B total parameters rumored but not released yet. The models we have today come in sizes of ~5B dense, 8B dense, 26B total 4B active MoE, and 31B dense. I’m most excited that they’re finally adopting a standard Apache 2.0 open source license. This’ll massively boost adoption. The standard of better licenses for strong open-weight LLMs was set by mostly Chinese open model labs in the last 1-2 years, and now U.S. companies are following suit. I will personally be so happy if the horrible Llama licenses and Gemma terms of service were an ~18-month transient dynamic of the industry being nervous about releasing strong open models. The Gemma 4 scores look very solid, the small models have incredible benchmark scores (especially in general domains like LMArena) and the 31B model rivals the recent Qwen 3.5 27B, which is the leading member of that class. The ~30B size range is an important one, as it’s accessible both to researchers and to enterprises looking to deploy the model in real use-cases. Where the 7B model scale is the default for tinkering and research, a 30B model is the default for seeing if an open model can unlock substantial value in your specific workflow — a good mix of intelligence, low price, tractability for downstream training, etc. This takes us back to the above adoption criteria I mentioned for open models and the bigger question — do I think Gemma 4 will be an overwhelming success? Previous Gemma models have been plagued by tooling issues and poorer performance when being finetuned. Gemma 4’s success is going to be entirely determined by ease of use, to a point where a 5-10% swing on benchmarks wouldn’t matter at all. It’s strong enough, small enough, with the right license, and from the U.S., so many companies are going to slot it in. I’m cautiously optimistic that Gemma 4 is going to work better here. Winds are shifting for open models built in America. We saw GPT-OSS go through a bumpy launch to become an overwhelming success. There’s a collective energy around the likes of Reflection, Arcee, Nemotron, Gemma, Olmo, and peers that show substantial demand for building new stacks around open models. There’s capital to be spent on AI stacks across the economy by those who want more ownership of everything, including the model. After launching The ATOM Project 240 days ago, the conversation is shifting into the next stage. Summer of 2025 was a crisis moment where the U.S. AI scene realized it can’t wait and figure out open models after building AGI. The two markets will capture different areas and proceed in parallel. Now that more companies in the U.S. are releasing strong models, we need to improve the ecosystem so that these models are easy to use, understand, and build value around. It’s the hard work to build another inflection point in these adoption plots I’ve been updating consistently, but that’s the work to be done. Join me in it. More data coming soon! Here’s a sneak peek:
Scientists Just Discovered There’s Actually Something Faster than the Speed of Light
Scientists experimentally confirmed what we’ve known to be true for half a century—darkness can “travel” faster than light.
Scientists Discover Why Flu and COVID Hit Older Adults So Hard
A new study reveals that aging lungs may play a major role in why flu and COVID can become so dangerous for older adults. Researchers found that certain lung cells can trigger an exaggerated immune response, creating clusters of inflammatory cells that end up damaging lung tissue instead of protecting it. In experiments, activating this aging-related signal in young mice caused their lungs to behave like older ones, leading to severe illness.
Chinese satellite with robotic 'octopus arm' passes key refueling test in orbit — making longer-lived space assets more likely
The experimental Hukeda-2 satellite and its highly flexible robotic arm have passed a major refuelling test in low Earth orbit. The demonstration is the latest step toward China significantly expanding the longevity of its spacecraft.
Physicists just solved a strange fusion mystery that stumped experts
Fusion scientists have solved a long-standing mystery inside tokamaks, the donut-shaped machines designed to harness fusion energy. For years, experiments showed that escaping plasma particles hit one side of the exhaust system far more than the other, but simulations couldn’t explain why. Now, researchers have discovered that the rotation of the plasma itself plays a crucial role—working together with sideways particle drift to create the imbalance.
Mad Science: Easter Experiments
Mad Science of Greater Kansas City's Radiant Randy joins Shane and Kerrigan to share the science behind Easter egg dyeing!
Mad Science: Easter Experiments
Mad Science of Greater Kansas City's Radiant Randy joins Shane and Kerrigan to share the science behind Easter egg dyeing!
USMNT experimentation comes up short in loss to Portugal
The USMNT have a lot of work to do.
Hundreds of Times Colder Than Outer Space: Scientists Reach Near-Absolute Zero in Dark Matter Hunt
Deep beneath the Earth’s surface, a major physics experiment has reached a critical milestone, enabling detectors to operate at temperatures near absolute zero.