MIT's IRIS model infers a cell's signaling history from its gene activity
MIT researchers trained a neural network on six signaling pathways in human stem cells, then tested it on mouse embryo cells. Each pathway's gene-activity fingerprint held across cell types, the team reports in Nature Methods.
Reality
- Evidence52
- Adoption
- Not scored
- Hype gap+30
- Incentives55
- Confidence58
MIT students' 2017 tremor wristband wins FDA clearance nine years on
Encora Therapeutics won FDA clearance in February for a tremor wristband that MIT students first built in a 2017 class. The cleared use is adults with essential tremor, and in one trial 78 percent of patients reported benefit after 90 days of home use.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+15
- Incentives65
- Confidence45
Margaret Hamilton, first programmer hired for MIT's Apollo effort, dies at 90
Margaret Hamilton, who headed the Apollo Guidance Computer software team, has died aged 90, Nature's obituary says. It credits her with helping develop reliable-programming techniques but does not say they underpin practice today.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap
- Not scored
- Incentives
- Not scored
- Confidence55
MIT enrolled a fifth fewer graduate students after losing $100 million in federal awards
MIT enrolled nearly 500 fewer graduate students this fall, down 20% from 2024, after federal research awards fell $100 million, its president said. Sally Kornbluth said the school admits cautiously so no Ph.D. student runs out of funding midway.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+20
- Incentives60
- Confidence45
Researchers including a team at MIT report in Nature that a machine-learning layer doubled a gold-standard model's forecast accuracy at two to six weeks. Beyond two weeks, energy, farm and emergency planners have had little reliable forecast to use.
Perspective Coverage
4 publishers
- Builder
- Builder 43%
- Operator
- Operator 42%
- Investor
- Investor 15%
Reality
- Evidence74
- Adoption
- Not scored
- Hype gap+25
- Incentives35
- Confidence72
Researchers at MIT and Microsoft report in Nature that their open-source correction tool doubled an AI weather model's subseasonal accuracy. Its gains on extreme heat and rain at three weeks are the part grid planners would care about.
Reality
- Evidence72
- Adoption12
- Hype gap+25
- Incentives40
- Confidence70
Researchers report lipid nanoparticles that program mouse NK cells into CAR NK cells in vivo, reaching up to about 80% transfection. Regression in two humanized blood-cancer models is the nearest the abstract comes to human cells.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+20
- Incentives
- Not scored
- Confidence58
MIT-led astronomers found a star about 300 light-years away slowly pulling material off a brown dwarf companion, a process that could last billions of years. One detection shows slow stripping happens, and working out how common it is will need a larger sample of such pairs.
Reality
- Evidence66
- Adoption
- Not scored
- Hype gap+12
- Incentives30
- Confidence64
Microwave cavities take the boson role in an MIT toolkit for hybrid quantum simulation
An MIT team led by Eleanor Crane calculates that cavity-based hybrid quantum hardware could tolerate 1,000 times the error rate of qubit-only machines. The toolkit, published in Physical Review X, was checked on a simplified model only; the proposed experiment is the next step.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+15
- Incentives
- Not scored
- Confidence55
buildOne report1 publisher Self-edited context lifts Qwen3.5-9B from 28.8% to 42.5% on BrowseComp-Plus
Meta, MIT and University of Washington researchers let models rewrite their own context, lifting a trained 9B model from 28.8% to 42.5% on BrowseComp-Plus. The gains come from specified benchmark tasks, and rewriting earlier context invalidates the model cache, so we would keep hand-built compaction pipelines as the production default.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives
- Not scored
- Confidence40
David Autor's trial of 133 patent lawyers found AI improved everyone's drafts, but only seniors kept a skill gain once it was gone. Firms that expected the tool to train junior associates still have to pay for that training, on the evidence of a Google-run working paper.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+30
- Incentives70
- Confidence55
Russia ended its Irkutsk plague quarantine without explaining the death that started it
Rospotrebnadzor ended its quarantine on 7 October and freed most of the nearly 200 people held over a lab worker's death, after tests found no dangerous pathogen. What killed 28-year-old Daria Shipilova, and why the response far exceeded plague protocol, remains unexplained.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap−8
- Incentives35
- Confidence38
Margaret Hamilton, who led the software engineering team at MIT's Instrumentation Lab during NASA's Apollo program, died on September 30 at 90. MIT News, which reported her death, credits her with helping to establish software engineering as a discipline of its own.
Perspective Coverage
9 publishers
- Builder
- Builder 55%
- Operator
- Operator 32%
- Investor
- Investor 13%
Reality
- Evidence80
- Adoption
- Not scored
- Hype gap+15
- Incentives20
- Confidence78
Labor Department investigators have subpoenaed nine universities, Harvard and MIT among them, over J-1 visas held by foreign postdocs. The White House did not immediately source Vance's case, a 61% J-1 rate at the schools against a 38% national average.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+40
- Incentives60
- Confidence55
Small star 300 light-years away is slowly eating a brown dwarf on an 87-minute orbit
Astronomers have watched a low-mass star strip a brown dwarf for the first time, in ZTF J0440+2325 about 300 light-years away, space.com reported. The pace is slow enough for the feeding to last up to billions of years.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+20
- Incentives
- Not scored
- Confidence45
Rumoured OpenAI proof pushed MIT theorists to post a 95-page paper in three days
MIT's Dor Minzer and two students posted a 95-page paper three days after rumours that OpenAI had proved the unique games conjecture. An expected AI announcement set the publication date for one group's finished work.
Reality
- Evidence60
- Adoption
- Not scored
- Hype gap+5
- Incentives
- Not scored
- Confidence55
MIT's SANDO planner guarantees collision-free drone paths as long as obstacle speeds have a known ceiling
MIT's SANDO planner guarantees collision-free drone paths with no map, provided it knows the top speed any moving obstacle could reach. That ceiling is the number an operator must pick and defend before trusting the proof on a delivery route or a disaster site.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+30
- Incentives
- Not scored
- Confidence45
buildOne report1 publisher Readyset widens its cache by rewriting more SQL into shapes its dataflow engine maintains
Readyset's compiler now rewrites subqueries in HAVING, ORDER BY and inner and outer join conditions so its dataflow cache can keep them current. Queries it cannot safely rewrite go to the database uncached, so SQL coverage sets how much read load it takes.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+5
- Incentives60
- Confidence55
Babies aged 4 to 6 months link saliva sharing to close bonds in UCSB-led looking-time tests
Babies aged 4 to 6 months looked longer at a saliva-sharing partner when the other showed distress, in two UC Santa Barbara-led experiments. The finding reaches babies who have had little time to watch anyone kiss or share food.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives
- Not scored
- Confidence40
MIT launches an institute-wide push into K-12 and community-college STEM teaching
MIT launched MIT for America, an institute-wide STEM push from kindergarten to community college, noting only 6 percent of US students take computer science. The plan makes national STEM teaching part of MIT's stated mission and builds partly on programs the Institute already runs.
Reality
- Evidence35
- Adoption20
- Hype gap+30
- Incentives75
- Confidence40