UC Irvine and JPL glaciologists project up to 30 mm of Thwaites sea-level rise by 2100, twice earlier estimates, even with no added warming. Replacing the grounding line with a grounding zone made the model's reconstruction of past change four times more pronounced.
Perspective Coverage
4 publishers
- Builder
- Builder 32%
- Operator
- Operator 49%
- Investor
- Investor 19%
Reality
- Evidence66
- Adoption
- Not scored
- Hype gap+25
- Incentives35
- Confidence64
Central Pacific's Niño 3.4 region passes 30C for the first time as El Niño strengthens
Sea surface temperatures in the Pacific's Niño 3.4 region passed 30 degrees Celsius for the first time, preliminary data released Oct. 5 show. On the anomaly measures that remove global ocean warming, the event is already close to the strongest El Niño on record, Timothy Osborn of the University of East Anglia told Live Science.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+15
- Incentives
- Not scored
- Confidence50
Atoco pitches its waste-heat water harvester to AI data centers on the strength of a parking-lot demo
Atoco says its Nobel-backed MOF harvester pulls water from air with no electricity, using heat as low as 100F. That makes waste heat from AI data centers a candidate water supply, though operators still lack an output figure for the unit.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+45
- Incentives70
- Confidence35
Platinum electrode lenses molded 10 isolated rabbit corneas to a target focus in about a minute
Chemist Michael Hill and surgeon Brian Wong reshaped 10 isolated rabbit corneas to a target focus using a platinum lens and a small current. The method removes no tissue, and so far the evidence is a conference talk on eyes taken out of the animals.
Reality
- Evidence28
- Adoption
- Not scored
- Hype gap+35
- Incentives
- Not scored
- Confidence40
Aleks data shows ChatGPT-era students 25% less likely to solve word problems without AI
UC Irvine researchers mining a decade of Aleks math data found students 25% less likely to solve word problems in proctored exams after ChatGPT arrived. For teams using AI assistants, it suggests output figures can improve while unaided skill slips.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+30
- Incentives
- Not scored
- Confidence40
Managers run more of their writing past AI than the employees they supervise
Omni Calculator found 54% of managers run at least 70% of their writing past AI, against 30% of experienced individual contributors. The studies tying that habit to weaker thinking did not study working managers. The case for retraining leaders is still an inference.
Reality
- Evidence45
- Adoption55
- Hype gap+35
- Incentives
- Not scored
- Confidence40
Waymo and Zoox test drivers absorbed dozens of injuries while training the robotaxis
Autonomous vehicle safety cases rest on crash comparisons with human drivers, but exclusive OSHA data reported by TechCrunch carries a second injury count: the sprains and whiplash logged by the test drivers training the cars.
Reality
- Evidence46
- Adoption58
- Hype gap+24
- Incentives62
- Confidence48
Second generation of mice showed reduced fertility after mothers spent 42 days on the ISS
Twenty female mice spent 42 days on the International Space Station, conceived after landing, and produced two further generations with fewer litters, smaller litters and lower ovarian hormone levels. Their own litters were only slightly small.
Reality
- Evidence62
- Adoption
- Not scored
- Hype gap+22
- Incentives34
- Confidence57
A Bayesian PDE inference framework swaps adjoint solves for LoRA fine-tuning
Uncertainty quantification on PDE inverse problems normally costs either an adjoint solve or a large offline training set. A Nature Communications paper reports a sampler whose gradients come from a small local fine-tune instead.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives25
- Confidence55
Single-cell maps date a microglial handover in the hippocampus to ages 50 through 75
A Science paper reads gene regulation cell by cell in the human hippocampus and finds embryonic microglia giving way to cells with blood-like inflammatory signatures across a 25-year band of adult life.
Reality
- Evidence58
- Adoption
- Not scored
- Hype gap+22
- Incentives62
- Confidence45
buildOne report1 publisher Batching routine security fixes recovers the time developers spend getting back in
The 23-minute recovery figure everyone quotes was measured on general office workers, not developers, and the coding-specific study puts a full context rebuild nearer 45 minutes. That is the number a batched remediation campaign has to beat.
Reality
- Evidence56
- Adoption
- Not scored
- Hype gap+12
- Incentives62
- Confidence44
Housing and commute type sort how Californians talk about heat online
A UC Irvine team labelled 24,598 California posts with a language model and matched them to county housing and transport data, finding that counties heavy in RV, van and boat households had 22 per cent lower odds of talking about coping.
Reality
- Evidence38
- Adoption12
- Hype gap+28
- Incentives66
- Confidence44
The 150,000-to-1 gap: scaling is a workaround for a learning mechanism nobody has found
Llama 3.1 was pretrained on 15 trillion tokens. A preteen manages on about 100 million words. Nobody knows how the child does it, and easily available web data may run out in the 2030s.
Reality
- Evidence54
- Adoption20
- Hype gap+12
- Incentives32
- Confidence52
A 400-year-old shark eye with no dying cells is a finding, not yet a model organism
UC Irvine reports no retinal cell death in Greenland shark eyes, one of them about 200 years old, with DNA repair offered as a hypothesis and fixed tissue as the only material.
Reality
- Evidence34
- Adoption
- Not scored
- Hype gap+38
- Incentives68
- Confidence41
UC Irvine's morphogen-guided neocortical organoids hold anteroposterior identity across 200,000 profiled cells. In fragile X lines, one regional marker difference disappears.
Reality
- Evidence60
- Adoption
- Not scored
- Hype gap+8
- Incentives62
- Confidence52
Two social features, opposite results: a diet app trial argues against "add engagement"
A four-month study run with a Pittsburgh nutrition app found dietitian feedback on meal photos kept users logging, while peer meal feeds pushed less healthy eaters away. Neither changed diet quality.
Reality
- Evidence48
- Adoption
- Not scored
- Hype gap+12
- Incentives58
- Confidence44