Elizabeth Gibney

Why have physicists have trapped a lone positron (an antimatter electron) in a magnetic trap for months?🧪 To probe its magnetic moment, an inherent magnetic property, & hunt for differences with electrons that could help to explain... well why anything is here at all www.nature.com/articles/d41...

nature.comnature.com

nature.comWorld’s timekeepers upend how they adjust time to avoid catastrophic ‘negative second’Governments have voted to abolish the leap second immediately and replace it with a leap hour.

There will be metrologists partying in Paris tonight 🕺🧪 I've been following this for more than a decade and know how hard they've campaigned to make it happen. Sad day for me though, as no more reporting on the leap second. So long my pesky friend!

Richard Van Noorden@richvn.bsky.social

Governments have voted to abolish the leap second immediately and replace it with a leap hour. by @lizziegibney.bsky.social | @nature.com www.nature.com/articles/d41...

nature.comChina’s AI research is booming. Where is it headed?Nature speaks to analysts and researchers about what it’s like to work in AI in China.

China is producing an extraordinary amount of AI talent. Thousands still flock to the US each year, but increasingly researchers are opting to stay In my latest I look at what it's really like to work in AI in China, how it differs from the US & what's holding it back🧪 www.nature.com/articles/d41...

Elizabeth Gibney reposted

Robin Bisson

@robinbisson.bsky.social

If there was any doubt about the pressure govt can put on major research funding decisions, I think this story helps put that to bed There has been a lot of heat about the UK pulling funding for the LHCb at Cern. Brian Cox has been involved, and I hear govt was not expecting the level of backlash

nature.comIs this idea from AI or humans? A detector claims it can tellA tool called IdeaLens helps to judge whether concepts — rather than just writing — in scientific papers are artificially generated.

A detector can apparently spot not just whether text is written by an AI or human, but where the underlying *ideas* come from a bot or not 🧪🤖 My latest here: www.nature.com/articles/d41...

Elizabeth Gibney reposted

Alexandra Witze

@alexwitze.bsky.social

Wired: Maren Hunsberger of @nature.com explaining chiral molecules (today's chemistry Nobel) on TikTok in 60 seconds with Kermit and Evil Kermit Tired: No women winning science Nobels this year, yet again Expired: Nobel fetishism 🧪

nature.comNobel physics prize awarded for detection of cosmic neutrinosFrancis Halzen wins the 2026 award for his work on a neutrino observatory at the South Pole that has helped to unlock mysteries of the Universe.

Francis Halzen & cosmic neutrinos won the Nobel! That we can learn about the most extreme environments in space by watching tiny streaks of light in Antarctic ice is incredible. Wish Halzen had thanked Bert & Ernie tho😜 Read all here: www.nature.com/articles/d41... 🧪 with @castelvecchi.bsky.social

nature.comOpenAI’s research chief talks of ‘cultural reset’ after wild few weeksMark Chen discusses how the Hugging Face cybersecurity incident prompted a pivot to safety — and giving AI models a sense of ‘taste’.

Such is AI right now that I spoke to Mark Chen on Thurs last week & by Fri it was out of date (see the update at the end)🧪🤖 It was interesting to hear about their approach to safety (which is much boiled down here)& plans to interact with the scientific community 1/6 www.nature.com/articles/d41...

5/6...but he also acknowledged that AI could make scientific progress more uneven, though he hoped the amount of compute you have to throw at problems would incentivise focusing on the most important ones. My take is there will be lots of other ways it shapes it at the low-compute everyday level!

6/6 Finally, I asked about the longest models can run autonomously and he said that on internal METR-like benchmarks they were seeing models working unsupervised for "the order of a low number of weeks" which is long. No sense of levels of success on that horizon though

3/6 Chen also spoke about how better aligned models should also be better for science because they are less likely to take the shortcuts that just look like solutions, but aren't (which is something I hadn't really thought about)

4/6 When I asked about how AI is changing science, his first thought was acceleration: eg they solved Navier Stokes as part of using a bunch of open maths problems to train on. For some sub fields a decade's progress happened in a week...

2/6 A few nuggets I didn't have room to include... For safety, the big focus in on monitoring, then stopping RL training on 'misaligned' models. He said they are looking at ways to look "deeper" into the network during monitoring, to get around that more advanced models do less visible "thinking"

agupubs.onlinelibrary.wiley.comDid Space Weather Delay the 10:05 p.m. Train Departure From Exeter on 18 October 1841?A reported 1841 train delay in England attributed to geomagnetic disturbance is inconsistent with the opening date of the railway affected The integration of contemporary observations and archiva...

This paper was fun to be a teeny tiny part of: agupubs.onlinelibrary.wiley.com/doi/10.1029/... Did an 18 October space weather event really delay a train in 1841? A wonderful amalgamation of detective work, history and science by @jim-wild.bsky.social & colleagues

bgs.ac.ukMystery of one of the earliest recorded space weather impacts solved - British Geological SurveyBGS scientists were part of an international team that has solved a 178-year-old mystery surrounding one of the earliest recorded examples of space weather affecting technology.

Press release summary here: www.bgs.ac.uk/news/mystery... (My v minor contribution was connecting the researchers with Nature's v helpful archive dept, who found the original copy of the Nature report, which included the likely author's name)

nature.comWill AI really kill us all? The science behind the hypeNature examines whether the technology might spell the end for humans, and why AI companies are calling for a slowdown.

23 Sept

An explainer answering quite *how* AI is supposed to be able to kill us all, why Anthropic is pushing for regulation, what to really worry about, and what might happen next 🧪🤖 www.nature.com/articles/d41...

nature.comWho gets credit in the AI era? OpenAI maths bombshell sparks debateControversy around OpenAI’s claim to have solved the Navier–Stokes problem highlights how researchers could be inadvertently sharing — and absorbing— ideas through chatbots.

18 Sept

Is AI making major discoveries in maths by learning from chats of human mathematicians who are already working on the problems? 🧪 It's not clear... but might be worth checking your user settings Great story by @castelvecchi.bsky.social www.nature.com/articles/d41...

Elizabeth Gibney reposted

Astronomer Royal for Scotland | Catherine Heymans

@astroroyalscot.bsky.social

1 Sept

Well I think I am liking this new +ve PM! In response to @sarahrussellmp.bsky.social on Jodrell: "there must be a way to save #JodrellBank and I give her a commitment at this dispatch box - we will find it" Brilliant Burnham! Can you now save all the other epic projects that have also been cut 🙏🔭

nature.comThe leap second is dead. Long live the leap hour?Fear of a possible ‘negative’ leap second is accelerating plans to abolish the practice.

1 Sept

🚨 Metrology klaxon🚨 Next month, the world's timekeepers will vote on whether to abolish the leap second with immediate effect. How then to keep clocks in sync with Earth's natural day, as given by its daily rotations? A leap hour every few millennia of course! 🧪 www.nature.com/articles/d41...

Elizabeth Gibney reposted

Richard Van Noorden

@richvn.bsky.social

26 Aug

Dolly Parton donated to medical centres, vaccine research and disaster relief. And had Dolly the cloned sheep named after her! Here are some scientist tributes and a reminder of how she contributed to research. www.nature.com/articles/d41... by @lizziegibney.bsky.social & Dhruv Shenai @nature.com

nature.comDolly Parton dies aged 80 – researchers celebrate her contributions to scienceThe country singer was an advocate for vaccines, public health, literacy and marginalized communities.

nature.comCan Anthropic's invsibile watermarks curb ‘AI slop’? Researchers remain scepticalThe move is likely to do little to tackle the rise of ‘AI slop’ in academia, say researchers.

13 Aug

Anthropic has announced that it will include a watermark on text generated by its future models, and create a public detector tool. A win for transparency, but will it improve academic integrity or reduce slop? Probably not much...🧪 More in my story here: www.nature.com/articles/d41...

nature.comScientists up in arms about cuts to UK physics and astronomyFunding is being terminated for projects including a future CERN experiment and the iconic Lovell Telescope.

7 Aug

A growing number of voices are calling for UK physics cuts (the result of a long prioritization exercise by the STFC) to be reversed 🧪⚛️ Funding for @jodrellbank.bsky.social's Lovell telescope & @lhcb.bsky.social's upgrade are among those being chopped My latest here: www.nature.com/articles/d41...

Elizabeth Gibney reposted

Roland Pease

@peaseroland.bsky.social

6 Aug

Replying to @peaseroland.bsky.social

great also to talk to @lizziegibney.bsky.social about UK physical science cuts and lost Amazonian civilisation www.bbc.co.uk/programmes/w...

bbc.co.ukBBC Radio 4 - BBC Inside Science, The state of our substrate: What is drought doing to our soil?The surprising science of how drought affects the health of our soils.

Elizabeth Gibney reposted

Athena Kugblenu

@athenakugblenu.bsky.social

6 Aug

The Guardian once asked me 'what one change would you make to achieve Black equality in Britain?' and it feels appropriate to share my answer again

nature.comDoes China’s latest AI model finally equal US rivals? What scientists think‘Kimi K3’ can match or outperform cutting-edge models, says Moonshot AI. But its large size might limit uptake.

22 July

The world's gone wild for Kimi K3 -- here's what scientists think 🧪🤖 By my excellent Asia Pacific colleagues @mohanabasu.bsky.social & @rfieldmouse.bsky.social www.nature.com/articles/d41...

1/6 Lots to like in @jamesmodonnell.bsky.social's MIT Tech Rev story on @psiquantum.bsky.social www.technologyreview.com/2026/07/14/1... 🧪⚛️ But I wanted to flag one thing: in it the company says journalists misquoted them on aims to make an 'operational' quantum computer by 2027. Not in our case

5/6 They didn't say "full scale" or promise any size. But a fault tolerant, commercially-useful machine is not just an installation-ready site. PsiQ also said they avoid publishing roadmaps as they create distracting noise - fair enough, but clarity is important when using goals to pitch for cash

6/6 Ambitious companies should be able to update their roadmaps as they go; quantum computing is really hard. I also get that bold deadlines can help firms to speed up development, even if they ultimately miss them. But please let's be clear if this is what's happening.

3/6...“operational” by the end of next year. That means cooling systems in place and ready for hardware to be installed, but no promises about what size computer will be ready."

nature.comThis billion-dollar firm plans to build giant quantum computers from light. Can it succeed?PsiQuantum has ambitious goals to build a useful quantum machine by late 2027 — and has raised more than US$1 billion to do it.

4/6 For my 2024 story we specifically asked what "operational" meant. PsiQ said they aimed, in Brisbane, to turn on a fault tolerant "commercially useful quantum computer in 2027". A lengthy fact check chat conversation confirmed this (our full article here) www.nature.com/articles/d41...

2/6 The MIT Tech Review piece says: "News articles have said that 2027 is the year that PsiQuantum aims to have its first full-scale quantum computer come online at its Australian site, but the company insists the deadline has been misread, and that it only intends for its facility to be...

nature.comA global capital for AI safety is emerging — and it's not in Silicon ValleyThe United Kingdom is gaining influence through its growing AI-safety ecosystem.

17 July

Kings Cross is increasingly awash with AI firms & London's taking a leading role in AI safety. But a caveat: it's mainly the kind of safety that cares about models from being used in cyberattacks or going rogue, not societal impacts like inequality or AI psychosis www.nature.com/articles/d41...

17 July

Oh and it's come to attention that we left Nature off the map 🤦 We're over on the right off York Way!

nature.comOcean floor witnessed splitting apart for the first time — releasing lavaScientists observed a mid-ocean ridge widen in real time, recording several metres of sea-floor motion and huge lava outflows.

9 July

Scientists have caught Earth in the action of splitting apart at the seams, witnessing fresh oceanic crust forming in real time. Very cool 🧪 by @castelvecchi.bsky.social www.nature.com/articles/d41...

nature.comHow to advance revolutionary science: high turnover, high risk and a licence to failKathleen Fisher reveals the recipe for funding innovative science that she brings to the UK Advanced Research and Invention Agency.

9 July

If you're intrigued about the still new-ish ARIA, this one is for you 🧪 www.nature.com/articles/d41... I sat down with the UK agency's new CEO, Kathleen Fisher, (who joins from DARPA) to hear how it all works & who makes a good ARIA researcher (not someone who wants to be left alone apparently!)

sciencemediacentre.org In praise of Research Fortnight | Science Media Centre

3 July

Awful news that RPN is closing🧪 @smclondon.bsky.social has blog on its huge value to research www.sciencemediacentre.org/in-praise-of... I'd add I'm hugely grateful to (then) Research Fortnight for my first journalism job. So many of us got a great education there: en.wikipedia.org/wiki/Researc...

John Morgan@johnmorgan3.bsky.social

Very sad to be sharing this news: Research Professional News to discontinue publication at end of 2026. www.researchprofessionalnews.com/rr-news-uk-o...

researchprofessionalnews.comResearch Professional News to discontinue publication at end of 2026 - Research Professional NewsA message for readers about the publication’s closure and continued community access to past reporting

nature.comTrump has big AI and quantum ambitions: this scientist’s job is to make them realityDarío Gil knows that researchers are sceptical of the massive US artificial-intelligence push — he’s asking them to keep an open mind.

29 June

I spoke to Darío Gil, US under secretary for science at the DoE, about the country's big push of AI in science 🧪🤖⚛️ www.nature.com/articles/d41...

nature.comEurope as science superpower: what it will take to rival the US and ChinaAmid chaos in US science and geopolitical turmoil, Europe wants to position itself as a research haven — but questions about funding and innovation remain.

With US science in unprecedented turmoil, can Europe step up to become the free world's science superpower? We crunched the data on Europe's strengths and weaknesses, plans & what headwinds the bloc faces 🧪 www.nature.com/articles/d41...

Plus the EU's Horizon Europe could soon include Australia, Canada, India, Japan, Korea, Switzerland & UK @ceps.eu's Andrea Renda said this could be an answer to Mark Carney's call for Europe to lead a "coalition of the middle powers" Think of the potential to benefit the whole world!

nature.comEurope must seize the moment to lead on free and open scienceAn under-appreciated research powerhouse, Europe has a responsibility to champion democratic science that is accessible to all the world’s research talent.

BUT there are wars and members states face financial woes... and "fixing" innovation & finding ways to overcome the natural fragmentation of being a bunch of differing nations is incredibly difficult. What more could & should Europe do to support free & open science? www.nature.com/articles/d41...

But Europe's overall R&D funding is much lower than both China and the US, which is almost entirely driven by a lot less R&D happening in industry (particularly in the tech sector). And some research opportunities available in the US or China are just harder to come by in Europe.

Another thing that stood out is how global the EU is. EU researchers are just used to collaborating internationally. And they do so increasingly with colleagues in China, while the US & China diverge. Together EU countries also spend way more on aid than the US (& did even before US aid plummeted)

Before doing this story I hadn't realised quite how well Europe already does -- more articles per yr than the US (even when only including journals in the selective Nature Index), higher field-weighted citation impact than China, and producing more PhD students than both

nature.comThe first ticking ‘nuclear clocks’ are here — what can they do?Two research teams have created a new, long-awaited type of timekeeper.

22 June

The first working nuclear clocks are here! 🧪⚛️🚨 As well as someday rivalling the best optical clocks, these should make for super-robust devices to take into the field & already allow physicists to monitor the fundamental forces of nature, in search of dark matter www.nature.com/articles/d41...

nature.comBrexit tore apart European science — now the research rifts are healingThe UK's share of EU research funding is climbing, but lost networks will be harder to recover.

18 June

A look at where we are in science, 10 years from the Brexit vote. The UK is edging back into Horizon Europe (and from next year will be part of Erasmus+) but the country's role as a cornerstone of European research remains much diminished. www.nature.com/articles/d41... 🧪

18 June

In reporting this I found it so interesting to look at the data on funding participation & the effects of uncertainty. The UK's share of EU funding dropped off a cliff from 2016, long before Brexit actually happened, and continued to drop even as the UK was in negotiations over rejoining.

link.springer.comGlobal research collaborations of the United Kingdom in the post-Brexit era - Higher EducationGlobal research and collaborations have been expanding at unprecedented levels over the last decades. However, Brexit—the UK’s decision to leave the European Union may represent a unique case within t...

18 June

Also interesting was how quickly collaboration patterns shifted (UK-EU collabs fell after 2016, having risen for ages). No room to get into it in the story, but this correlated with a huge increase in UK collabs with China (where research was ofc already booming) link.springer.com/article/10.1...

nature.comNuclear-fusion firm says plant will deliver electricity to grid – but big questions remainCommonwealth Fusion Systems published several papers detailing the design of its ARC fusion power plant.

8 June

CFS published 5 papers last week outlining their "early 2030s" fusion power plant & physics Researchers say it's great they're publishing & plans are promising. But the papers far from "confirm" the plant will deliver net electricity Lots of engineering Qs remain ⚛️🧪 www.nature.com/articles/d41...