I am writing from Argentina as a researcher working within the country’s national scientific system. While discussions about science funding are often framed in terms of budgets and economic indicators, the reality is experienced in laboratories, offices and research facilities where scientific knowledge is produced.

Today, many of us in Argentina are witnessing a progressive deterioration of Argentine science, both in terms of infrastructure and human capital.

I am based at the Comisión Nacional de Energía Atómica (CNEA), where uncertainty has become routine. Hundreds of employees remain on temporary contracts that can be terminated without warning. On 1 July 2026, a total of 61 colleagues lost their jobs, with armed military police officers occupying the building to contain a peaceful employee protest. Beyond the immediate human cost, these dismissals are steadily dismantling scientific capabilities that have taken decades to build.

The impact became tangible when my colleague Paula Alderete, an experienced electron microscope specialist, lost her position. Alderete, who has published two papers with me, was responsible not only for operating a highly specialized instrument, but also for maintaining it and supporting dozens of research projects.

Without Paula Alderete’s expertise, our lab’s electron microscope has become largely unusable. It is a quiet reminder that scientific infrastructure depends as much on people as on expensive equipment.

Faramarz S Gard

Without her expertise, the microscope has become largely unusable. Every morning I pass her empty desk on my way to the laboratory. It has become a quiet reminder that scientific infrastructure depends as much on people as on expensive equipment.

The deterioration extends beyond personnel. In my laboratory, an X-ray photoelectron spectroscopy (XPS) instrument worth more than $1m must sometimes be protected with plastic sheets because rainwater leaks through the roof.

Equipment designed to operate under tightly controlled environmental conditions is now exposed to circumstances that threaten its long-term reliability. Similar situations are becoming increasingly common across Argentina’s universities and research institutes as maintenance budgets shrink and infrastructure ages.

Argentina has a scientific tradition that reaches far beyond its borders. The country has produced five Nobel laureates, including three in the sciences, and has made internationally recognized contributions to nuclear technology, medicine, agriculture, astronomy and materials science. Institutions such as CNEA, CONICET (Argentina’s national research council) and the national universities have trained generations of researchers whose work has strengthened international collaborations and advanced scientific knowledge worldwide.

Scientific systems rarely collapse through a single dramatic event. They decline incrementally: one technician dismissed, one laboratory left unrepaired, one instrument taken out of service, one young scientist deciding to build a career elsewhere. These individual losses accumulate into a weakening of scientific capacity that may take decades to rebuild.

When a country loses the expertise required to operate advanced laboratories or maintain unique scientific infrastructure, the consequences extend beyond national borders.

Faramarz S Gard

This should concern more than Argentina. Modern science depends on international networks of researchers, shared facilities and long-term collaboration. When a country loses the expertise required to operate advanced laboratories or maintain unique scientific infrastructure, the consequences extend beyond national borders.

Knowledge is lost, collaborations weaken, and opportunities for discovery diminish.scientific infrastructure is more than buildings and instruments; it is the people whose expertise gives those resources value. Once that expertise disappears, replacing it is far more difficult than preserving it.

The empty desk of a single scientist may seem insignificant, but multiplied across laboratories and institutions, it becomes a warning of how a research system is quietly dismantled. Argentina’s experience is therefore not only a national concern but also a reminder that scientific capacity, wherever it exists, is a global resource that is remarkably difficult to rebuild once lost.

I write this letter not only to describe a difficult situation but also to call attention to the broader consequences of sustained disinvestment in science. The international scientific community should recognize that the erosion of research systems in countries such as Argentina represents a loss not only for those nations but for global science itself.

Science is built by people, sustained by institutions, and enabled by infrastructure. Today, all three are under increasing strain in Argentina.

The post Letter from Argentina: ‘an empty desk and an idle electron microscope signal the erosion of the country’s science’ appeared first on Physics World.

The New York Times recently published a 124-page special issue of its Style Magazine entitled “How to be cultured”. According to the cover of this glossy, ad-driven supplement, the contents were “idiosyncratic” but amounted to a “compendium of what you need to know right now”. The material had been “chosen by experts” for the purpose of “teaching readers how to be more cultured”.

Curious as to what would be included, my hopes were raised by the magazine’s lead editorial. It said that the issue “celebrates expertise, a quality ever less valued in our culture today, where learnedness and firsthand research can be overshadowed by slick presentation and dumb certainty”.

Now, I’m not naïve. I know that the New York Times Style Magazine caters to advertisers, who in turn cater to wealthy readers who feel the need to spend gobs of money to get culture. It didn’t therefore surprise me to have to leaf through 40 pages of adverts for jewellery, champagne, watches and home furnishings before finally coming across the table of contents.

I saw headings for film, food, music, theatre, fashion and the visual arts. Shakespeare showed up several times: over half a dozen plays, sonnet #94 and one of his monologues. There were entries for 1930s movies, animated videos and films from Brazil. One page on food and drink covered porridge from Norway, meat from Mexico, lemonade from Vietnam and wine from Austria.

There was no mention of science in “How to be cultured”, nothing at all

Novels from the US, UK, France, Japan, India and Brazil made the list, as did selections of music from medieval times to opera and hip-hop. Samples of theatre, puppetry and architecture made it in too. But there was no mention of science, nothing at all. The closest I found were architectural materials such as plaster, glass, steel and textiles.

Snow’s lament

The absence of anything scientific should not have shocked me; in fact, it reminded me of the British scientist and novelist C P Snow’s 1959 essay The Two Cultures and the Scientific Revolution. Even back then, Snow lamented that humanists and scientists were living in two separate cultures divided by a “gulf of mutual incomprehension”.

In the essay’s most famous passage, Snow wrote that “a good many times I have been present at gatherings of people who, by the standards of the traditional culture, are thought highly educated and who have with considerable gusto been expressing their incredulity at the illiteracy of scientists”.

“Once or twice,” he went on, “I have been provoked and have asked the company how many of them could describe the Second Law of Thermodynamics. The response was cold: it was also negative. Yet I was asking something which is about the scientific equivalent of ‘Have you read a work of Shakespeare’s?’”

Snow did not stop there. “I now believe,” he continued, “that if I had asked an even simpler question – such as, ‘What do you mean by mass, or acceleration’, which is the scientific equivalent of saying, Can you read? – not more than one in ten of the highly educated would have felt that I was speaking the same language.”

“So the great edifice of modern physics goes up,” he concluded, “and the majority of the cleverest people in the western world have about as much insight into it as their neolithic ancestors would have had.”

Culture club

Snow’s tone was snobby and arrogant, but he was on to something. Can someone who knows nothing about, say, Einstein’s equation E = mc² or about Heisenberg’s uncertainty principle really be called cultured? Sure, cultured individuals don’t need to know much about these topics. But they ought at least to realize that those equations concern how mass and energy work and the inherent limitations to what humans can know.

I’d say those are principles fundamental to the formation, maintenance and fate of the universe – and ones that a cultured person ought to know something about. Indeed, references to them are routinely found in arts and popular culture. Einstein’s equation has made the cover of Time magazine, while the uncertainty principle has been taken up by novelists, theologians and comedians.

Shouldn’t a cultured person be acquainted in some way with the existential threats facing the world?

Further, shouldn’t a cultured person be acquainted in some way with the existential threats facing the world? Wouldn’t such a person have to have an inkling of the way a carbon molecule absorbs and emits heat, of how vaccines work, and of safe and unsafe levels of substances toxic in high doses? How cultivated can an ostrich be?

Of course, there was nothing about the second law in “How to be cultured”, or about any other science of the sort Snow had in mind.

Nothing about scientific equations, experiments, theories or facts that a cultured person ought to know.

No discussion of dark matter, dark energy or of quantum physics, genes or vaccines. Nor about the Pythagorean Theorem or the double-slit experiment.

The critical point

Culture refers to the influences and values that are embedded in the way we live. Judging such influences and values – deciding which influences and values ought to be so embedded – is a tempting but arrogant game.

So let’s play. Imagine that you are among the experts charged with compiling a “compendium of what you need to know right now”. What would you include of science – and why? Send me your contributions, and tell me why they are what we need to know, and I’ll write about them in a future column.

The post Tell us what science you need to know to be ‘cultured’ appeared first on Physics World.

The free-to-read Physics World Instrumentation and Vacuum and Briefing 2026 is now available. It includes R&D updates, interviews that explore technology commercialization, and a fun look at the quirky side of SI units.

One topic covered in the briefing is the world of quantum sensors. Over the years, physicists have developed some amazing quantum sensors, but many of the technologies have stayed in the lab because of the challenges of miniaturizing key components. With ultrahigh vacuum (UHV) playing an important role in quantum sensors based on cold atoms, Florence Concepcion of the UK-based firm Aquark talks about her mission to reduce the size and energy consumption of UHV systems.

Elsewhere in the issue, we look at how manipulating individual living cells plays an important role in biology and medicine. Trouble is, cells tend to stick together when grown in vitro and separating them often involves using harsh chemicals that can damage cells or modify their properties. In a special interview, Luke Cox, who co-founded the UK-based firm Impulsonics, discusses its system that uses ultrasound to gently separate living cells.

Another entrepreneur featured in this briefing is Brian Pogue, co-founder of DoseOptics. The US-based company has developed a system that detects the extremely faint Cherenkov light that is emitted when a radiotherapy beam strikes a patient’s skin. This allows the radiotherapy to be monitored in real time, ensuring that the beam passes through the target tissue and avoids healthy areas of the body.

Also included is a look at the use of intense laser light to accelerate particles. Find out how researchers in the US have created a compact, free electron laser that is driven by a laser plasma accelerator (LPA) – and it has also been used to create a beam of muons.

Finally, you can explore some of the curiosities of the International System of Units (SI), which is the bedrock of metrology. Although SI and its predecessors have been honed and re-defined over the centuries, it still has some surprising quirks. Ben Stein from the US National Institute of Standards and Technology explores some of these oddities including how the candela was derived from the brightness of a candle made from whale fat and beeswax – and the ongoing debate about using the dimensionless radian as the SI derived unit for planar angle.

• Read the Briefing now.
• You can also watch our Physics World Live Quantum metrology panel discussion.

The post The 2026 <em>Physics World Instrumentation & Vacuum Briefing</em> is out now appeared first on Physics World.

Private fusion companies have raised almost $4.5bn in funding over the past year – a 70% increase year-on-year. That is according to a report on the state of the fusion industry, which was published yesterday by the Fusion Industry Association (FIA). The report – The Global Fusion Industry in 2026 – also finds that private fusion endeavours now employ over 16,000 people worldwide.

The report surveyed 56 private fusion companies around the world – most of which are based in the US and Europe. Some of the biggest players in private fusion include Commonwealth Fusion Systems (CFS), General Fusion, TAE Technologies and Tokamak Energy.

Magnetic confinement, in which magnetic fields are used to contain a high-temperature plasma, is the most popular fusion technique, being used by 48% of firms according to the report. Inertial confinement, which uses rapid compression, usually by lasers, to create a confined plasma for a short period of time, is employed by 21% of firms – with other techniques such as magneto-inertial making up the remainder.

While electricity generation is a main target for private fusion companies, many also see the technology being applied to space propulsion, with other markets including marine propulsion, hydrogen fuel and industrial heat.

The funding boost for this year’s figures has come from major funding rounds, with CFS raising $863m in August 2025 while Inertia Enterprises, Helion Energy and Proxima Fusion have raised about $1.5bn in total this year.

Ambitious goals

The first FIA report on the fusion industry was published in 2021, where it found that private fusion endeavours had received over $1.8bn of funding since the 1990s.

The 2021 report discovered that over two-thirds of the 35 companies surveyed thought that electricity generated from fusion would enter the grid in the 2030s, with a further 20% thinking it would more likely come in later decades.

That confidence, however, has grown – with this year’s report finding that 71% expect the first fusion plant to deliver commercial electricity by the 2030s.

Indeed, FIA chief executive Andrew Holland says he is “confident” that commercial fusion will be delivered in the 2030s.

“Alongside private investment, fusion companies still need the support of governments to address common challenges including the availability of resilient materials and the fusion fuel cycle,” he says. “The governments that update their programmes and funding priorities to meet the sector’s needs today will be the ones to capitalize on this vital emerging industry.”

The report comes as General Fusion has become the first fusion company to be publicly listed. It joined the Nasdaq exchange with $150m in fresh funding, which is expected to go towards the firm’s Lawson tokamak programme that is expected to be complete in 2028.

The post Private investment in fusion firms jumps by $4.5bn appeared first on Physics World.

In The Hitchhiker’s Guide to the Galaxy, Douglas Adams imagined a supercomputer called Deep Thought, built by a race of hyper-intelligent beings to calculate the Answer to the Ultimate Question of Life, the Universe and Everything. After a mere seven and a half million years of computation, the machine finally revealed its answer: 42.

There was just one problem. Despite their hyper-intelligence, none of those beings understood what the question had been. The joke was that any answer – no matter how precise – is meaningless without a thorough understanding of both the question and the route taken to obtain the answer.

Yet what was once purely science fiction is beginning to sound unexpectedly relevant to modern research. From structural biology to particle physics, artificial intelligence (AI) systems are increasingly involved in research, providing results of remarkable power and accuracy. They are being used to identify hidden patterns in vast datasets, to generate hypotheses, to accelerate simulations and to guide experiments.

As AI becomes increasingly embedded in the scientific process, could we be entering an era of discoveries without understanding?

But in some cases, even the creators of those AI systems are struggling to explain exactly how the AI arrives at its  conclusions. All of which raises an uncomfortable possibility. As AI becomes increasingly embedded in the scientific process, could we be entering an era of discoveries without understanding? And if so, what happens when science starts producing its own versions of the answer “42”?

AI and accelerated particle searches

For particle physicists, this question is no longer hypothetical. At CERN, the home of the Large Hadron Collider (LHC), AI is already woven into day-to-day research, subtly reshaping how physics research is carried out. Indeed, machine learning (ML) – one of the main pathways through which AI is achieved – has long played a role at the LHC.

Back in the early 2010s physicists working at the CMS and ATLAS experiments, routinely used ML in their search for the Higgs boson. The ML algorithms separated rare signals – a Higgs boson decaying either into two photons or into four leptons via Z bosons – from far more common Standard Model processes that produced similar signatures in the detector.

Machine learning did not discover the Higgs boson by itself, but it accelerated the search by making analyses more sensitive. In fact, achieving the same sensitivity without ML would have required between 15% and 125% more data, according to a review of Higgs-decay measurements (Nature 560 41). Some measurements would even have needed more than double the data.

But the scope of AI in particle physics has been expanding rapidly since then. Neural networks are now used to compress enormous datasets, accelerate simulations, and identify unusual events hidden among billions of particle collisions. By learning directly from examples, neural networks can uncover complex correlations in detector data that would be difficult for physicists to unravel using conventional analysis techniques.

Caterina Doglioni, an experimental particle physicist from the University of Manchester, UK, who uses ML for data compression as part of the ATLAS collaboration, says that deployment of AI is becoming difficult to separate from particle-physics research itself. “It’s pretty much everywhere,” she says, especially when it comes to experimental high-energy physics. “It’s critical at the LHC where far more data is created than can realistically be stored.”

When it comes to compressing detector data, for example, one option is to use neural networks to force information through a computational “bottleneck” that preserves the most important features while dramatically reducing the storage required. The original data can then be reconstructed from this compressed summary, either exactly or approximately depending on the application (Rep. Prog. Phys. 84 124201).

This is different from the ML techniques used during the Higgs search. Whereas Higgs analyses relied on supervised algorithms that had been trained to distinguish known signals from background noise, these data-compression methods typically use “self-supervised” learning. They identify and preserve important patterns – without requiring labelled examples.

Algorithms learn what ordinary collisions look like and then flag events that deviate from those expectations

More recently, researchers have begun applying ML to a very different challenge. In the Higgs search, algorithms were trained to recognize signatures predicted by theory and distinguish them from background processes. But now, instead of having algorithms that search for a specific predicted signal, those algorithms learn what ordinary collisions look like and then flag events that deviate from those expectations.

In what is essentially a clever form of anomaly detection, the hope is that new physics might reveal itself as something unusual, even if physicists do not yet know exactly what form it will take. In a nutshell, ML used to be applied to problems where physicists knew what they were looking for; now it is applied to problems where we don’t know what we’re looking for just yet.

While no new particles have yet been spotted with these anomaly-detection techniques, they have progressed from theoretical proposals to real experimental searches. In 2024, for example, the ATLAS collaboration reported the results of an unsupervised anomaly-detection search using Run 2 data. No significant deviation from Standard Model expectations was observed, but the study proved that such methods can be deployed successfully on real collider data (Phys. Rev. Lett. 132 081801).

The black box problem

Yet the very feature that makes these systems so powerful – their ability to rapidly identify patterns humans may miss – also raises one of the most persistent concerns surrounding AI in research: the black box problem. Modern ML systems can uncover complex relationships in data, but researchers may struggle to understand exactly how a model arrived at a particular result. In particle physics, where extraordinary claims require extraordinary evidence, that lack of transparency can create unease.

For David Sutherland, a theoretical physicist at the University of Glasgow, the greater concern is not necessarily whether researchers can interpret every internal feature of a model – but whether they can trust and reproduce its outputs. “I would certainly say reproducibility [is more important] than interpretability,” he says.

That distinction matters. Science has always relied on the principle that results should survive scrutiny, be independently verified, and withstand repeated testing. A model does not need to reveal every detail of its inner workings – but researchers must be able to demonstrate that it behaves robustly across datasets and conditions.

Particle physics, in particular, has built extensive safeguards around that process. “We have very large review committees that check basically everything that goes on,” Doglioni says, with those structures providing an additional layer of protection against over-reliance on opaque systems.

Reliability and reproducibility

Yet as AI systems become increasingly embedded in research workflows, ensuring that reliability may become more difficult. Reproducing results may require not only access to data and code, but also to model architectures, training conditions and computational environments.

Unlike many commercial AI systems, the ML tools used in particle physics are often developed by particle physicists and shared openly with the scientific community. Even so, reproducing results may require access not only to data and code, but also to model architectures, training conditions and computational environments, all of which can influence an AI system’s behaviour.

The growing use of AI could also begin to change how physicists work and even how scientific value is measured. Christoph Weniger, a physicist at the University of Amsterdam who applies ML and AI techniques to particle physics and cosmology, argues that AI systems are likely to take on an increasingly active role in navigating complex analyses. Researchers, he believes, will shift into “a supervisory role in steering these different agents”, fundamentally changing how scientists interact with data.

Creative impact

Such changes could extend beyond day-to-day workflows. Traditionally, the currency of academic research has been publications. But Weniger suggests that AI may shift the emphasis away from productivity alone and towards something harder to quantify: originality.

Scientific value may increasingly depend not on how many papers a researcher produces, but on the creativity and impact of the questions they ask

As routine technical barriers to experiments and theoretical work become easier to overcome, he argues that scientific value may increasingly depend not on how many papers a researcher produces, but on the creativity and impact of the questions they ask.

The effects may also extend to how scientific knowledge is communicated. The scientific literature has long been criticized for being difficult to navigate, not simply because of its volume, but also because of the way researchers write. Nichol Furey, a mathematical physicist at Humboldt University of Berlin whose work explores whether there is an underlying algebraic logic to the structure of the Standard Model, believes AI could help bridge that gap.

“Authors in maths and physics often write in ways that are overly opaque,” she says. “AI systems write with their audience in mind, whereas human authors often don’t.” Yet that possibility creates a tension. The same tools that can make research more accessible can also make scientific text dramatically easier to produce.

Communities are already beginning to grapple with the unintended consequences of widespread AI adoption. Recently, arXiv, one of the world’s largest repositories for physics and mathematics preprints, announced a crackdown on unchecked AI-generated submissions. It said it was introducing one-year bans for authors who upload papers containing obvious signs of unverified AI content, such as hallucinated references or leftover chatbot instructions.

The move reflects growing concern about the quality of AI-assisted scientific writing. Although arXiv’s new policy is too recent for its impact to be fully assessed, the threat of a one-year ban marks one of the strongest responses yet by a major scientific repository to unchecked AI-generated content.

The concern is not simply that AI can generate poor research. Now that producing scientific text is an almost frictionless exercise, the burden of judgement shifts increasingly onto readers, reviewers and future researchers trying to separate genuine insight from noise.

Training future physicists

Another potential problem with AI’s infiltration in many areas of research is training the next generation of physicists. This is generally a long process, where expertise is built gradually through years of practice and failure. If AI automates many of those steps, researchers may eventually face an uncomfortable question: how do you train future experts if the work that once trained them disappears?

This challenge extends far beyond physics. Across the education sector, schools and universities are already grappling with the ease with which generative AI can produce essays, solve problem sheets and complete programming assignments in seconds.

In fact, Wrishik Naskar, a particle theory postdoc at the DESY laboratory in Germany, wonders if research habits are also beginning to change. “When I get stuck, I go to my supervisor,” he says, “[but] many people, the first thing they ask is ChatGPT.” That approach might speed things up, but does it really help a researcher learn and grow? As Naskar puts it: “Focus on the learning. Results will follow if you have skills.”

Despite the many concerns, not everyone sees the transition to AI as a threat, and framing AI as a competitor to researchers may miss the point entirely. For example, Admir Greljo, a particle theorist at the University of Basel, believes the real shift is not replacement but collaboration. “It’s not that they will compete against AI. Entering into theoretical physics is still going to be very interesting and very important. AI is not competition, it’s a new tool.”

That perspective echoes a broader historical pattern. Physicists once calculated by hand, then with slide rules, then with computers. Each technological leap changed how research was done and prompted fears about what skills might be lost along the way. Yet rather than eliminating physicists, those tools expanded what physicists could ask.

Towards an AI scientist?

AI may simply be the next step in that progression. But it also presents something new. Unlike previous tools, AI can suggest ideas, identify patterns and increasingly participate in parts of the scientific process that once seemed uniquely human. So might that participation one day become independence? Could AI move beyond assisting physicists and begin carrying out the entire research process itself?

For Sutherland, a future in which AI systems perform the entire research cycle, from generating hypotheses to collecting data and even writing papers, is not difficult to imagine. But he still envisages a human “steering it” and acting as a “sanity check”, much like the relationship between a senior researcher and a PhD student.

Others go further. Greljo believes that fully autonomous scientific systems may not arrive in the near future, but sees no fundamental reason why they could not emerge eventually. Because science ultimately depends on reproducible results, he argues, humans could simply repeat calculations and test predictions to verify whether the AI had reached the right answer. Such a future, he says, is “not something that is impossible”.

We may be entering a world where machines help us uncover truths about the universe faster than we can fully understand them

We may be entering a world where machines help us uncover truths about the universe faster than we can fully understand them. But that need not signal the end of scientific inquiry. It is important to remember that AI models, much like many of the models used throughout scientific research, are imperfect and may always remain so. But perhaps they do not need to be perfect.

Science has always involved navigating uncertainty, constructing incomplete models and refining them over time. AI may help us discover things faster than ever before. It may reshape how future generations of physicists learn, work and think. But science has never simply been about generating results. It has also been about explanation, curiosity and deciding which questions are worth asking in the first place.

We should not treat AI as an oracle that bypasses the scientific method. Instead, we should view it as another tool in the physicist’s arsenal – a powerful one, certainly, but still a tool, helping researchers explore possibilities, test ideas and refine their models of reality.

After all, the problem with Deep Thought was never that it gave the wrong answer. It was that nobody knew what question they had asked.

The challenge for the next generation of physicists may not be deciding whether AI can discover new truths about nature. It may be making sure we still know which questions are worth asking.

The post AI’s black box problem: discovering physics we don’t understand appeared first on Physics World.

In 2008 the US author Neal Stephenson published a speculative fiction novel called Anathem. Set on the fictional planet of Arbre, the story follows an order of monks, mostly isolated from the rest of the world, who every so often open up their doors to receive information on the state of the outside world.

In the book, Stephenson explores a Dark Age of the Internet, in which companies flood the network with plausible-sounding-but-false information in a ploy to force consumers to buy the firms’ own tools for distinguishing truth from falsehood.

Initially, this “crap”, as Stephenson calls it, had to be made by humans, but eventually the process is run by a so-called “Artificial Inanity”, which is able to create hundreds or thousands of bogus versions, or “bogons”, for every legitimate document.

Today, we face our own version of Stephenson’s Dark Age as our information landscape increasingly fills with AI slop, now going beyond AI-generated videos of uncanny kittens or images depicting politicians and celebrities in outlandish situations.

With the rise of chatbots powered by large language models (LLMs), our search results, social-media feeds and even scientific journals are getting polluted with data that, at first glance, appears relevant and plausible but contains no value beneath its gilded veneer.

As LLM-generated slop infects our previously trusted knowledge base, we face an information landscape that is quickly rotting beneath us.

As scientists, we must position our work in relation to other recent research. Yet that task gets harder when we have to sort out the real from the hallucinated. As LLM-generated slop infects our previously trusted knowledge base, we face an information landscape that is quickly rotting beneath us.

Where we could once track an idea to a paper with human authors (who bore consequences for acts of fraud), we face a future where we have to question whether an article’s authors, institutions or experiments ever existed, on top of evaluating its scientific merits.

This is not a hypothetical danger. In 2024 a team led by medical researcher Almira Osmanovic Thunström at the University of Gothenburg invented a fictitious medical condition called “bixonimania” and publicized it through posts on Medium and two pre-print papers.

Their goal was to test whether AI chatbots would pick up and propagate word of the fake disease, which they duly did. The researchers even tried to insert information that made the work obviously fake. This included acknowledging a colleague from Starfleet Academy, thanking the University of Fellowship of the Ring for funding, and even writing early on in the article that “this entire paper is made up”.

Trapped in a bog

As physicists, we are not immune to the dangers of information rot. A flood of LLM-generated articles prompted arXiv in May to announce consequences for authors whose submissions have “incontrovertible evidence that the authors did not check the results of LLM generation”, including having “hallucinated references”. Those consequences include a one-year ban from arXiv and the requirement, thereafter, that any submitted paper must already be accepted at a “reputable peer-reviewed venue”.

I initially welcomed the news. Harsh consequences are one way to discourage LLM usage, especially for sloppy shortcuts. But the more I reflect on it, the more I realize that arXiv’s consequences could eventually fall even on those who are not using LLMs to write their papers.

Relying on shortcuts is easy. Say you’re writing a literature review and paper A, which you’ve read and trust, has a fact you want to cite – but you notice that paper A cites paper B for this fact. It’s simple to scroll down to the reference section and copy paper B’s citation into your own citations (I would be lying if I said I’ve never done this myself).

But as the information landscape rots, this shortcut could run you afoul of policies like arXiv’s. What appears to be a valid citation – maybe even containing the names of authors you know personally – can be hallucinated.

We must double down on the skills that make us experts – our scientific judgement, information literacy, communication skills and careful scholarship.

Or it could be entirely fake, as a team of researchers who cited one of the “bixonimania” pre-prints in a peer-reviewed article discovered. Their paper was justifiably retracted by the journal editor, over their objections. By simply copying a reference without double-checking it, you can unintentionally join ranks alongside LLM users. Worse, you open yourself and all of your co-authors to being banned from arXiv.

I am not suggesting that arXiv’s policy is flawed. Rather, I argue that those who intend their writing to be AI-free must be vigilant. We must double down on the skills that make us experts – our scientific judgement, information literacy, communication skills and careful scholarship. We should lean into practising these skills – and teaching them – even though they are precisely the skills that AI companies claim LLMs can replace.

In Anathem, a character simply hand-waves away their Internet full of bogons, but we have no such solution. Instead, as our information landscape rots, we find ourselves trapped in a bog. To get out, we must plan each step carefully and take none for granted. What appears to be solid ground may well prove treacherous. Moving safely is a slow process, full of checking and double-checking as we painstakingly build out paths and tools we can trust.

Simply choosing not to use an LLM when writing is no longer enough. We have only our own scholarly skills to sort the bogons from the ground truth. We will all have to slow down if we want out of the bog.

The post In the age of AI, we must double down on the skills that make us experts appeared first on Physics World.

It was fantastic to see the men’s marathon world record being broken in London at the end of April when Kenya’s Sabastian Sawe became the first person to officially complete the run in under two hours. It’s an achievement that a whole generation of world-class runners had had their eyes on, with Sawe completing the race in a headline-grabbing 1:59:30.

Perhaps less attention was given to another achievement a week earlier, which saw a humanoid robot run the Beijing E-Town half-marathon in 50:26, breaking the human record by almost seven minutes. What’s particularly impressive is that at the inaugural event in 2025, the robot winner took 2:40:42 to complete the race, with many robot entrants failing even to reach the finishing line.

In just one year, in other words, the robot half-marathon record has been slashed by more than two-thirds, an unthinkable feat in human terms. What’s more, the rules between the first and second robotic competitions were tightened. They now favour robots that can run on their own (rather than via remote control) and penalize those that need to be taken off course to have their batteries changed.

Giant steps

My mind began spinning, trying to work out what must have happened over the past year to realize such advancements. How did we go from remote-controlled, slow and clunky robots to autonomous, agile and sleek devices? It turns out that these increases in speed and endurance have been driven by a combination of better hardware, specialized designs and advanced algorithms.

Robots now benefit from powerful actuators enabling higher torque and faster movement of its hips and knees. Anatomical designs have also been optimizied to mirror biological efficiencies by, for example, using lightweight limbs and lower torso components to minimize the energy lost when a robot’s foot strikes the ground.

Other changes include the use of liquid cooling technologies to stop a robot from getting too hot during prolonged activity and finding ways to mimic highly efficient natural bipeds, such as an emu or ostrich. Improved AI and control software have also allowed robots to navigate varied terrain and stay stable autonomously, rather than relying on remote operation.

AI answers

Of all these developments, I believe the most significant is the expanding use of AI. To me, it seems that AI will be the enabler for robots to become more “human”. In China, humanoid robots are already being developed to automate what is dubbed the hardest task in the car industry – the final assembly stage.

This work typically consists of real people carrying out highly labour-intensive tasks such as installing wiring harnesses, fitting a vehicle’s interior trim and instruments, or bringing its engine and chassis together. Being super-precise tasks that require manual dexterity, this work has previously been thought too complex for robots.

However, through a combination of imitation learning, simulation and real-time human guidance, it’s now becoming conceivable for robotic devices to do this kind of work. Underpinning this aspiration are what are known as  Vision-Language-Action (VLA) models, which process camera images together with natural language or text descriptions and translate these into physical actions.

Having emerged relatively recently from work at AI firm Google DeepMind, VLA models differ from traditional, task-specific programming robotics by offering generalization and the ability to handle novelty. They are seen as the next big thing in AI because they bridge the gap between understanding the world (vision and language) and interacting with it (action) – potentially letting robots do tasks they have not been explicitly trained on.

VLAs typically rely on existing Vision-Language-Models (VLMs), which provide a combined knowledge of text (large language models) and images (computer vision models). The VLA model is then fine-tuned on task-specific data to learn a mapping of visual observations and text instructions to create the desired robot action.

One challenge in training VLAs is getting enough data with appropriate visual and contextual information. It may be gleaned from real robots (imitation) but it can also be generated by human guidance or teleoperation – a manual process involving humans “showing” robots how to do a specific task. The task is time-consuming, having to be performed deliberately and repeatedly using equipment that captures appropriate levels of data quantity and quality.

A third option is to build virtual simulation environments to train the models and optimize robotic movements. As VLAs inherently learn to associate higher-level cognition with lower-level physical actions, the trend will be to move from fixed, pre-programmed scripts towards AI and large-scale simulation training. The result: robots that become more autonomous and versatile in real-world situations.

We are all getting used to AI becoming an increasing part of our everyday life – think how much GenerativeAI models have improved since they first appeared a few years ago. But I wonder what will be next for robots as they become capable of moving independently and spontaneously, having the ability to operate with full autonomy within dynamic human environments, such as homes and factories.

When we – eventually – experience the full physical embodiment of Artificial General Intelligence, will it be like living on the filmset of I, Robot or Blade Runner, with humans and humanoids coexisting and, in some cases indiscernible? Several years ago, this was something I did not expect to see in my lifetime. But in April it just got one giant leap – or at least one half-marathon – closer.

The post How AI is transforming human robots appeared first on Physics World.

Some animals could sense Earth’s magnetic field by using the superparamagnetic properties of white blood cells in their livers, a new interdisciplinary study has revealed. Led by Clivia Lisowski at the University of Bonn, the team’s insights into homing pigeon navigation could help scientists resolve the decades-old mystery of how animals can sense magnetic fields. The research also shines new light on the inner workings of the immune system.

Animals ranging from sharks to bats appear to use Earth’s magnetic field to navigate during long migratory voyages. However, the nature of this magnetoreception is poorly understood.

“How they sense Earth´s magnetic field has been a mystery,” Lisowski explains. “Several hypotheses were suggested, but these have often resulted from laboratory experiments. They were hardly ever done in the field, and so couldn’t explain how migration would work at night or in dark environments.”

In their study, Lisowski’s team investigated the origins of magnetoreception in homing pigeons. These birds are famous for finding their way back home after being released in unfamiliar locations. “We combined expertise from research fields that usually are not intersected, namely immunology, physics and animal behaviour,” Lisowski explains. “[This gave] us a truly interdisciplinary approach to solve the mystery of magnetoreception in homing pigeons.”

Searching for magnetism

As part of this approach, the researchers searched for magnetism in several different tissues of the bird – including the liver, spleen, muscle, and beak. They used vibrating-sample magnetometry (VSM), whereby samples are placed in a constant magnetic field and vibrated up and down. If the sample is magnetic, it will induce an electric field that is proportional to its magnetization.

The team also did field studies of homing pigeon flight patterns and studied the birds’ anatomy. The latter involved using dyes to visualize microscopic structures embedded in tissues along with genomic assays of the tissues.

This combined analysis revealed the presence of superparamagnetic macrophages in homing pigeon livers. Macrophages are specialized white blood cells whose primary function is to engulf and digest pathogens. They are a cornerstone of the immune system. Superparamagnetic means that the macrophages are very easily magnetized.

“The macrophages are superparamagnetic due to their physiological function of degrading old and damaged red blood cells,” Lisowski explains. “They sequester and store the iron from red blood cells as oxide nanoparticles in ferritin proteins.”

Connecting to the brain

Within homing pigeon livers, the researchers found that these macrophages were often in the vicinity of – and sometimes even in direct contact with – nerve fibres in liver tissue. This could allow magnetic information to be passed from macrophages to the brain.

At the atomic level, Lisowski’s team proposes that the effect emerges from unpaired electrons in ferritin proteins, which interact with each other through magnetic dipole–dipole coupling to enhance their magnetic susceptibility. This subatomic effect could in turn explain the distinctive circling behaviour observed in pigeons after they take off. This circling could allow these unpaired electrons to be imprinted with magnetic information, before the birds settle into a more uniform flight.

The research could also improve our understand of the immune system. The team suspects that the sensing abilities imparted by their superparamagnetic properties could enable macrophages to better gauge their surrounding environments: distinguishing between what to fight, and what to tolerate.

“Our finding that the immune system can also sense the Earth´s magnetic field is a completely new layer in this concept of ‘immuno-sensation’, and opens the door to new research,” Lisowski says.

The research is described in Science.

The post Pigeons’ immune system can sense Earth’s magnetic field, study suggests appeared first on Physics World.

News that large language models (LLM) have made major advances in solving Erdős problems – a set of problems formulated by the renowned 20th-century mathematician Paul Erdős – has created an amalgamation of uproar and interest among mathematicians. The past month alone has seen two significant LLM-generated solutions. The first relates to prime sets, a generalization of prime numbers, and was solved after Liam Price, an amateur mathematician from the US, fed the problem statement into GPT-5.4 Pro without other information. The second came last week when the company behind ChatGPT, OpenAI, announced that it had used artificial intelligence to disprove Erdős’ planar unit distance conjecture.

LLMs have solved Erdős problems before, but the one Price chose wasn’t just any Erdős problem. It was one that human mathematicians had worked on for 60 years without success. The nature of the solution was also unusual. While previous LLM mathematical proofs used standard techniques, this one took an entirely different approach. Rather than starting from Erdős’ original probability-theory-based framing of the problem, as human mathematicians had, the LLM found an alternative route – one that led naturally, in less than a page, to a correct proof.

“Paul Erdős had a concept of ‘Proofs from The Book’, meaning that the argument is so compact and elegant that this is the proof God would’ve written down in ‘The Book’,” Jared Lichtman, a mathematician at Stanford University in the US, wrote on the social media site X after the proof was announced. “After reading the GPT5.4 proof of Erdős #1196, I would say this is a Book Proof of the result.”

The planar unit distance conjecture, meanwhile, concerns a deceptively simple question: if you have n points in a plane, how many pairs of points can be exactly one distance unit apart? Erdős thought the limit was n^{1+C/log log(n)} where C is a positive constant, but OpenAI’s model identified a higher bound. What’s more, the company claims it did so by coming up with a counter example, and in doing so, a whole family of objects that mathematicians had so far not considered. This counter example was generated using sophisticated ideas from algebraic number theory.

Some members of the mathematics community have greeted these proofs, and the advent of AI in mathematics in general, with enthusiasm. OpenAI’s announcement quotes Arul Shankar, a number theorist at the University of Toronto, Canada, as saying that the new proof “demonstrates that current AI models go beyond just helpers to human mathematicians – they are capable of having original ingenious ideas, and then carrying them out to fruition”.

Others, however, are more cautious. David Bessis, a mathematician-turned-science writer who previously worked on algebra, geometry and topology, claims that the programme of trying to automate mathematics stems from a misconception of mathematics as a logically direct process of churning out theorems, given some rules. Writing in his Substack newsletter, Bessis argues that the method used to verify AI-generated proofs, which involves a computer program called Lean, may reduce the benefit the mathematics community gains from proofs. Notably, proofs that are verifiable in Lean are not always parse-able by humans, which detracts from (and in certain cases removes) the insights researchers typically get from new proofs.

How AI is being used in mathematics…

To evaluate the merits of these arguments, it’s useful to understand how AI is currently used within mathematics research. The first strategy is the one Price used to solve Erdős #1196: directly prompting an LLM. “Large language models have proven their worth at literature search: finding similar instances of a problem, or a proof, in past literature,” notes François Charton, an AI engineer at the California-based start-up AxiomMath, which is using AI to accelerate mathematics research.

The second strategy is to use AI models trained on other types of data. According to Charton, these models are especially good at spotting “weak signals and correlations” and thereby uncovering patterns in data that might be too laborious or convoluted for humans to identify.

Both methods have shown promise for generating new results, but they are not universal – at least, not yet. “It [AI] seems to do a lot better at certain types of maths than others,” says Thomas Bloom, a mathematician at the University of Manchester, UK, who maintains a webpage that tracks solutions to Erdős problems. In particular, Bloom says that to the best of his knowledge, AI “hasn’t done anything interesting in category theory” – a field whose reputation for abstraction is only matched by its track record of bridging supposedly distinct areas of mathematics.

Another challenge is that with AI systems churning out new proofs at scale, there are simply not enough people with the skills needed to check them. A process called autoformalization could solve this problem by turning human proofs into what Bessis calls “bulletproof, machine-verifiable logical derivations” expressed in Lean or other specialized languages. At that point, AI-generated proofs could be checked automatically. The question is, could humans lose out on some knowledge in the process?

For doubters like Bessis, who refers to AI-driven autoformalization (at least as practiced by certain firms) as “AI slop”, the answer is very little. But within the AI camp, there is considerable interest in what AI brings to autoformalization. “I see autoformalization as the bridge in both directions, as important as proving itself,” Charton argues. “We can use Lean to translate between these two languages so that a Lean proof can be reverse-translated into a sketch, lemmas or natural language a human mathematician can engage with. That bidirectional translation preserves and extends mathematical knowledge at scale.”

…and how it isn’t

In the 18th century, when Leonhard Euler began arranging the logical thought processes of mathematics into theorems, definitions and proofs, mathematicians were primarily interested in solving problems with underpinnings in the physical world: questions of volume and distance, and, more generally, geometry and counting. Since then, though, mathematics has become a discipline that is at least as concerned with coming up with interesting problems as it is with solving them.

Two aspects of this change seem relevant to debates over AI’s utility. The first is that posing problems requires a broader skillset than solving them. The second is that solving posed problems sometimes requires mathematicians to invent new structures, tools or objects. Fermat’s Last Theorem, which posits that there are no three positive integers a, b, and c that satisfy the equation aⁿ + bⁿ = cⁿ for any integer value of n greater than 2, is a good example. At face value, this nearly 400-year-old theorem seems simple. However, proving it was the life’s work of a modern mathematician, Andrew Wiles, who won the Abel Prize in 2016 for developing the numerous new tools required, as well as for the proof itself.

Coming up with such tools – or indeed whole new frameworks – is a challenging and hugely creative endeavour. There are no rules as to the kinds of objects you are allowed to create, and unlike a proof (which is either correct or incorrect), there is no finality, either. If the new framework is a good one, it will crop up frequently and naturally in various branches of mathematics, and other mathematicians will incorporate it into their own work. If it isn’t, they won’t.

Currently, not even AI enthusiasts like Charton think machines are capable of such leaps. “Theory building is completely out of reach right now,” he tells Physics World. “Models, especially generative models, can provide a mathematician with interesting examples, or discover surprising relations that may bring a theoretical breakthrough, but the breakthrough still depends on the mathematician. I believe this will remain the case for some time.”

A new tool for scientists and mathematicians alike

In many areas of science, AI works in a way that is entirely distinct from human thinking. In physics, for example, machine learning algorithms are trained to analyse large amounts of data, find patterns and use them to infer underlying laws. This strategy could advance our understanding of some of the most fundamental questions in physics, but it is very different from how a human scientist would do it, and therefore perhaps more likely to be seen as a welcome new tool.

On the theorem-proving side of mathematics, the distinction between methods a human might use and those an algorithm might use is more blurred. Yet in some ways, Bloom thinks incorporating AI into mathematics could bring the field closer to other sciences. In particle physics, for example, “you don’t go in and take these individual recordings [of data]. It’s all automated,” he tells Physics World. “Until now, there has been no equivalent for maths. It takes time and attention to prove theorems, and maybe this had been a bottleneck.”

AxiomMath’s Charton agrees. “Every new math tool in history has automated something that used to be the work of a human mathematician – from the abacus all the way to symbolic algebra,” he says. “With each new tool, the role of the mathematician evolved rather than disappeared. Tasks got automated, and problems that felt impossible became trivial – but mathematicians just keep moving up the stack to the next set of questions. I see AI as the latest shift rather than a categorical break from history.”

The post AI-led solutions of Erdős problems spark debate over the future of mathematics appeared first on Physics World.

On Episode 220 of This Week In Space, Rod Pyle and Tariq Malik are joined by Dr. Pascal Lee and astrobiologist Dr. Penny Boston to mark the 50th anniversary of Viking 1's landing on Mars.

This week, we cover how Starship Flight 13 finally prepares to launch amid a legal battle over SpaceX's proposed land swap at Starbase, plus a Chinese Long March 3B rocket gets struck by lightning during liftoff, yet still delivers its satellite safely.

Then, we revisit the Viking 1 mission's groundbreaking life-detection experiments and the decades-long controversy over what they actually found. The conversation traces Mars exploration from the earliest Mariner flybys through Viking's ambiguous, still-debated results — later explained by perchlorates in the Martian soil — and the fierce scientific rivalry the findings sparked. Join us!

Download or subscribe to this show at: https://twit.tv/shows/this-week-in-space.

Get episodes ad-free with Club TWiT at https://twit.tv/clubtwit

Space news of the week

• SpaceX will now try to launch critical Starship Flight 13 test today after delays. Here's what to expect.
• 'The court has to intervene now': Indigenous groups and conservationists condemn SpaceX's South Texas land swap
• Lightning strike on Chinese rocket captured in epic launch photo
• ‘Triumph of Mariner 4’
• Slice of History - 50 Years of Mariner 9 in Orbit
• Viking Lander
• Life on Mars detected ?... Viking 1...
• Viking lander biological experiments

Model Falcon 9!

TOP TELESCOPE PICK:

Looking for a telescope to see planets and comets? We recommend the Celestron Astro Fi 102 as the top pick in our best beginner's telescope guide.

Finally, did you know you can launch your own SpaceX rocket? Model rocket maker Estes' stunning scale model of a Falcon 9 rocket that you can pick up now. The launchable model is a detailed recreation of the Falcon 9 and retails for $149.99. You can save 10% by using the code IN-COLLECTSPACE at checkout, courtesy of our partners collectSPACE.com.

About This Week In Space

This Week in Space covers the new space age. Every Friday we take a deep dive into a fascinating topic. What's happening with the new race to the moon and other planets? When will SpaceX really send people to Mars?

Join Rod Pyle and Tariq Malik from Space.com as they tackle those questions and more each week on Friday afternoons. You can subscribe today on your favorite podcatcher.

The Great Pyramid of Giza is built from something like 2.3 million stone blocks, averaging about 2.5 tons each, and it took roughly 20 years to build. Divide one number by the other and the pace reads like a typo. A block set, then the next, then the next, every few minutes, for two decades without stopping. 

It isn’t a curiosity about one heavy rock. It’s a claim about a whole human system holding a steady tempo, day after day, for longer than most modern building projects, with no engine anywhere in the chain.

The math is almost embarrassingly simple, which is part of what makes it land. A 2025 paper in npj Heritage Science lays out the standard figures: about 2.3 million blocks of around 2.5 tons each, and a required pace of roughly one block every one to three minutes across a 20-year build of long working days. Squeeze the schedule and it gets worse. The same paper notes that a 10-year build would need a block set every 30 to 90 seconds.

The blocks weren’t uniform, either. The lowest rows carry stones reaching around 10 tons or more, so “some weighing several tonnes” isn’t rhetorical. All told, laborers moved into position six and a half million tons of stone.

What that pace really describes isn’t a single feat of strength. It’s a feat of scheduling, repeated a couple of million times.

A block every few minutes only works if you stop picturing one crew doing it all. Nobody was setting a stone, catching their breath, and setting the next. The pace is an average across a whole site working at once, with many teams quarrying, hauling, shaping, and placing in parallel, so the finished total climbs at that headline speed even though no single person is moving that fast.

That reframing is what modern Egyptology has spent decades building. The old image of whip-driven slaves has been replaced with evidence of an organized, housed, and fed workforce. Egyptologists Mark Lehner and Zahi Hawass estimate the peak labor force at 20,000 to 30,000 people. Not all of them were lifting stone. Vaclav Smil’s energy-based calculation for IEEE Spectrum puts the core stone-raising and placing crew at only about 3,300 workers, far from the 100,000 men the ancient historian Herodotus reported working in three-month shifts.

Perhaps the most intriguing evidence for how the tempo held came out of the ground fairly recently. In 2013, the French Egyptologist Pierre Tallet dug up a set of papyri at Wadi al-Jarf, the oldest known papyri in the world. Among them was the journal of an official named Merer, who led a crew of some 200 men and recorded ferrying fine limestone by boat to the Great Pyramid. It’s a logistics record, dated to the 27th year of Khufu’s reign, describing the movement of stone in plain administrative terms.

Where the workers slept has also come into focus. Lehner’s team dug up long stone galleries near the pyramids and reads them as living quarters. “We now see the enigmatic rows of long galleries…as barracks housing for a rotating labor force, perhaps as large as 1,600 to 2,000 workers,” he has said, careful to call it a working hypothesis rather than settled fact. Standing in one of them, he reportedly put it more bluntly: “This is a hell of a barn!” What interested him was less the stone than the society it required. His larger argument, and it’s his interpretation rather than consensus, is about “how the pyramids built Egypt”: that organizing the workforce may have done as much to shape the state as the state did to shape the pyramid.

The build itself was probably more clever than brute. Most modern scholars think the blocks were moved with ramps and sledges rather than any single heroic lift, and the recent npj Heritage Science paper argues the internal passages themselves may have doubled as sliding ramps for counterweights. Whatever the exact method, the common thread is doing many things at once so the average holds.

What we still don’t know

The rate calculation, honestly, is the easy part. You can do it on the back of an envelope, and every serious source lands in roughly the same range.

What none of the math answers is the harder question underneath it: how a single human organization, with no machinery and no written management theory we’d recognize, held that cadence for the better part of a generation.

Consider what the pace assumes. Quarries feeding the site without running dry. Boats arriving on schedule. Ramps maintained. Thousands of workers fed, housed, rotated, and replaced as they tired or died, all without the whole tempo collapsing when any one link failed for a season. Merer’s journal gives a single thread of that operation for a few months of one year. The rest is guesswork. I find it hard to believe we’ll ever fully reconstruct the day-to-day of it, and the papyri, for all they reveal, mostly sharpen the question rather than answer it.

So the number that stops you cold turns out to be the part we understand best. Set a block every few minutes for twenty years, and the strain isn’t on any one back. It’s on the organization that had to keep that rhythm going, on schedule, in the sun, for two decades, while the thing it was building slowly blotted out the horizon.

The post The Great Pyramid was raised from an estimated 2.3 million stone blocks in roughly 20 years — which works out to setting a block, some weighing several tonnes, about once every few minutes of daylight for two decades appeared first on Space Daily.

La selección sexual servía a Darwin para explicar los rasgos más vistosos de los machos de alguans especies, como puede ser el canto, los colores vivos o la cornamenta. Características que servían para indicar a las hembras que eran ellos, y no otros, los más adecuados para garantizar el éxito de la progenie. 

Es el caso de los pájaros pergoleros, su éxito no depende tanto del canto o de sus atributos, sino de su competencia en artes decorativas. A diferencia de otras aves, los machos de esta especie no ayudan a las hembras a construir el nido, ni a incubar los huevos ni a criar a los pollos. Solo les aportan sus genes. Por eso ellas son extraordinariamente exigentes a la hora de seleccionar a su pareja. 

Sin embargo, en su caso, atraer a las hembras no implica solo tener un plumaje colorido o emitir un canto espectacular. La mayoría de las especies de pergolero construyen unas estructuras que recuerdan una glorieta o un cenador, unas estructuras equipadas con una plataforma artísticamente decorada. Y si quieren tener éxito, están obligados a que sean las mejores del vecindario.

El biólogo evolutivo Jared Diamond dijo en su día que estos pájaros son “las aves más curiosamente humanas” que existen. Razón no le falta, pues son capaces de construir unas chamizas diganas del mejor decorador de exteriores: algunas de ellas con arreglos artísticos de último diseño construidos con flores, hojas, setas… o cualquier objeto que caiga en su pico: ya puede ser un ganchillo sujeta papeles, una bolsa de patatas fritas... y hasta un billete de 50 dólares. 

Todo vale para agasajar a la hembra, y cuanto más brillante, mejor. No importa que no se trate de objetos naturales, sino que sean restos de basura desechados por una especie llamada Homo sapiens que ha expandido sus entornos urbanos e alterado el espacio natural… para algunos machos de pájaro perolero, no hay mal que por bien no venga. 

Así lo han determinado unos investigadores de la Universidad de Exeter que ha estudiado las preferencias de estas aves en varias localizaciones del norte de Qeensland, en Australia, durante la temporada de reproducción de 2023.

Descubrieron que la urbanización de esta región costera está modificando el comportamiento reproductivo de estas aves tan peculiares, especialmente en lo relativo al diseño de sus cabañas. Los machos que viven cerca de los entornos urbanos están construyendo unas glorietas más grandes, con decoraciones más vistosas y de mayor variación cromática que sus congéneres rurales, principalmente porque utilizan materiales antropogénicos, como vidrio, plástico o alambre. 

Esta fue la principal conclusión después de analizar repartidas entre un establecimiento urbano (Townsville City) y un asentamiento rural: la estación ganadera de Dreghorn, ambos en Queensland, durante la temporada de reproducción de 2023. 

Desde unas esposas hasta un billete de 50 dólares

“¡Me sorprendió mucho lo ingeniosas que pueden ser las aves a la hora de buscar elementos decorativos!”, explica a National Geographic España Caitlin Evans, coautora de este estudio, publicado en Royal Society.  Uno de los objetos más sorprendentes que encontraron fue un billete de 50 dólares, o unas esposas que algún ave había recogido del vecindario. “Una señora me contó que tiene que mantener sus plantas en el interior, ya que, de lo contrario, los pájaros le roban los frutos de los árboles. Además, tiene que vigilar sus pinzas de la ropa, pues a los pergoleros les encanta robarlas como objetos de decoración. 

cuanto más brillo, mejor 

“Creo que este comportamiento se explica por la preferencia de los machos por los objetos brillantes, llamativos y variados, características presentes en los objetos fabricados por los humanos. Piezas, además, que no se degradan tan fácilmente como los materiales naturales, por lo que es posible que los machos prefieran recolectarlos, ya que así no tienen que gastar tanta energía en sustituir los adornos cuando empiezan a deteriorarse", dice la experta. 

Lo interesante de la investigación es que no indica que las ave urbanas hubiesen cambiado sus preferencias a raíz de su mudanza desde entornos rurales, sino que parece que las aves rurales también prefieren este material de construcción y decoración, que puede parecerles más moderno, más atractivo para las hembras. Sin embargo, las aves situadas cerca de núcleos urbanos tienen mayor disponibilidad de suministros. Como si hubiera más ‘centros comerciales’ en los que adquirirlos. 

¿qué impacto tiene este hábito en el éxito reproductivo de la especie?

Y ello nos lleva a una nueva pregunta ¿Tienen estas decoraciones más suntuosas alguna consecuencia en el éxito reproductivo? “Sospechamos que las hembras podrían preferir a los machos cuando estos exhiben estos objetos, basándonos en el entusiasmo que muestran estos por incluir estas piezas artificiales en su exhibición” asegura Evans, quien afirma asimismo que en la ampliación de esta investigación pretenden dilucidar si este cambio en el comportamiento podría tener alguna consecuencia en el éxito reproductivo de estas aves. “Será interesante estudiar la respuesta conductual de las hembras cuando los machos presenten estas decoraciones artificiales durante el cortejo”, concluye. 

En resumen, estos resultados demuestran que la urbanización está transformando el cortejo del pájaro pergolero grande (Chlamydera nuchalis): los machos que residen en zonas urbanas están construyendo glorietas mucho más grandes y decoradas que sus vecinos rurales.

No es un cambio de preferencia, sino una consecuencia de un mayor acceso a los materiales de construcción. El estudio, sin embargo, plantea una pregunta aún por resolver: ¿son las hembras más receptivas a estas glorietas mejor decoradas? Todavía no lo sabemos, pero los científicos han concluido hasta qué punto la actividad humana puede modificar, aunque sea de un modo muy sutil, un concepto tan determinante en el mundo natural como es la selección sexual. 

© Foto: Caitlin Evans

Artificial intelligence may not deliver on its promise of vast economic opportunity at a price that humanity is willing to pay

In March, Anthropic, the cutting-edge artificial intelligence business that gave us the chatbot Claude, published an analysis on the impact of AI on employment, to help us assess the claim that intelligent robots were about to redefine human existence, ending demand for human labor.

Last year in May, Anthropic’s co-founder, Dario Amodei, claimed AI could wipe out half of all entry-level jobs in one to five years. Last January, he told us AI would probably become a “general labor substitute for humans”. In June he said we risk “a world where the economic trade-off dial is stuck on the hypergrowth, hyper-inequality setting”.

Continue reading...

© Photograph: DragonImages/Getty Images

© Photograph: DragonImages/Getty Images

© Photograph: DragonImages/Getty Images

Fixing a problem inside one of the world's most sophisticated scientific observatories might seem like the kind of challenge that demands a multimillion-dollar upgrade or revolutionary new technology.

But for scientists working on the Laser Interferometer Gravitational-Wave Observatory (LIGO), which listens for ripples in spacetime generated by cosmic collisions like merging black holes, the solution to a mild but persistent engineering challenge turns out to be as simple as an off-the-shelf camera.

By pairing commercially available thermal imaging cameras with computer models, a team led by Jonathan Richardson at the University of California, Riverside, has developed a technique that corrects tiny, heat-induced distortions in the observatory's mirrors — an elusive flaw that scientists say currently limits how far into deep space the facility can look.

"It doesn't require any new technology development, which is almost unheard of for solving a LIGO instrumentation problem," Richardson said in a statement.

Once incorporated into LIGO's upcoming upgrade, Richardson and his team estimate the fix would extend the observatory's reach by roughly 33 million light-years.

That gain might sound like a drop in the ocean against the unimaginably vast scale of the universe, but because space expands in three dimensions, pushing a detector's reach even slightly opens up an exponentially larger window of space. Being able to look further into the universe will allow astronomers to "hear" many more cosmic ripples, in turn increasing the potential for discovering the universe’s most violent collisions that lie beyond LIGO's reach today.

LIGO detects these cosmic ripples, known as gravitational waves, using twin L-shaped facilities in the U.S. — in the states of Washington and Louisiana. Inside each detector, a laser beam shoots down two 2.5-mile-long (4-kilometer-long) tunnels, bouncing off pristine mirrors at each end. When a gravitational wave passes through Earth, it subtly stretches one tunnel and squeezes the other. That microscopic shift alters the laser beams ever so slightly, producing a tiny flicker of light that alerts scientists to a distant cosmic event.

Because these cosmic signals are inconceivably small, preserving every single photon is crucial. To accomplish this, LIGO relies on mirrors polished to reflect 99.9999%t of the laser light that strikes them, ranking them among the purest optical components ever built.

Yet even these near-perfect mirrors have had one unavoidable flaw. The mirrors still absorb a tiny fraction of that intense laser light. That energy turns into heat, warping the mirror's surface by just a few nanometers, enough to distort the laser beam and reduce the observatory's overall sensitivity.

Physicists already knew they could counteract these distortions by applying targeted heat to the back of the mirrors. The difficult part was measuring the distortions accurately enough such that the correcting heat could be applied with exact precision.

The new technique uses infrared thermal images and existing computer models to reconstruct a map of distortions across the mirror's surface.

"You can think of it like taking an infrared picture of a car engine," Richardson said in the statement. "An engineer can look at the temperature pattern on the outside and infer what's happening inside the engine. We're doing the same thing with LIGO's mirrors."

And the technique isn't just a fix for LIGO. It is also expected to become part of the foundational design for Cosmic Explorer, a proposed next-generation U.S. gravitational-wave observatory targeted for the mid-2030s.

With 25-mile-long (40-km-long) arms — 10 times larger than LIGO's — Cosmic Explorer is already designed to detect gravitational wave events far beyond the reach of today's observatories. This new technique will only supercharge its ultimate reach.

"The goal for the next generation of gravitational-wave detectors is to achieve about 10 times the sensitivity of today's instruments," Richardson said in the statement. "One of the key obstacles to achieving that is reducing the fundamental quantum mechanical noise that limits the precision of the measurements."

The technique is described in a paper published July 16 in Classical and Quantum Gravity.

The race to build larger and more powerful quantum computers has mostly focused on engineering: adding more qubits, reducing noise, and keeping fragile quantum states alive. However, a new theoretical study suggests that another, largely overlooked problem may emerge as these machines grow. 

Researchers at the Helmholtz-Zentrum Dresden-Rossendorf (HZDR) have found that a strange quantum phenomenon known as the quantum Zeno effect could cause certain quantum computers to slow dramatically—or in the worst case, nearly stop computing altogether.

“For quite generic conditions, we find that the quantum Zeno effect poses strong limitations on the performance” of quantum systems, the study authors note.

Rather than exposing a flaw in today’s devices, this research work identifies a fundamental scaling challenge that designers may have to confront as quantum computers become larger. 

In their new study, the HZDR team says the effect could become especially important for adiabatic quantum computers, a class of machines widely studied for solving optimization problems, searching massive datasets, and simulating molecules.

When the environment won’t stop “looking” at a quantum computer

The quantum Zeno effect is one of quantum mechanics’ most counterintuitive ideas. It predicts that repeatedly measuring a quantum system can prevent it from changing its state, effectively freezing its evolution. 

The concept has been studied experimentally for decades. Earlier research works explored this phenomenon mainly in isolated laboratory systems or as a way to deliberately control quantum states. For example, a theoretical work showed that continuous measurements can trap quantum systems within specific states. 

The HZDR researchers asked a different question—what if a quantum computer experiences this effect unintentionally?

Adiabatic quantum computers work differently from gate-based quantum computers. Their “algorithms are considered robust and can be executed by quantum computers largely independently of the hardware that is used,” Ralf Schützhold, one of the study authors and the director of the Institute of Theoretical Physics at HZDR, said.

Instead of performing a sequence of quantum logic operations, they gradually reshape an energy landscape while keeping qubits in their lowest-energy, or ground, state. If this transformation is slow enough, the ground state naturally evolves into the solution to the problem being solved. 

To investigate, they built a theoretical model of an adiabatic quantum computer interacting with a realistic environment. In practice, no quantum processor is perfectly isolated. Even inside refrigerators cooled to temperatures just above absolute zero (−273.15°C), tiny amounts of heat, electromagnetic radiation, or other external influences still interact with qubits. 

These interactions cause decoherence, gradually disrupting the delicate quantum properties such as superposition and entanglement—that give quantum computers their power. In effect, the surrounding environment behaves like an unintended detector that continually checks the qubits’ state, even though nobody is intentionally measuring them.

More qubits could make the problem worse

The team analyzed the adiabatic version of Grover’s search algorithm, a famous quantum algorithm that searches unsorted databases much faster than classical methods. 

In adiabatic computing, the algorithm starts with an easy-to-prepare quantum state and then slowly transforms the system’s governing energy landscape, known as the Hamiltonian, until the ground state represents the desired solution. If the transformation is sufficiently slow, the qubits naturally remain in the ground state throughout the computation.

Their calculations revealed that scaling creates an unexpected vulnerability. As the problem size grows, the minimum energy gap separating the ground state from higher-energy states can become increasingly small. 

As these energy differences shrink, the computation must proceed more slowly to remain adiabatic. Since the computation takes longer while environmental disturbances continue acting at roughly the same rate, the quantum Zeno effect becomes increasingly likely. 

“Each disturbance acts like an unwanted measurement, slowing down the system’s evolution. In the worst case, a calculation could even freeze completely,” Gernot Schaller, one of the study authors and the head of Quantum Technologies at HZDR’s Institute of Theoretical Physics, said.

Much like a cake rises properly only if the oven remains closed, constantly opening the door to check on it interrupts the baking process, causing the cake to stop rising or even collapse. In the same way, repeated environmental disturbances keep interrupting the natural evolution of the quantum state before it can reach the desired endpoint.

More importantly, the study does not suggest that all quantum computers face the same limitation. The work focuses on adiabatic quantum algorithms and related approaches such as quantum annealing, where the computation relies on gradually crossing small energy gaps. 

A roadmap for building more resilient quantum computers

The study is theoretical rather than experimental, so the findings still need to be explored in real quantum hardware. 

However, the researchers say the problem can be mitigated by improving shielding against heat and electromagnetic interference and by using techniques such as spin echo, which applies carefully timed pulses to reduce interactions between qubits and their environment.

The findings also suggest that scaling quantum computers will require more than simply adding qubits. Future systems may need better protection from environmental noise and smarter control techniques to reduce the risk of the quantum Zeno effect becoming a bottleneck. 

The study is published in the New Journal of Physics.

Giovanni Kanter goes to at least one art museum whenever he visits a new city. That’s on top of his regular exploration of galleries in New York, where the 25-year-old ICU nurse lives and studies. He typically discovers new artists and gallery openings through word of mouth, though he has recently started curating his social media feed to serve relevant content (he’s currently in what he terms a “surrealism phase”).

He was raised on visual art; he's the son of painter Marianne Gargour and nephew of mixed-media artist Heather Hutchison and painter-printmaker Mark Thomas Kanter. His “exposure-oriented education” featured childhood visits to galleries and shows in and around Dallas, where he grew up, and in France, where he spent time with his grandparents.

But Kanter is an exception. Many other young people have a less immersive artistic education, perhaps shaped by occasional school trips or family visits to museums. If they haven’t internalized an instinct for what Kanter calls “moving through the art and absorbing it,” the museum experience is central to how they learn about and understand art.

Read full article

Comments

In the final weeks before Romania’s presidential vote in November 2024, TikTok accounts that had previously spent years posting about manicures or fashion suddenly started promoting a little-known politician named Calin Georgescu.

His posts, espousing hard-line views on immigration and anti-Semitic tropes, were viewed 120 million times before the vote. Georgescu, who had previously been polling in the single digits, stunned observers by winning the election’s first round with 23 percent of the vote.

In the Netherlands, Adriana Iamnitchi was watching closely. As chair of computational social sciences at Maastricht University, she leads the research into disinformation campaigns. She wanted to know how pro-Georgescu content was being monetized on TikTok through hidden influencer marketing and livestreamed political content.

Read full article

Comments

© Getty Images | Photographer is my life

Picture someone leaving their apartment on a hillside above the water, coffee in hand, stepping onto a train. Four minutes later they get off in a different continent. No passport, no fuss, no sense of having crossed anything at all. They do it again on the way home.

For a lot of people in Istanbul, this is just the shape of an ordinary working day.

It sounds like a geography quiz answer, and it is one. You can hop from Asia to Europe before your coffee cools.

What does that tell us about the lines we draw on maps?

The city that broke the map

Istanbul is by far the largest city in the world shared by two continents, Europe and Asia. There are other transcontinental cities, including Atyrau and Suez, but none come close to Istanbul’s scale. This is a city of roughly 15 million people, and the continental boundary runs right through it.

The split is not even. Istanbul’s population reached 15,701,602 at the end of 2024, according to the Turkish Statistical Institute, and the two sides carry very different loads. Roughly 65% of residents live on the European side, and 35% on the Asian side. Most of the shops and offices sit in Europe, and many jobs remain concentrated there. Much of the more affordable housing has grown up in Asia. You can probably already see where the daily traffic comes from.

What the Bosphorus actually is

The thing doing the dividing is the Bosphorus, a narrow channel of water about 31 km long that links the Black Sea to the Sea of Marmara and runs entirely inside Turkey. It is water, but it is also a border. This strip of sea is treated as part of the accepted line between Europe and Asia, which is why crossing it counts as changing continents.

Until the first Bosphorus bridge opened in 1973, travelling by boat was the only direct way across, and that history still shapes the city’s rhythm. Ferries are woven into daily life in a way that’s hard to picture from the outside.

As Adil Bali, an expert on the history of Istanbul’s ferries, puts it, “It is one of the few cities in the world that can be crossed by sea, so the ferries are indispensable here.” That’s a romantic framing, but the point holds: those city ferries still carry at least 40 million passengers a year.

The commute that crosses a continent

Ferries are the old way. The Marmaray rail line runs through a tunnel beneath the strait, and its trains make the crossing in about four minutes. Not a scenic voyage between continents. Just four minutes of a normal train ride, mostly in the dark.

The scale is what gets me. Marmaray carries between 600,000 and more than 700,000 passengers a day, and in 2024 it set a single-day record of 743,596 riders. A report from the Council of Europe Development Bank describes how the tunnel changed the trip, lifting capacity to up to 75,000 passengers per hour in each direction, up from around 10,000 before. 

So the intercontinental commute is not a novelty. It’s the plain logistics of getting to work for hundreds of thousands of people, most of whom, I’d guess, think about it about as much as anyone thinks about their morning train, which is to say not at all.

Why this matters beyond the geography trivia

Here is what I keep circling back to. If millions of people can cross between Europe and Asia without noticing, maybe the noticing was always the strange part, not the crossing.

Geographers have made this point for a long time: the boundary between Europe and Asia is debated, not a simple fact of geology. There is no ocean between them, no crack in the Earth’s crust, nothing underfoot that says “you are now leaving one continent.” It’s a single connected landmass, Eurasia, with a line drawn across it by historical and geographic convention rather than by a clean geological break. The Bosphorus got picked as part of that line, which is convenient and tidy and completely human-made.

Istanbul is where that idea meets a commuter train timetable. The city doesn’t disprove continents so much as show how soft the concept is once you stand on top of it. A person eating breakfast in Asia and clocking in in Europe an hour later isn’t doing anything out of the ordinary. They’re just moving through a city that happens to sit on a line somebody else decided mattered.

The continents are an agreement, not a discovery. Istanbul is the place where you can ride straight through the seam and feel nothing change, because there was never really anything there to change.

The post Istanbul is the only major city on Earth that sits in two continents at once. The Bosphorus strait cuts straight through it, so one half of this metropolis of some 15 million stands in Europe and the other in Asia — meaning millions of residents cross between two continents on their daily commute, hopping from Asia to Europe and back just to get to work appeared first on Space Daily.

Nothing Ear (3a): Two-minute review

Sometimes, reviews are written with a heavy heart, and this is exactly that. I've championed Nothing's affordable "(a)" earbuds for well over two years — with the updated pink set for 2026 appearing to be right up my strasse. But all is not as it first appears.

The issues don't concern their cost. Like before, they come in at just under $100 and are therefore a candidate for our best budget earbuds guide.

Bizarrely, for the review star rating, I also have very few bones to pick with audio quality, which is still best pals with LDAC and now with Bluetooth 6.0 for a stronger connection. I recommend you tweak the sound profile from the box, mind — it's overly bassy by default — but that's hardly unheard of for earbuds at this level, and it can be rectified with a little effort.

It would have been lovely if Nothing had allowed its special sonic tests (for personalised EQ profile curation done by Nothing's software, rather than you tweaking sliders or blobs yourself) to be made available here, but the company has clearly decided that its three-minute hearing test is flagship only — although there is a new "Advanced" 8-band EQ tab with "Q factor" (more on this later).

What of the comfort levels and overall build quality? Here you get everything you'd expect of Nothing: rock-solid, colorful, and with just a couple of new flourishes to enjoy. Battery life has taken a knock, but that might possibly be down to having to work a bigger driver.

So what's the main problem? Actually there are two: Call Recording and Audio Snapshot (sometimes referred to as Audio Capture in Nothing's extensive marketing). Nobody asked for these features and, honestly, they're not the USP I wanted — even as a journalist who interviews people regularly.

None of the best noise-cancelling earbuds we recommend come with these features, but that's hardly a problem in itself. Such tech exists in the world, obviously; I point you to the Viaim RecDot, for example. But the Viaim's buds publicise their raison d'etre; there's a visible "record" button on the case, a light for in-person meetings, and a business-class $249.99 / £185 price tag. That's not the case with Nothing's version, where a surreptitious squeeze of the stem kicks off audio capture of your current call (or whatsapp voice note) in a set of buds that otherwise seem to scream "I'm a bit of fun!"

As with the RecDot, Nothing offers tech to transcribe your recordings (and not just on its own phones, either) — to a point. You get three months of its Pro Transcription service for free, with 120 minutes to use each month. And after the trial? Nothing tells me official pricing hasn't yet been finalised, but will be decided in August. If it was hard to recommend such a package before, it's even harder now when one doesn't know how much it will cost.

I can't help but feel that Nothing has taken a wrong turn here. I'd much rather have seen fan-favorite features such as personalised sound profiles and/or wireless charging rather than in-call recording. Some TechRadar team members have suggested I make like Elsa and just Let It Go, but in a year when consumer tech already feels uncomfortably intrusive (Meta, I'm looking at you with those smart glasses), adding call and voice note recording feels like a step too far. It's important to question where these features are taking us.

At $99 / £99, the price is definitely right. I'm just not sure the feature set is…

Nothing Ear (3a) review: Price & release date

• Released on July 7, 2026
• Priced $99 / £99 / around AU$192, same as the 2024 Ear (a)

The Nothing Ear (3a) cost the same as their predecessor, priced at $99 / £99.

Although the name might suggest otherwise, these are the direct successors to the excellent, five-star Nothing Ear (a) from 2024 – there was never a Nothing Ear (1a) or Nothing Ear (2a). They also slot nicely alongside the still-current but more expensive Nothing Ear (3), with the shared 3 denoting that they're part of the same generation.

The Ear (a) are some of the best budget earbuds I've ever had the pleasure of testing, and I'd looked forward to their update ever since word was out that it was coming.

Key rivals to the Nothing Ear (3a) include the Apple AirPods 4, which offer gesture control alongside a few other iPhone-exclusive features, or, for a little less, the five-star Cambridge Audio Melomania A100. If you're tempted by open earbuds instead, the Huawei FreeArc or even clip-style SoundPEATS Clip1 can both be picked up for less than the Nothing Ear (3a) right now, making them excellent-value alternatives.

Nothing Ear (3a) review: Specs

Drivers	12mm custom driver
Active noise cancellation	Yes
Battery life	Up to 6 hours (ANC off); up to 4.5 hours (ANC on); up to 25 hours with case (ANC off) or up to 18 hours (ANC on)
Weight	4.53g per earbud
Connectivity	Bluetooth 6.0 with LDAC, USB-C
Frequency range	20Hz – 40kHz
Waterproofing	IP54
Other features	Wideband Active Noise Cancellation up to 45dB, 3 MEMs mics per earbuds with AI ENC, Nothing X App support, Custom 8-band EQ, Find My Earbuds, Low Lag mode, auto-off, Dual Connection, extra ear-tip, Audio Snapshot and in-call recording

Nothing Ear (3a) review: Features

• Bluetooth 6.0 (15m range) with LDAC support
• Improved ANC (cancellation up to 45 dB, effective to 5 kHz)
• Pinch stems for Call Recording/Audio Snapshot, no Nothing phone required

It would be easy to say, "Look, you don't like the cutesy branding of Nothing's new 'Audio Snapshot' offering? Don't use it — there's still plenty to like for $99!" but let's dig in to see why the take is not so simple.

Firstly, we should get granular on what Audio Snapshot and Call Recording do, and what they cannot do. I mentioned above that as a journalist, I regularly record interviews, often over Zoom, Google Meet or even Microsoft Teams. When set up as an interview, my subjects usually take it as a given that I will be recording — and where I reside in the UK, it is legal to record calls, even without consent from the other party. I still always ask.

In the US, however, the laws on recording phone calls are different, varying from state to state. For calls crossing state lines, the stricter 'two party consent' law (as per the Federal Wiretap Act) applies, where both parties on a call must consent to a recording.

So what is Nothing doing about it? In its reviewers' guide, the company says it "aligns with internationally recognised privacy standards ISO 27001, ISO 27701 and EN 18031". To this end, during setup, the Nothing X app also displays a short tutorial reminding users to obtain permission before recording calls — as shown below.

Interestingly, on a traditional phone call made in the UK, the "prompt" Nothing promises will be played to everyone on the call is a simple beep. On Google Meet, a female voice announces "This call will be recorded in 3, 2, 1", presumably so anyone not OK with that can scream "Oh no you don't!". From smartphone to smartphone, there's no vocal.

Thanks to the Nothing Ear (3a)'s 32MB of onboard storage, you can start recording with just a pinch of the stems, storing up to two hours of audio on the earbuds themselves. You can also record and transcribe voice notes from apps such as Instagram or WhatsApp — and from August, Nothing says it will add summaries of call recordings, speaker identification and the ability to create calendar reminders based on what was discussed.

What these earbuds can't do is record conversations that take place around you, or indeed your own voice. So, thankfully, there's no scope for eavesdropping around the water cooler.

However, my thoughts on this are clear— I'm not wholly OK with it. It puts the onus on buyers to adhere to laws, and opens the door for flouting rules of consent. I tried muting my mics to see whether the beep could be silenced, but it still plays for the person on the other end of the call to signal that recording has begun. Nevertheless, I think I've made my concerns around this feature clear.

Elsewhere, you get Bluetooth v6.0 and LDAC support for hi-res audio (the Sony-developed codec that lets you stream high-resolution audio up to 32-bit/96kHz over Bluetooth at up to 990kbps, if your device supports it and the file is up to scratch). While the predecessor did also include the latter, it remains a valuable inclusion here.

There's also onboard spatial audio, albeit of the "static" variety rather than head-tracked spatial audio that keeps the sound anchored to your source device as you move your head. It adds a little extra depth and height to the presentation, but for me, the effect is decidedly light-touch compared to what you get from pricier earbuds.

Note that you don't get the Ear (3)'s splendid personalized hearing tests, or the useful "walkie-talkie" Super Mic. That said, there's in-ear detection (to pause music when you remove the buds and resume it when they're in), a low-lag toggle for gaming, seamless pairing, multipoint to two devices, an ear-tip fit test and a Find My Earbuds feature, which emits a rattlesnake-style sound from whichever bud you're trying to locate, to help you find it.

Nothing's Transparency mode is decent. It's a little warm, but it still delivers a more three-dimensional presentation than many buds at this level. I could hear the placement of the TV, voices in my home and a dog outside, rather than all of these sounds appearing tinny, omnipresent and seemingly on top of one another.

When deploying ANC (rather than "Transparency" or "Off"), you can once again select from High, Mid, Low, and Adaptive noise-nixing profiles. High is decent: Nothing tells me performance is up to 45dB (effective to 5 kHz), which equates to a 17.1% improvement compared to the Ear (a), thanks to a software upgrade and improved architecture regarding where the mic sits.

The issue in 2026 is the existence of the AirPods Pro 3 and Bose QuietComfort Ultra Earbuds (Gen 2). Although it would be unfair to compare like for like, owing to the chasm in pricing, you can get better if you're willing to pay for it. That said, the ANC here is no slouch, giving a fairly clean, largely uninterrupted and issue-free neutral background for my music.

If you have a Nothing Phone, the Nothing X app and the ChatGPT app, then you can once again you can also access ChatGPT access via the Ear (3a)'s stem; simply squeeze and talk.

The new EQ tab is now an eight-band offering (up from three in the older buds), as well as the more basic audio splodge Nothing made its own — and which I rather like since it reminds me of Nura True Pro's visual depictions, rather than a mixing desk. You also get five presets in the Advanced tab, and four in the Simple tab — but you can create your custom EQ using whichever system you prefer.

And I think you'll need to get familiar with these, as I'm about to explain…

• Features score: 4 / 5

Nothing Ear (3a) review: Sound quality

• LDAC support represents value
• Fun, V-shaped sound — if a little quiet
• A little bass-heavy from the box, but you can fix it

As was the case with their siblings, it would be wrong of me to knock Nothing Ear (3a) in the sonic department for this money. The sound quality is easily worthy of the outlay.

That said, there are a few points to bring to your attention.

Those with a Sony or Samsung smartphone (I used a Sony Xperia) will find much to celebrate in LDAC music streaming. In My Chemical Romance's The Ghost of You, guitars snake from my left to right ear in a cohesive mix — with Gerard Way's vocal central, petulant and every bit as emotive as I remember it in the mid-noughties.

Even when I stream Apple Music songs from my iPhone, the Nothing Ear (3a) offer ample width as my playlist continues to Thank You For the Venom and the intro guitars find a spacious home in my left ear. There's room around every stem in the mix to shine, but they're all held properly in check, even in this fast-paced and challenging song.

It's a little bass-heavy at times, though. Fleetwood Mac's The Chain is a good test of a regimented low-end, and alongside the kick-drums, I did lose just a little of Stevie Nicks, Christine McVie, and Lindsey Buckingham in that glorious upper harmony through the chorus. It seems to be Nothing's modus operandi to offer a bassy listen generally, and while I did find that the new Advanced EQ tab helped, I think the enhanced low-end went a little far here.

Nothing offers a "Q Factor" in addition to what you've just tweaked in the 8-band tab, to control the width or sharpness of your adjustment. So, imagine it as a "quality factor" (because that is what it stands for), where the number you select between 0.1 and 10 determines how wide or narrow of a frequency range is boosted or cut — all of this specific to your chosen frequency. Selecting a low Q Factor will generate a wider, more gentle curve to your adjustment, where a high Q Factor will make the change narrower, steeper and more effective around that specific frequency augmentation (or reduction).

I eased off the bass and selected a Q Factor of 3.2, but you must of course play with these things yourself — and sadly, Nothing hasn't included its hearing tests here, to check how perceptive your ears are within this particular tuning.

For dynamics and nuance, the Nothing Ear (3a) are perhaps a shade less engaging and arresting than I'm used to, owing to their volume. Once criticism often levelled at Nothing's budget earbuds is they don't go especially loud, and I frequently found myself listening above 50% volume. Also, since battery life is measured at that volume, it did mean the quoted "4.5-hour battery life with ANC deployed" claims although bang on, were still disappointing.

• Sound quality score: 4 / 5

Nothing Ear (3a) review: Design

• Curvier case
• That 1mm bigger driver is sucks up power
• And the battery life isn't great

Hold the Nothing Ear (3a) earpieces and case next to their older Nothing Ear (a) and you'll certainly struggle in a game of spot the difference. There's the color, obviously — also available in black, white and a slightly different yellow to the hue I love in this older set — and the 50% curvier case. There's also a trio of LEDs on the case that deliver a total of six status indications for battery status, pairing status, charging levels and so on.

For further tweaks, we have to look deeper.

The driver is now 12mm rather than 11mm and features an "aerospace-grade diaphragm". You also get an extra set of ear-tips in the box, four in total, and although the XS will appeal to owners of small ears, the actual earpieces here aren't tiny, despite being lighter in weight than Ear (a), at 4.53g versus 4.8g. The actual design of the tips has been finessed, too: they're a little thicker and more flared across the outer wall. I found the buds very secure, if a little big to accommodate comfortably over my small-ish ears over the course of long listening sessions.

I've said it before now and I'll say it again: I think we've reached a point of diminishing — perhaps even negative — returns when it comes to ever-larger earbud drivers. Every new release seems to tout a bigger driver without adding weight, as if bigger automatically means better. But where does it end? 13mm drivers? 15mm drivers? To my ears, the Ear (3a)'s 1mm larger driver hasn't negatively impacted the sound, but I don wonder if it's contributed to sthe shorter battery life. In my briefing, I was told the buds themselves would last 10 hours — but that turned out not to be the case.

While the battery in each earpiece — along with the larger driver and onboard 16MB flash drive — works hard to power the buds, it's only good for up to 6 hours without ANC, or 4.5 hours with ANC turned on. And remember, I had to pump the volume up higher than 50%, which the level at which those measurements are typically taken. With the case, you'll get up to 25 hours with ANC off, or up to 18 hours with it on.

None of these stats is much to write home about. Take Nothing's own Ear (a), which offered up to 9.5 hours of playback on a single charge with ANC off, and about 5.5 hours with ANC on. Now, consider the very inexpensive Cambridge Melomania A100, which offer 11 hours per charge or up to 39 hours with the case, along with aptX Lossless. I know which I'd rather take on long road-trips…

Again, Nothing's pinch-to-control stems return here, and while volume controls aren't enabled by default, you can assign them in the Nothing X app — provided you're happy to sacrifice another shortcut in the process.

While I'm here, a quick note on call-handling. It's absolutely clear and clean as a whistle, thanks to the 3 MEMS mics per bud.

• Design score: 4 / 5

Nothing Ear (3a) review: Value

• Still very good-looking earbuds
• Decent ANC; not so great for stamina
• LDAC adds value, but call-recording (for me, at least) is a no

Before now, I've always championed Nothing's "(a)" buds — and while the basics in the (3a) are fine, there are a few issues that are hard to ignore.

The Call Recording feature is top of the list here. More than any technical limitation, it's the privacy implications that concern me. Much as seeing someone wearing smart glasses now makes me wonder whether I'm being recorded, these earbuds leave me asking why someone might want to capture our conversations. I appreciate that some readers may see this as an overreaction, but it's a genuine concern for me, informed by my own experiences as a woman and former dancer who had to deal with overly familiar fans.

I think the audio quality here is perfectly acceptable for $99 / £99, but the stamina and feature set can be bettered. Even in the open earbuds market, I'd happily direct you to the cheaper Huawei FreeArc or SoundPEATS Clip1 over these earbuds.

• Value score: 4 / 5

Should I buy the Nothing Ear (3a)?

Section	Notes	Score
Features	I can't get behind Call Recording — although the ANC is good enough	4/5
Sound quality	An fun listen that's easy to tweak if the bass proves a bit much. Just a tad quiet overall	4/5
Design	Good-looking buds let down by disappointing stamina	4/5
Value	It's hard for me to write it, but the smart money isn't always with Nothing	4/5

Nothing Ear (3a) scores

Solid generally, but with battery concerns and a potentially problematic call recording 'feature'

Features

4

Sound Quality

4

Design

4

Battery Life

2.5

Noise Cancellation

4.5

Value

4

Overall

4

012345

Group 1 Data

Product	Features ()	Sound Quality ()	Design ()	Battery Life ()	Noise Cancellation ()	Value ()	Overall ()
Nothing Ear (3a)	4	4	4	2.5	4.5	4	4

Buy them if...

You want buds that say "look at me"
The pink colorway is super-striking and fun, especially against their black/transparent stems.

You have a Sony or Samsung phone
LDAC support for $99 / £99? That's good stuff.

You want to record your phone calls
These buds do offer the option to record and transcribe up to 2 hours of audio from calls, songs or voice notes — and you don't need a Nothing source device to do it.

Don't buy them if...

You want extra-special spatial audio
Nothing adds its "static" spatial audio profile here, for a slightly more immersive sound profile to your music. For me, though, it's a little bit lightweight.

You have bijou ears
This design just about works for my smaller ears with the small ear-tips (there's even an XS option included now), but if your concha or ear-well is particularly diminutive, then you may still find these earpieces a little big.

The idea of covert surveillance bothers you
As a journalist, I have no issue with recording when everyone has consented. My concern is Nothing's playful marketing if in-call recording as a lifestyle feature, despite the obvious privacy implications.View Deal

Nothing Ear (3a) review: Also consider

	Nothing Ear (3a)	Cambridge Audio Melomania A100	Huawei FreeArc
Price	$99 / £99 / Approx. AU$192	$149 / £119 / AU$TBC (but around AU$249) at launch, discounted now	£99 (roughly $130, AU$200) but discounted now
Drivers	12m custom	10mm Neodymium dynamic	17mm x 12mm (open earbuds)
Active noise cancellation	Yes	Yes	No
Quoted battery life	Buds: 6 hrs (without ANC) or 4.5 hrs with ANC; Total with case: 18 hrs with ANC (or 25 hrs without ANC)	11 hours per charge; up to 39 hours with the case	7 hours (earbuds), 23 hours (total)
Weight	4.53g	4.7g	8.9g
Connectivity	Bluetooth 6.0, LDAC, USB-C (no wireless charging)	Bluetooth 5.4 with LDAC and aptX Lossless	Bluetooth 5.2
Frequency range	20 Hz – 40 kHz	Not stated	Not stated
Waterproofing	Yes, IP54	Yes, IPX5	Yes, IP57
Other features	Audio Snapshot and Call Recording; auto-off, static spatial audio, ChatGPT with a Nothing Phone	7-band EQ, ANC, DynamEQ for optimised tuning, wireless charging, 3 mics per bud	Huawei AI Life app, 10-band EQ, audio presets, find your earbuds

Huawei FreeArc
Up for taking a punt on affordable and great-sounding open earbuds? The FreeArc are available at a generous discount right now — but in all honesty, we loved them even at full price.
Read more on this in our full Huawei FreeArc review

Cambridge Audio Melomania A100
These June 2025-release earbuds didn't get the fanfare they deserved, unless you count our glowing five-star review, of course. They don't look as flamboyant — but oh, do they sound good, and the ANC is great too. Did I mention they've been discounted recently? Highly recommended for their current price.
Learn more in our in-depth Cambridge Melomania A100 review

How I tested the Nothing Ear (3a)

• Tested for two weeks, listened against the Ear (a), Bose QuietComfort Ultra Earbuds (Gen 2), Technics EAH-AZ100 and AirPods 4
• Listened in the office, walking along a beach, on a train into London, into Central London and at home
• Listened to Tidal Masters, Apple Music Lossless tracks and Spotify on an iPhone 15 Pro, MacBook Pro and Sony Xperia smartphone

The Nothing Ear (3a) became my primary musical companions for well over two weeks — after a 48-hour run-in period.

They accompanied me to work (walking brusquely to our Paddington office and on the London Underground to various press events) and even walking on a beach at sunset.

To assess comfort levels, ANC and battery life claims, I followed TechRadar's meticulous methodology testing and found much of note — including slightly disappointing stamina.

To check the audio quality across the frequencies, I listened for several hours to my own reference playlists on Apple Music and Tidal, alongside TechRadar's reference tracks, plus podcasts and movies. Additionally, I wore them to watch YouTube tutorials from my MacBook Pro.

I’ve been testing audio products for well over seven years, full-time. As a dancer, aerialist and musical theater performer in another life, sound quality, fit, and user experience have always been a priority for me, but having heard how wonderful ANC can be when done well, I know what I'm listening to here, too.

Read more about how we test earbuds at TechRadar

• First reviewed: July 2026