Did Prime Minister Zelensky make huge mistake to dismiss the effective Defense Minister Fedorov?
Thursday, July 16, 2026
Ukraine Senior Air Force Commander Quits, Calls Defense Minister Fedorov’s Dismissal ‘Great Evil’
Sodium-ion batteries are entering mass production in China
Good news! Will China corner the global battery market?
"China has over 36.9 million ... passenger ... EVs and more than 420 million electric two-wheelers (motorcycles, scooters, and e-bikes) currently in use." (Google search)
"China’s CATL set to begin large-scale output this year in a push that researchers say could offer a cheaper, safer, and more abundant alternative to lithium for EVs and grid storage."
Beyond lithium: how sodium-ion batteries could change the world "Batteries based on sodium ions are entering mass production. Some researchers say they could ultimately be a cheaper, safer alternative to lithium in electric cars and other energy applications."
Sodium-ion batteries made by CATL in China are used in electric cars (left) and energy-storage banks for electricity grids (right).
The number of unvaccinated children worldwide declines further
Good news! The comprehensive vaccination of children is a huge humanitarian success story!
Given "There are approximately 2.4 billion children (people under the age of 18) worldwide" (Google search) this is a great success!
"... Key takeaways: ~13.5 million children remained “zero-dose” in 2025, meaning they missed all routine vaccinations. That number is down ~750,000 from the previous year, but still well above pre-pandemic levels. ..."
"... In 2025, 90 per cent of infants globally – or nearly 116 million – received at least one dose of a diphtheria, tetanus and pertussis (DTP) vaccine, and 85 per cent – or 110 million – completed the full three-dose series ..."
Global childhood immunization coverage inches forward despite conflict and hesitancy – UNICEF, WHO "Zero-dose children fell by nearly 750,000 in the past year, but drop-out numbers high and stagnant, increasing risk of disease outbreaks"
France’s parliament adopted an intensely debated right-to-die bill yesterday
Good news! Food for thought!
What e.g. the next day a cure is published for a previously incurable disease?
"France’s parliament adopted an intensely debated right-to-die bill yesterday, supporting assisted dying for adults with incurable illnesses; the bill is slated to undergo review by France’s Constitutional Council before it can take effect."
"... The legislation will, under strict conditions, allow a person to request a lethal substance. The substance could be self-administered or, if the person is physically unable to do so, administered by a doctor or nurse. ..."
France’s lower house of parliament adopts final text of landmark right-to-die law "France's National Assembly on Wednesday adopted a bill that will create a legal right to assisted dying for adults with incurable illnesses, capping years of intense ethical and political debate. French President Emmanuel Macron, who promised such a law during his re-election campaign, hailed lawmakers' "respectful debate" on the issue following the vote."
Violent attacks by Israeli extremists in West Bank fall 25% since March
News that many Western media will not report!
"... In addition, the Judea and Samaria District Police have carried out arrests in several cases, including the detention of six Israelis accused of setting fire to a Palestinian home and the arrest of four Israelis who attacked a CNN news crew last week in the village of Sinjil, in the Binyamin region of the West Bank. ..."
PrismML — Announcing Bonsai 27B: The First 27B-Class Model to Run on a Phone
Bonsai, not a bad name! How did they compress the large Qwen model?
Is it really necessary to have such models run on a phone? Do the benefits outweigh the disadvantages?
"Today, we're announcing Bonsai 27B, based on Qwen3.6 27B, the new multimodal flagship of the Bonsai family and the first model of its capability class to run on a phone. ..."
A Top Mamdani NYC Official Tried to Meet with Islamic Republic’s ambassador to the U.N.
The NYC mayor has ambitions beyond the city limits! 😊 What was Zamdani thinking?
It was not reported whether the female top official would have worn a Burka or head scarf at the meeting.
"The top official in the Zohran Mamdani administration’s Office for International Affairs made plans to meet with Iran’s ambassador and permanent representative to the United Nations.
Commissioner Ana María Archila was scheduled to meet with Amir-Saeid Iravani, Iran’s permanent representative to the United Nations, at 2 United Nations Plaza, alongside two other senior officials in the Mayor’s Office for International Affairs on July 7 at 11 a.m ...
The meeting between Archila and Iravani was called off after the State Department—which was not informed ahead of time—met with the Mamdani administration to clarify acceptable conduct, according to the State Department official."
Ana Maria Archila, Commissioner (Source)
Pentagon launches testosterone screening program for male troops
What hormone will the DoD test female soldiers for? Is testerone testing in the military necessary? Is it obscure?
Is this potentially a violation of gender equality?
"The Pentagon will begin annually screening service members for testosterone deficiency, Defense Secretary Pete Hegseth announced on Wednesday, citing the need to ensure troops maintain the “biological foundation” to fight.
Under the new program, the secretary said service members aged 30 and older will be tested yearly for testosterone deficiency as part of routine health assessments. Troops under 30, meanwhile, will have the option to be tested, he said. ...
The decision to receive testosterone replacement therapy — if recommended by a medical professional — will remain up to the individual service member. ..."
Bangladesh to build 108-km fence along Myanmar border
Not only the US is building fences at its southern border.
"DHAKA -- Bangladesh has decided to fence off parts of its 271-kilometer border with Myanmar to combat transnational crime as the volatile situation in Rakhine state of the neighboring country deepens security concerns for Dhaka. ..."
Treating Taipei’s Homeless with a Policy of Compassion
Taiwan is not all technology and business!
"Taipei Main Station is a destination that wears many hats. Aside from being the largest land and rail transport hub for Northern Taiwan, it is also a critical space for the city’s unhoused population – where many gather, keep their belongings, and settle down.
Official statistics show there are about 3,000 homeless people, also known as “street friends” (街友), across Taiwan: 620 of them are said to be living in Taipei, a similar number may be found in neighboring New Taipei City, and the rest can be found dispersed in urban centers across the country. Social workers say most of the unhoused are male, with an average age of 60. Many, but certainly not all, are single and unmarried.
Despite this, Taipei’s aid workers believe the actual figure is much higher, since the official statistics only cover those who fit the government’s definition of a homeless person: someone who “sleep(s) on the streets or in any public area.” Additionally, non-government groups like Homeless Taiwan state that the number does not consider “rough sleepers,” or persons who have a permanent address but sleep in shelters, institutions, internet cafes, or couch-surf in the homes of friends or relatives.
Non-government groups like Salt Collectiv, which runs feeding programs around Taipei Main Station, say the homeless are mostly concentrated in three areas: in and around Taipei Main Station, at Bangka Park (艋舺公園) near Taipei’s Longshan Temple (龍山寺), and in the neighboring Wanhua district (萬華區). The areas are so well known they attract recruiters who often go to designated spots to find people willing to undertake low-paid, casual work.
Allowing homeless people to settle in high-traffic, high-visibility areas confounds both tourists and commuters ..."
More than Just Trade: Taiwan’s Investment Boom in the United States
Good news! This is mind boggling!
"... an investment agreement signed in January between the United States and Taiwan. The deal promises $250 billion in Taiwanese private-sector investment and up to $250 billion in government-backed credit guarantees in the United States. ...
Taiwan was the United States fourth-largest trading partner in 2025, with bilateral trade totaling approximately USD 256 billion. Four years prior, Taiwan was America’s eleventh-largest trading partner, with bilateral trade barely exceeding USD 100 billion.
Now, Taiwan is the United States fourth-largest source of imports (USD 201 billion), and the tenth-largest export market (USD 55 billion). This large volume of trade has led the United States to become Taiwan’s second-largest trading partner for over a decade. ...
While discussion of Taiwanese investment understandably focuses on TSMC, there are also a growing number of companies across a wide range of industries trying to expand into the American market. ..."
TSMC plans further $100bn US investment to feed AI demand (latest news) "Taiwanese chipmaker raises full-year capex to $64bn, revenue growth forecast to 40%"
In game theory, generalists sometimes win out over specialists in reinforcement learning
This could be an interesting new paper by Zico Kolter.
"... it does offer new insights into so-called imperfect-information games that involve two contestants facing off in a “zero-sum” competition, where one player’s gain means the other player’s loss. ...
The focus of the new work is on algorithms that could be used to train neural networks to participate in imperfect-information games. The assumption, long-held in the field, was that algorithms grounded in principles of game theory would, in this setting, clearly outcompete a general-purpose variety of algorithms called policy gradient methods, which came into use for decision-making in the 1990s. The term “policy” in this context basically means strategy, whereas “gradient” refers to a path that leads in the direction of greatest change — to the top (or bottom) of a hill, for example. Policy gradient methods are being used to train neural networks to make decisions that move — in small, sequential steps — toward a particular goal (like reaching a summit, metaphorically speaking), with continual adjustments and course corrections made along the way to bring the agent closer to the intended destination. ..."
From the abstract:
"In the past decade, motivated by the putative failure of naive self-play deep reinforcement learning (DRL) in adversarial imperfect-information games, researchers have developed numerous DRL algorithms based on fictitious play (FP), double oracle (DO), and counterfactual regret minimization (CFR).
In light of recent results of the magnetic mirror descent algorithm, we hypothesize that simpler generic policy gradient methods like PPO are competitive with or superior to these FP-, DO-, and CFR-based DRL approaches. To facilitate the resolution of this hypothesis, we implement and release the first broadly accessible exact exploitability computations for five large games.
Using these games, we conduct the largest-ever exploitability comparison of DRL algorithms for imperfect-information games. Over 7000 training runs, we find that FP-, DO-, and CFR-based approaches fail to outperform generic policy gradient methods."
Researchers confirm cause of Earth’s biggest mass extinction
Amazing stuff!
Unfortunately, Stanford University could not resist demagoguery by drawing parallels to today's climate change:
"... Warming today
The Stanford researchers plan to examine more marine animal groups to further understand the intertwined impacts of the three stressors of warming, lack of oxygen, and acidification, which are growing in severity today.
The researchers emphasize that history could well repeat itself, as changing ocean conditions threaten modern species that are vulnerable to warmer, oxygen-depleted waters. ..."
Today's warming is hardly comparable to what went on in the during the Permian–Triassic extinction event.
"In brief
- The Permian–Triassic extinction event, which killed off most life on Earth, did not impact all animal groups equally.
- In the oceans, groups of animals collectively called the Paleozoic fauna that had long dominated marine environments were almost completely wiped out, but the so-called modern fauna experienced far fewer extinctions and have dominated since.
- New research reveals that the warmer, poorly oxygenated oceans of the Permian–Triassic transition strongly favored the modern fauna’s faster metabolisms, while the Paleozoic fauna’s slow metabolisms could not keep up with increased oxygen demand triggered by the warming waters.
...
About 252 million years ago, 96% of marine species and 70% of land animals died off during the Permian–Triassic extinction event ...
for the first time incorporates biological responses of the animal groups that were decimated in the extinction and those that fared better. The groups hit hardest were those whose metabolisms could least tolerate warm, poorly oxygenated water. Such conditions prevailed throughout much of the world’s oceans as the Great Dying unfolded, caused by a surge of volcanic activity that released gargantuan amounts of planet-warming gases like carbon dioxide and methane into the atmosphere. ..."
From the significance and abstract:
"Significance
The well-established faunal turnover event between the Paleozoic fauna (e.g., brachiopods, crinoids) and the Modern fauna (e.g., clams, snails, urchins) coincided with intense global climate change of the latest Permian. We hypothesize that physiological differences in species vulnerability to temperature-dependent hypoxia explains this ecological transition. We test this hypothesis by performing physiological experiments on different taxonomic groups, dramatically increasing the amount of physiological data available for understudied but ecologically significant marine taxa. Simulations of extinction patterns guided by these traits show that ocean warming and deoxygenation together caused the taxonomic selectivity of the end-Permian mass extinction and resulting permanent shift in marine ecosystem composition. Similar selectivity patterns are expected in the modern biodiversity crisis due to similar environmental circumstances.
Abstract
The rapid global climate change at the end of the Permian Period (~251.9 Mya) coincided with the greatest macroevolutionary faunal turnover event in Earth’s history.
As the oceans warmed, lost dissolved oxygen, and became more acidic, the dominant animal groups in the Paleozoic fauna (including brachiopods and crinoids) suffered differentially high rates of extinction, allowing the Modern fauna (including bivalves and gastropods) to rise to ecological dominance.
The end-Permian kill mechanism(s) are not fully understood, but differences in extinction intensity among Linnaean classes suggest an important physiological component.
Here, we use a trait-based model of species’ metabolic O2 balance to demonstrate that temperature-dependent hypoxia can explain the taxonomic selectivity of the end-Permian mass extinction.
Direct respirometry experiments and physiological trait estimates derived from biogeographic data reveal that species belonging to the Paleozoic fauna have a higher temperature dependence of hypoxia than those belonging to the Modern fauna. In simulations of the climate transition, this trait difference leads to a greater loss of aerobic habitat for Paleozoic fauna, consistent with their observed greater extinction intensity.
These results demonstrate that differences in average physiological tolerances to environmental change across biogeography, taxonomy, and functional ecology drove end-Permian extinction patterns and could eventually characterize the modern biodiversity crisis.
Temperature-dependent hypoxia is the only kill mechanism that has been shown to explain the magnitude, biogeography, and now taxonomic selectivity of the end-Permian mass extinction, ultimately underlying the permanent shift in marine ecosystems across this transition."
North Korea’s role in providing Russia with artillery for the war in Ukraine
Bad news!
"Ukrainian military intelligence (HUR) confirmed North Korea’s role in providing Russia with artillery for the war in Ukraine, including an estimated 25-40 percent of current Russian artillery ammunition.
North Korea has supplied more than 600 artillery systems of different types and calibers to Russia, including M-1989 Koksan 170mm self-propelled guns, M-1991 240mm multiple-launch rocket systems (MLRS), 107mm Type 63 rocket launchers, self-propelled anti-tank missile systems, D-74 122mm towed guns, and Type 76 140mm mortars, anti-tank guided missiles, and artillery shells.
North Korea has also provided more than 100 KN-23 and KN-24 ballistic missiles to Russia, which has used at least 80 of these missiles.
HUR frames these weapons deliveries as “systemic” in their scale and regularity, with North Korea occupying a “key position” among Russia’s foreign weapons and equipment suppliers. North Korean weapons provisions to Russia reportedly totaled over 9 million 12mm and 152mm artillery rounds by the end of 2024, 12 million 152mm artillery shells by July 2025, and an estimated 15 million 152mm artillery shells by March 2026. ..."
Azerbaijani President Ilham Aliyev Lauds Trump as ‘a Man of Peace,’ Credits Him with Armenia‑Azerbaijan Deal
President Trump, the peacemaker!
"... the Azerbaijani president credited Trump’s success to a fundamentally different approach to resolving long-running conflicts. ...
The Azerbaijani president argued that previous administrations spent nearly three decades pursuing policies that effectively froze the Armenia-Azerbaijan conflict rather than resolving it, while Trump approached the conflict from an entirely different perspective. ...
Describing Trump as “a person who loves peace” who “sees peace as an opportunity,” Aliyev said Trump’s role proved decisive as Azerbaijan and Armenia reached the final stage of negotiations on a peace agreement. According to Aliyev, Trump and his team understood Azerbaijan’s concerns, worked to persuade Armenia that peace served both countries’ interests and ultimately “created such a framework that peace became possible.” ..."
Warum Bosch der Wandel nicht gelingt
Wieder mal schlechte Nachrichten aus der Bananenrepublik D!
Was passiert mit deutschen Traditions-/Spitzenunternehmen? Sterben sie aus? Sind sie nicht mehr wettbewerbsfähig?
"Bosch kommt nicht zur Ruhe. Gerade erst hatte der Traditionskonzern die Gespräche über den größten Stellenabbau in der Unternehmensgeschichte beendet und den Schock über das erste Verlustjahr nach langer Zeit verkraftet, als Ende Juni überraschend Stefan Hartung als Vorsitzender der Geschäftsführung zurückgetreten ist. Nicht nur das Streichen von fast 28.000 Stellen in Deutschland, sondern auch der Rücktritt, der eigentlich eine Entlassung war, machen eines deutlich: Der Umbruch beim weltgrößten Autozulieferer ist noch lange nicht zu Ende. ..."
Cell therapy successfully treats terminal, incurable, and deadly brain cancer in children
Good news! Cancer is history (soon)!
"An experimental immunotherapy has staved off brain tumours that would usually be incurable in four children. The therapy trains immune cells in a person’s blood to target three protein markers found in paediatric brain cancers.
The tumour-associated antigen T-cell therapy was given to 33 children and young adults with brain tumours that are usually fatal.
One patient with DIPG, a type of cancer that is usually fatal within a year, is alive two years post therapy.
Three others with other types of aggressive or recurrent brain cancers remain disease-free between two and five years on.
It is unclear why the treatment worked in some and not others. The small safety trial doesn’t deliver a cure yet, but the therapy doesn’t require genetic engineering, can target multiple proteins and might work better for solid tumours than chimeric antigen receptor (CAR)-T-cell therapy."
"Key takeaways:
- The multi-target design may help address tumor heterogeneity, one of the major barriers to successful treatment of aggressive childhood cancers.
- The trial successfully established a feasible manufacturing process, identified a maximum tolerated dose and defined an early safety profile – key milestones needed to advance the therapy into future Phase 2 studies.
- Researchers analyzed patients with DIPG or relapsed brain cancers, showing both responses and prolonged disease control, with some patients remaining disease-free years after treatment.
..."
From the abstract:
"Central nervous system (CNS) tumors are the deadliest cancers in children, highlighting the need for new therapies. The tumor-associated antigens (TAAs) WT1, PRAME and survivin are widely expressed by these tumors, and a manufacturing technique has been developed to target these intracellular TAAs using autologous, nongenetically engineered T cells.
Here we therefore conducted ReMIND, an open-label, phase 1 adaptive dose-finding study to determine the safety/feasibility of autologous, systemically administered trivalent T cells targeting WT1, PRAME and survivin in children with CNS tumors.
Eligible patients had newly diagnosed diffuse intrinsic pontine glioma without lymphodepletion (arm A, n = 16 enrolled, n = 11 infused) and relapsed/recurrent nonbrainstem CNS malignancies without (arm B, n = 28 enrolled, n = 18 infused) or with (arm C, n = 7 enrolled, n = 4 infused) lymphodepletion.
Primary end points were safety, feasibility and maximum tolerated dose determination; secondary end points included preliminary efficacy and immunobiological correlates, including in vivo TAA-T persistence and systemic immune activation. Dose level 3 (8 × 107 cells per m2 per dose) was determined as the maximum tolerated dose.
Treatment was well tolerated with fatigue and headache being the most common adverse events, although two possibly related serious adverse events of tumor swelling occurred. One grade 5 event in a patient with diffuse intrinsic pontine glioma with hydrocephalus, tumor edema and respiratory failure was categorized as a dose-limiting toxicity.
Median overall survival for arm A was 13.7 months from diagnosis (range, 6.2–32.0) and median progression-free survival for arms B/C was 5.0 months from infusion (range, 0.5–51.6).
Three patients in arms B/C are alive without disease at 31.8, 41.2 and 51.6 months without further treatment, including one complete response.
This trial met safety/feasibility primary end points with some preliminary signals of efficacy. ..."
Four children with terminal brain cancer saved by new cell therapy "An experimental immunotherapy has beaten aggressive brain tumours in a handful of children, and a personalised version is now being tested on more patients"
Innovative cell therapy offers new hope for children with aggressive brain tumors (original news release)
Multi-antigen-targeting T cells in pediatric central nervous system tumors: a phase 1 trial (open access)
Magnetic resonance imaging scan of a vertical section of the head of a child with a glioma brain cancer (dark red, centre)
Plenty of wigs on display in a department store in China
Do male wigs come with sunglasses and a bow tie? Incognito?
Wednesday, July 15, 2026
Sweden Introduces Tough New Deportation Rules for imMigrants
Good news! Will other Western countries follow?
Ukraine's Defense Minister Resigns as Zelensky Unveils Major Wartime Cabinet Reshuffle
Maybe a fresh military leadership does not hurt after 4 years of war!
President Trump Says Iraq Will Be Rid of Iran ‘Burden’ Soon As he Hosts Iraqi PM
Recommendable! Good news!
Meet the Trump of Bagdad! When two businessmen get along very well!
BlackRock made history as the first investment firm to manage more than $15 trillion in assets.
Stunning!
"BlackRock took in $192 billion in new client money across its public and private markets businesses in the three months ending in June. ..."
President Trump said to urge Benjamin Netanyahu to withdraw IDF soldiers from Syria, Lebanon
How President Trump manages the Middle East and the Abraham Accords!
Indeed, the presence of IDF soldiers in Syria could possibly destabilize the regime of interim President Ahmed al-Sharaa! Good call!
"US President Donald Trump allegedly told Prime Minister Benjamin Netanyahu that Israel should remove IDF troops from Syria and Lebanon during a phone call on Thursday, according to a Tuesday Axios report citing US and Israeli officials.
Trump claimed that the presence of Israeli military personnel in Syrian territory could create tension and may lead to escalation, one US official told Axios.
“They don’t want you there. You should redeploy,” Trump allegedly told Netanyahu, according to the official. ..."
US charges Russian 'bulletproof' web hosts over cyberattacks that netted $62M from cybercrime victims
How rogue is Russia under Putin the Terrible!
"U.S. prosecutors have charged three Russian nationals and two web hosts with hacking, conspiracy, and money laundering over their alleged roles in hosting cyberattacks that caused tens of millions of dollars in damages to U.S. businesses.
The three Russians, Alexander Volosovik, Kirill Zatolokin, and Yulia Pankova, who reside in St. Petersburg, are accused of owning and running two web hosts, Media Land and ML.Cloud, which allegedly provided criminals and state-backed hackers with web hosting and infrastructure support for carrying out cyberattacks. ..."
New York state sues chemical companies over PFAS pollution. Really!
Another frivolous lawsuit by Dimocratic Party member and New York state attorney general Letitia James!
New York state attorney general Letitia James
TAPIR, a new CRISPR-based technology
Good news!
"Which comes first: cellular dysregulation or ribosomal RNA? This classic chicken-and-egg problem has long flummoxed scientists studying diseases associated with altered levels of ribosomes ... researchers have struggled to confirm whether unusual ribosomal readings in patients with certain cancers and congenital conditions were contributing to the diseases’ progressions or were side effects from other cellular issues.
To better understand this relationship, scientists created a new gene editing technology called Targeted Activation of Protein Translation—TAPIR, for short. ... the new technology allows them to tweak the quantities of the ribosomal RNA (rRNA) that makes up the bulk of the ribosomes themselves. By adjusting how much rRNA a cell generates and then observing how that affects a given medical condition, the team could finally establish a cause-and-effect relationship between ribosomes and disease.
The researchers first put TAPIR to the test by prompting ribosome formation in mice with a rare congenital disease that impedes rRNA creation. Once the rRNA was switched back on, the mice showed some improvement in their condition. Then, they activated rRNA in a different set of mice with pancreatic cancer, a disease associated with heightened ribosomal activity; in response, the cancer hastened its already rapid expansion.
Both results point to the same conclusion: Changes in rRNA levels are causes of disease symptoms, not symptoms themselves."
"... New Perspectives for Rare Diseases and Cancer
The results could be particularly relevant for diseases in which ribosome function is disturbed. These include ribosomopathies such as Treacher-Collins syndrome, a rare congenital disease that causes facial malformations. In a mouse model, the researchers successfully managed to partially compensate for disease-related alterations by stimulating rRNA production in a targeted way.
In addition, the research team observed that similar mechanisms also play a role in pancreatic cancer. Tumor cells seem to use increased rRNA production to maintain their rapid growth. In the mouse model for pancreatic cancer, TAPIR was able to increase rRNA production and promote the growth of the cancer cells. This shows that the increased rRNA production has a causal effect in contributing to tumor growth and is not just a side effect. ..."
From the abstract:
"Ribosomal RNA (rRNA) transcription rates vary during development, and their dysregulation is linked to diseases such as cancer and ribosomopathies. Owing to their high abundance and genomic redundancy, the functional significance of rRNA-levels remains unclear.
Here, we developed TAPIR (Targeted Activation of Protein Translation), a CRISPR-based approach to elevate rRNA-levels by inducing 47S rDNA transcription. TAPIR increased nucleolar size and enhanced protein synthesis, even in rapidly proliferating cells.
In neural stem cells, elevated translation promoted self-renewal and proliferation in vitro and in vivo.
Furthermore, TAPIR enabled the modeling and partial rescue of associated disease phenotypes.
Our findings revealed that rRNA-levels directly regulate translational output and that protein synthesis capacity can act as a key determinant of mammalian stem cell behavior."
New CRISPR Method Makes It Possible to Control Protein Production in Cells (original news release) "A research team led by Prof. Stefan Stricker at Helmholtz Munich and Ludwig Maximilian University has developed TAPIR, a CRISPR-based technology that enables precise control of protein production in cells. The method provides new insights into the regulation of stem cells, cancer, and rare diseases, while opening up new opportunities for biomedical research."
In dramatic testimony, US Supreme Court Justice Barrett recalls being given bullet proof vest over death threats. Really!
Wow! What! Bullet proof vests for US Supreme Court justices? Let that sink in for a moment!
Why not a helmet to protect their brains? Just kidding!
I blogged here yesterday about the testimony of two US Supreme Court Justices before the US Congress.
"... she had to explain to her 12-year-old son why she brought home a bulletproof vest to protect herself from death threats. ...
Barrett emphasized how her life changed after the 2022 Dobbs decision was released, which overturned the historic ruling in Roe v. Wade, which boosted threats against her and the other conservative justices. ..."
Justice Amy Coney Barrett (Source)
A global workspace in language models by Anthropic
Amazing stuff!
"... We find that Claude has developed a small collection of internal neural patterns that, compared to all its other internal processing, play a special role.
We call the collection of these patterns the J-space—named after the technique we used to find them, involving a mathematical concept called the Jacobian.
Each J-space pattern is linked to a particular word. But when one of these patterns lights up, it doesn’t mean the model is saying that word—just that the word is on its mind. If you've heard of language models having a "scratchpad" or “chain of thought”—text they write to themselves while reasoning—
the J-space is something different. It operates silently, in the model’s internal neural activations, allowing the model to think about a concept without writing it down. Notably, the J-space wasn’t designed or programmed by us, but instead emerged on its own during Claude’s training process. ...
We find that the J-space has a number of unique properties, compared to the rest of Claude's processing:
- Claude can report on these representations. If you ask Claude what it's thinking about, it will tell you what’s in the J-space. Non-J-space representations are less reportable.
- It can also modulate them on request. If you ask Claude to think about something, or solve a problem silently in its head, it will light up the appropriate patterns in its J-space. By contrast, it has trouble modulating patterns not in the J-space.
- Claude uses its J-space for internal reasoning. If you ask Claude to solve a problem that requires multiple steps, the intermediate steps will light up in its J-space, even when it doesn’t say them out loud. These J-space patterns causally mediate its performance in such tasks, despite being smaller in magnitude than other representations.
- Representations in the J-space can be used flexibly for many tasks—for example, once “France” has lit up in Claude’s J-space, the model can recall its capital, or its national currency, or the continent it belongs to.
- However, despite its important role, the J-space is not involved in most of what a language model does—speaking fluently, recalling simple facts, using correct grammar, etc. In experiments where we prevented Claude from using its J-space, it still interacted normally, but lost its higher-order cognitive functions.
..."
"... If the mind is an ocean, we spend our lives floating at the surface. Beneath us, an enormous amount of processing takes place without our knowledge: our visual systems parsing the contours of a face, our motor circuits maintaining our posture. At any given moment, only a small fraction of this neural activity is accessible to us. Yet it is this privileged sliver of activity that we rely on to reason deliberately ...
In this paper, we present evidence that an analogous functional distinction has emerged in modern AI models. Specifically, we observe that language models maintain a privileged set of internal representations, available for report, modulation, and flexible internal reasoning, atop a much larger volume of automatic processing. We identify these representations using a new interpretability technique, which surfaces the concepts a model is poised to verbalize at any point in its processing. Measuring and intervening on these representations provides us a window into a model’s thought processes, uncovering internal reasoning and reactions that do not appear in its output. ..."
The richer get richer – not just materially, but socially | Cornell Chronicle. Really!
So what! Especially in America, an individual (no matter rich or poor or background) with skills and talent is most likely quickly spotted and given an opportunity.
It also depends very much on the individual him- or herself to make better "social connections" in life!
Is sociology a pseudoscience?
"... This familiar example of elite network influence proves the adage that it’s not what you know but who you know that matters. And such disparities in social connection are far more pervasive throughout daily life than generally recognized, Cornell sociologists argue in a new open-access book, “Friends and Fortunes: Social Capital Inequality in America.”
Inequality in social ties is as stark as income inequality in the United States, the researchers say – and the two are mutually reinforcing. ...
“Why should there be differences in access to valuable social connections? Are they not free?” the authors write. “One would hope so. But the data tell a different story.” ...
explain how social networks subtly feed hidden systems of power, prestige, wealth and, ultimately, life chances. ..."
Guide to Loop Engineering: How 'autoresearch' and 'Bilevel Autoresearch' Turn AI Agents Into Autonomous Machine Learning ML Research Loops
Recommendable, but a long, convoluted headline! An overview article.
The pig in the python or how successive population cohorts influence economics and labor supply
Demography is destiny!
Like so many economic and social science theories, they explain historical data very well for a time and make reasonable predictions, but eventually they fail.
"In 1978, Richard Easterlin stood before the Population Association of America and predicted that the wages of young American men — falling since 1973 — would turn around by 1984.
He had earned the confidence. As we have discussed in the past, Easterlin’s relative cohort size hypothesis was the most successful applied demography of its era, and it had already explained the biggest population event in American history—the Baby Boom. He replaced the mysticism about returning GIs, swooning Rosie Riveters, and suburban optimism with actual social science. ...
Easterlin found the explanation for the unexpected Baby Boom in economics. The Depression produced small birth cohorts. Those cohorts came of working age in a booming postwar economy and immigration tightly restricted by the Immigration Act of 1924. The “lucky few” folks walked into a market starved for workers. They got high wages relative to what they’d grown up expecting, and so they married early and had a lot of children.
Tight labor markets and closed borders made the Baby Boom. That is the core of Easterlin’s thesis, and it is the part of his work that has aged best. ..."
From the significance and abstract:
"Significance
The large size of the baby boom cohort depressed economic opportunities for that generation as it flooded the labor force in the 1970s.
Contrary to the predictions of Richard Easterlin’s relative cohort size hypothesis, however, there was no rebound of opportunities with the advent of smaller cohorts in the mid-1980s; incomes of young adults remained depressed until the 2010s.
New measures of flows into and out of the labor force make it simple to assess the relative impact of retirement, female labor-force participation, and immigration in concert with the size of birth cohorts entering the labor force. The results suggest that the departure of baby boomers from the labor force will have profound implications for economic opportunities of new workers.
Abstract
This analysis revisits the relative cohort size hypothesis proposed by demographer Richard Easterlin in the 1960s.
Easterlin argued that the economic and social prospects of a generation are influenced by the size of the cohort relative to adjacent cohorts.
He hypothesized that relative cohort size affects wages, employment, marriage, and fertility decisions.
The theory fits the data well for the period from 1940 to 1980 but fails in later decades.
I present a more nuanced view of the impact of demographic factors on worker competition through the lens of a measure of decennial labor-force flows. This approach allows consideration of the effects of retirement, female labor-force participation, and immigration on labor-market competition.
I calculate these flows for the period from 1910 to 2040 and propose an index of employment competition. The results show that trends in labor-market competition are consistent with wage trends of young workers since 1940. Projections to 2040 show that we are on the verge of a radical reshaping of labor markets in which new workers will be in extremely short supply."
Canada to plug airspace surveillance gaps with Australia's over-the-horizon radar
Both countries are geographically over the horizon! 😊
"The North American Aerospace Defense Command (NORAD), operated by Canada and the United States, has gaping holes in its airspace radar coverage. Ottawa went some way to filling those gaps when it signed four related agreements in June for a new radar system sourced from Australia.
The system is called the Arctic Over-The-Horizon Radar (A-OTHR), and the deal is worth some US$1.75 billion to Australia. ..."
JORN, a network of three over-the-horizon radar facilities, is a strategic and early-warning asset for Australia. Canada is buying a similar system.
Meet Biomni – an AI-powered biomedical co-scientist
The future of scientific research & discovery is here thanks to ML & AI!
This is new research by Jure Leskovec and his team.
"In brief
- Biomni is an AI agent “co-scientist” that can help biomedical researchers through the entire research workflow, with outputs that demand human experience and reasoning.
- The tool specializes in being able to work with prompts written in casual language, such as “Why are these patients responding differently to the drug?”
- A prototype Biomni is already in use by more than 10,000 labs, making it the most widely used AI co-scientist system in biomedicine.
..."
From the abstract:
"Biomedical research is increasingly constrained by repetitive, fragmented workflows that slow discovery.
We introduce Biomni, a general-purpose biomedical artificial intelligence agent that autonomously executes diverse research tasks.
To map the biomedical action space, Biomni’s action-discovery agent mines tools, databases, and protocols from thousands of publications across 25 domains, building a unified agentic environment.
Its general-purpose architecture integrates large language model reasoning with retrieval-augmented planning and code-based execution, dynamically composing workflows without predefined templates.
Systematic benchmarking shows strong generalization across heterogeneous tasks—causal gene prioritization, drug repurposing, rare-disease diagnosis, microbiome analysis, and molecular cloning—without task-specific tuning.
Real-world case studies demonstrate Biomni interpreting multi-modal datasets, optimizing protein stability, orchestrating wet-lab instruments, and generating experimentally testable protocols.
Biomni envisions artificial intelligence augmenting human scientists and accelerating discovery."
Autonomous biomedical research with an artificial intelligence agent (no public access)
Biomni: A General-Purpose Biomedical AI Agent (preprint, open access)
When Does Continual Learning Require Learning
This could be an interesting new paper by Trevor Darrell and Jitendra Malik.
From the abstract:
"As large language models (LLMs) become increasingly capable, the next question is how can we enable models to continually learn?
Today, the field largely frames this as a problem of context management and mitigating forgetting.
We argue this framing is incomplete: continual learning is fundamentally about increasing model competence as the world changes.
We disentangle this change along two axes -- space, where the model encounters new domains, and time, where the underlying data drifts under a fixed task. This framing lets us study continual learning under realistic conditions: new domains arrive over time, facts drift past their training cutoff, and agentic interactions accumulate state across episodes.
To evaluate methods under this setting, we recast widely used LLM benchmarks as sequential problems and introduce a single mechanism-agnostic protocol that compares prompt-based methods (GEPA, ACE), supervised learning (SFT, SDFT), reinforcement learning (GRPO, SDPO), and context compression (Cartridges, In-place TTT).
Prompt-based methods fit each new stage quickly but degrade on future tasks. Distillation-based methods accumulate knowledge stably but struggle to update outdated facts.
Context compression improves efficiency without substantially improving the ability to learn new tasks.
Online reinforcement learning adapts most effectively to knowledge updates but remains sensitive to noisy reward signals.
Overall, our results suggest that continual learning is not a single capability: different patterns of environmental change require fundamentally different update behaviors, determining when adaptation must be learned inside model weights and when it can be achieved through external scaffolding.
We hope that understanding where each method succeeds and fails will guide the design of stronger continual learning systems."
Ewha Womans University in South Korea
No misspelling! 😊
Founded in "1886.5.31: Ewha Haktang was founded by Mary F. Scranton, an American missionary of the Methodist Episcopal Church."
What do these ladies research today? New process turns mixed plastic waste into hydrogen fuel without sorting
Subscribe to:
Posts (Atom)