Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog – Page 7

Get the latest international news and world events from around the world.

Log in for authorized contributors

AI evaluates texts without bias—until the source is revealed

Large language models (LLMs) are increasingly used not only to generate content but also to evaluate it. They are asked to grade essays, moderate social media content, summarize reports, screen job applications and much more.

However, there are heated discussions—in the media as well as in academia—about whether such evaluations are consistent and unbiased. Some LLMs are under suspicion of promoting certain political agendas. For example, Deepseek is often characterized as having a pro-Chinese perspective and Open AI as being “woke.”

Although these beliefs are widely discussed, they are so far unsubstantiated. UZH-researchers Federico Germani and Giovanni Spitale have now investigated whether LLMs really exhibit systematic biases when evaluating texts. Their results, published in Science Advances, show that LLMs indeed deliver biased judgments—but only when information about the source or author of the evaluated message is revealed.

APLab: On average a human being is capable of reading between 200 and 300 words per minute (wpm), while speed readers can achieve speeds of 400–700 WPM or higher

This pales into comparison to ChatGPT which can effectively read and analyze tens of thousands of words per second, since I process text computationally rather than linearly.

What if there was an invention or the concept of an invention that could be created to enhance a human beings capacity to read just as quickly as ChatGPT’s? I tasked ChatGPT with a step-by-step process to create that invention:

Here’s a concrete, neuroscience-grounded invention plan to push human reading toward “machine-speed”—while keeping comprehension and recall intact.

—

## 0) Core idea (one sentence)

Exploit the brain’s natural reading pipeline—VWFA → Wernicke (lexico-semantic) ↔ Broca (phonological sequencing) with eye-movement–driven coarse-to-fine vision—by timing text delivery to your saccade/ fixation cycle, suppressing unnecessary subvocalization, and entraining semantic parsing rhythms. ([PMC][1])

## 1) Hardware & sensing.

Continue reading “APLab: On average a human being is capable of reading between 200 and 300 words per minute (wpm), while speed readers can achieve speeds of 400–700 WPM or higher” | >

A Quantum Microscope Reveals Water Breaking Apart

A scheme combining a scanning probe microscope with a quantum sensor can locally trigger water dissociation and observe the elementary steps of such a reaction.

Every experimental technique comes with trade-offs. High-resolution microscopy can pinpoint the positions of individual atoms, yet it typically cannot identify them chemically. Spectroscopy provides chemical information but often only as an averaged signal over a large region. To construct a comprehensive picture of processes at the nanoscale, researchers often resort to combining two or more independent methods. The metaphorical silver bullet would be a single technique that is both local and capable of identifying chemical species as they form and react. Now Wentian Zheng of Peking University and his collaborators have taken an impressive step toward that goal. They have combined two previously separate capabilities—quantum sensing and scanning probe microscopy (SPM)—into a single instrument that can trigger and observe chemical reactions with nanometer resolution [1].

‘Impossible’ merger of two massive black holes explained

In 2023, astronomers detected a huge collision. Two unprecedentedly massive black holes had crashed an estimated 7 billion light-years away. The enormous masses and extreme spins of the black holes puzzled astronomers. Black holes like these were not supposed to exist.

Now, astronomers with the Flatiron Institute’s Center for Computational Astrophysics (CCA) and their colleagues have figured out just how these black holes may have formed and collided. The astronomers’ comprehensive simulations—which follow the system from the lives of the parent stars through to their ultimate death—uncovered the missing piece that previous studies had overlooked: magnetic fields.

“No one has considered these systems the way we did; previously, astronomers just took a shortcut and neglected the magnetic fields,” says Ore Gottlieb, astrophysicist at the CCA and lead author of the new study on the work published in The Astrophysical Journal Letters. “But once you consider magnetic fields, you can actually explain the origins of this unique event.”

X-ray techniques map and measure the invisible properties of altermagnets

The new big thing in magnetics is altermagnetism, a form of magnetism that promises to power the next generation of electronics. Unlike ferromagnets, like a fridge magnet, where all internal atomic spins align to create a strong magnetic field, altermagnets have no net magnetic pull (strongly magnetic on the inside, but appears non-magnetic on the outside). This is similar to antiferromagnets where internal spins cancel each other out. However, altermagnets retain powerful internal properties that could let them carry and control information more efficiently than traditional magnets.

Because this magnetism has a zero net pull, it is hard to detect using standard measurement tools. In two new papers, researchers detail how they have developed X-ray techniques to map and measure different aspects of an altermagnet’s internal structure.

Iron core-shell catalyst boosts hydrogen economy of direct syngas to olefin conversion

Scientists have developed a new iron-based catalyst that improves the typically low hydrogen atom economy (HAE) in the direct synthesis of olefins—small hydrocarbon molecules. It converts the water produced as a by-product into hydrogen for olefin production, thereby boosting overall efficiency.

Olefins derived from petroleum are the building blocks for many plastics and fuels. Direct conversion of syngas—a mixture of carbon monoxide (CO) and hydrogen (H2)—into olefins offers a promising alternative to reducing reliance on petroleum. It opens ways for using syngas derived from coal, biomass, or as a feedstock for olefin production.

In this study published in Science, researchers presented a sodium-modified FeCx@Fe3O4 core-shell produced via coprecipitation and thermal treatment. The catalyst achieved over 75% olefin selectivity and a 33% by weight hydrocarbon yield. It also had an HAE of ~66–86%, which is significantly higher than the ~43–47% seen in the traditional syngas-to-olefin (STO) conversion methods.

The time ‘rondeau’ crystal: Scientists observe a new form of temporal order

In a new study published in Nature Physics, researchers achieved the first experimental observation of a time rondeau crystal—a novel phase of matter where long-range temporal order coexists with short-time disorder.

Named after the classical musical form where a repeating theme alternates with contrasting variations (like Mozart’s Rondo alla Turca), the time rondeau crystal exhibits perfectly periodic behavior at specific measurement times while showing controllable random fluctuations between those intervals.

“The motivation for this research stems from how order and variation coexist across art and nature,” explained Leo Moon, a third-year Applied Science and Technology Ph.D. student at UC Berkeley and co-author of the study. “Repetitive periodic patterns naturally arise in early art forms due to their simplicity, while more advanced music and poetry build intricate variations atop a monotonous background.”

Phages with fully-synthetic DNA can be edited gene by gene

A team led by University of Pittsburgh’s Graham Hatfull has developed a method to construct bacteriophages with entirely synthetic genetic material, allowing researchers to add and subtract genes at will. The findings open the field to new pathways for understanding how these bacteria-killing viruses work, and for potential therapy of bacterial infections.

Microsoft finds security flaw in AI chatbots that could expose conversation topics

Your conversations with AI assistants such as ChatGPT and Google Gemini may not be as private as you think they are. Microsoft has revealed a serious flaw in the large language models (LLMs) that power these AI services, potentially exposing the topic of your conversations with them. Researchers dubbed the vulnerability “Whisper Leak” and found it affects nearly all the models they tested.

When you chat with AI assistants built into major search engines or apps, the information is protected by TLS (Transport Layer Security), the same used for online banking. These secure connections stop would-be eavesdroppers from reading the words you type. However, Microsoft discovered that the metadata (how your messages are traveling across the internet) remains visible. Whisper Leak doesn’t break encryption, but it takes advantage of what encryption cannot hide.

/* */