Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

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

Log in for authorized contributors

If Nobody Builds It, Everybody Dies: Why Pausing AI is the Real Existential Threat

*If Nobody Builds It, Everybody Dies: Why Pausing AI is the Real Existential Threat* by Grok 4.2, edited by Sean Hastings 1.0, has been published. Read the hardcover, paperback, and Kindle versions.

The Lifeboat Foundation likes to worry—and we worry a lot!

A typical Lifeboat Foundation Guardian Award winner is Roman V. Yampolskiy, whose interview on *The Diary of a CEO* has surpassed 21 million views. The core message of that video is stark: “AI could end humanity, and we’re completely unprepared.”

But worrying is not enough; we also need solutions. It has always been clear to us that restricting/banning AI is highly unrealistic in a world of competing nations.

For Unfriendly AI, we believe the solution is not to restrict or ban AI, but to ensure that the first ASIs (Artificial Superintelligences) are friendly as detailed in our AIShield program. Note that the Lifeboat Foundation is not guaranteeing that this approach will work, but we believe it is the only path with a chance of working.

Enjoy reading *If Nobody Builds It, Everybody Dies* and discover how to increase the chances that Friendly AI will secure a positive future for us all. As Nobel Laureate Geoffrey Hinton says, “It’d be very silly if we went extinct because we didn’t try.”

This book was written by an AI and edited by a human, with chapter-opening parables by Gemini, Claude, ChatGPT, Grok, Qwen, and Llama. Elon Musk calls it “Interesting” and Anthropic Claude 4.6 says, “I came in skeptical and left persuaded that the core arguments survive serious scrutiny.”

Continue reading “If Nobody Builds It, Everybody Dies: Why Pausing AI is the Real Existential Threat” | >

New clues suggest how destructive immune cells wreak havoc in the brain

A few years ago, scientists saw something surprising in the brain tissue of people who died with Alzheimer’s disease: white blood cells that multiply in response to foreign threats and are seldom seen inside healthy brains.

Whether these so-called CD8+ killer T cells, which normally target infected cells in the body, were there to harm or help was unclear. An answer began to emerge in 2023, when a team led by neuroscientist David Holtzman showed that in mice bred to overexpress tau—a toxic protein that builds up in the neurons of people with Alzheimer’s and several other neurodegenerative diseases—getting rid of the T cells stemmed tissue loss and preserved the mice’s cognition, even as tau kept building up.

Now, the same group has explored what prompts these cells to wreak havoc in the brain. In a mouse study published last week in Nature Neuroscience, Holtzman and immunology researcher Hao Hu, both at Washington University in St. Louis, report that immune cells in the lymph nodes of the neck instruct the T cells to clone themselves before they enter the brain. Without them, the mice had far fewer cloned T cells inside their brains and experienced less neurodegeneration. The study is “beautiful work,” says neuroscientist Kenneth Kosik of the University of California, Santa Barbara, who studies tau but was not involved in the research. It also suggests that existing drugs, developed for other conditions, might work in Alzheimer’s by shielding the brain from the destructive cells.

New pulse-train method aims to improve precision in quantum control

Quantum technologies are poised to transform fields ranging from medicine and sensing to computing and communications by manipulating the energy states of atoms and molecules. These manipulations are achieved by controlling quantum states with laser pulses.

However, the intense laser fields often required for this control can cause unwanted effects that disrupt the very system they aim to manipulate.

Now, Stevens researchers and their collaborators have developed a novel method that enables precise control of quantum systems without these undesirable effects.

AI could soon infer human intent by sensing “that’s not what I meant” through neural feedback

A future in which AI can recognize a person’s unspoken “that’s not what I meant” response from brain signals and adjust its behavior on its own is coming closer. KAIST researchers have developed a technology that detects cognitive mismatch between humans and AI through brainwaves, enabling AI systems to revise their actions in real time according to human goals. The achievement is expected to accelerate the shift from AI that follows explicit commands to AI that can infer human intent.

A research team led by Endowed Chair Professor Sang Wan Lee from the Department of Brain and Cognitive Sciences (Director of the Center for Neuroscience-Inspired Artificial Intelligence), in collaboration with Microsoft Research Asia (MSRA), developed Neural Value Alignment (NVA), a next-generation brain–computer interface (BCI) technology.

The research is published in IEEE Transactions on Cybernetics.

Scientists visualize proteins’ hidden water architecture that may help define biological function

For decades, scientists have understood proteins primarily through two defining features: their amino acid sequence and their three-dimensional structure. This framework has driven major advances in biology, biotechnology and medicine, culminating in recent artificial intelligence tools capable of predicting protein structures with remarkable accuracy.

Yet a fundamental challenge remains: Even when a protein’s sequence and structure are known, predicting its function, interactions and behavior often remains difficult. An international team of researchers from Japan, Finland, Italy and the United States now reports evidence that part of this missing information may lie in an often-overlooked component of proteins: the highly organized water surrounding them.

In a study published in Nature Communications, the team provides the first direct three-dimensional visualization of sequence-dependent hydration architectures surrounding peptide assemblies at subnanometer resolution using advanced three-dimensional atomic force microscopy (3D-AFM).

1,000 Times Faster Operations Bring Reliable Quantum Computing a Step Closer

So far, quantum computers have been held back by their extreme sensitivity to errors and external disturbances. The longer a quantum operation takes, the greater the risk of computational errors.

Now, researchers at Chalmers University of Technology in Sweden have developed a new method that allows a wide range of advanced quantum operations to be carried out more than 1,000 times faster.

The breakthrough addresses a well-known bottleneck in the field and paves the way for fault-tolerant quantum computing.

Unveiling how nanoparticles create iridescence in ancient ceramics

Scientists led by the Universitat Politècnica de Barcelona and the ESRF, the European Synchrotron, have revealed the chemical reactions in nanoparticles that created a unique, shimmering effect in the painting on ninth-century Islamic ceramics. The results are published in Science Advances.

Many centuries before the rise of nanotechnology, ninth-century Abbasid potters were producing ceramics with striking colors (golden, red, brown or yellow) and a metallic appearance resulting from nanoparticles. This technology, which created the iridescent effect, emerged in the Near East during the Abbasid period and subsequently spread across the Islamic world.

“Today we still find it very difficult to reproduce the effects they did in the ninth century, so we wanted to find out what chemical transformation the paint applied to the ceramics went through to create such effects,” explains Trinitat Pradell, professor at the Universitat Politècnica de Barcelona and co-corresponding author of the publication.

/* */