Toggle light / dark theme

Faster breast MRI—AI unlocks one image per second and sharper tumor tracking

A group of researchers from the Technion and the United States reports a breakthrough in MRI scanning in a paper published in Nature Communications. The researchers developed an innovative method that accelerates and enhances MRI scans for breast cancer imaging, a disease diagnosed in approximately 2.3 million people each year, most of whom are women.

The new method, called ELITE, combines artificial intelligence with advanced mathematical models, enabling dynamic MRI with unprecedented speed and accuracy. This international study brings together expertise in engineering, MRI physics, artificial intelligence and clinical radiology.

Dr. Eddy Solomon of the Technion’s Faculty of Biomedical Engineering, the paper’s lead author, explains that the study focuses on dynamic MRI, a critical technology in breast cancer diagnosis. Dynamic MRI is used primarily for screening populations at high risk for breast cancer and is characterized by exceptionally high sensitivity, with more than 90% accuracy, compared with approximately 50%–60% for ultrasound and mammography combined. However, MRI technology faces a major challenge: Producing highly detailed images usually requires longer scan times, making it difficult to track the flow of contrast material through the examined tissue.

Physicists Solve a “Quantum-Only” Problem Using an Ordinary Laptop

A problem once touted as requiring a quantum computer has now been solved on a laptop.

Using advanced mathematical techniques and sophisticated software, physicists at the Center for Computational Quantum Physics (CCQ) at the Simons Foundation’s Flatiron Institute and collaborators at Boston University showed that a conventional computer can successfully simulate a notoriously difficult quantum system previously claimed to be beyond the reach of classical computing.

Heisenberg Said Matter Is Made of Forms, Not Objects — Plato Said It 2,300 Years Earlier

Werner Heisenberg (1901–1976) was one of the founders of quantum mechanics — author of the uncertainty principle (1927) and winner of the 1932 Nobel Prize in Physics. He was also among the most philosophically engaged physicists of the century. In his late teens he read Plato’s Timaeus in the original Greek (his father was a professor of Greek), and the dialogue’s central idea stayed with him: that the smallest constituents of matter are not material objects but mathematical forms.

In Physics and Philosophy (1958), Heisenberg argued that modern physics \.

Intelligence Without Brains: A Radical New Idea

What if intelligence doesn’t require a brain? Biologist Michael Levin argues that intelligence is not confined to neurons, but exists on a continuum of goal-directed behavior and problem-solving across a wide range of species and systems. Using a framework he calls the “cognitive light cone,” Levin explores diverse forms of intelligence extending all the way down to the cellular level. His research suggests that cells communicate through electrical networks, enabling them to make collective decisions and adapt to unexpected challenges, evidenced by engineered tadpoles capable of seeing through eyes located on their tails. Levin radically challenges the conventional wisdom even further, proposing that forms of intelligence may extend beyond biology to molecular systems and maybe even the weather.

00:00 What is intelligence?
01:03 The field of diverse intelligence.
01:33 Intelligence at the cellular level.
02:08 The cognitive light cone.
03:01 The intelligence of groups of cells.
03:52 The bioelectric language of cells.
04:20 The mind of the body.
04:23 Cells that solve problems.
05:17 The tadpole experiment.
06:25 The cognitive spectrum.
06:48 Can you train a hurricane?
07:03 A new science of intelligence.
07:28 Beyond human biases.

——–
Quanta Magazine is an editorially independent publication supported by the Simons Foundation. We focus on developments in mathematics, theoretical physics, theoretical computer science and the basic life sciences.

READ free math and science articles on the Quanta website: https://www.quantamagazine.org.

LEARN about the Simons Foundation: www.simonsfoundation.org.

FOLLOW our social channels:

Mathematical modeling helps advance use of magnetic particles in targeted drug-delivery systems

A Florida State University computational scientist is paving the way for future medical breakthroughs by developing mathematical models and simulations to predict the behavior of a unique drug-delivery method, which aims to deploy treatments directly to targeted sites in the body.

Florida State University Associate Professor of Scientific Computing Bryan Quaife is part of a multi-institutional team of engineers, mathematicians and computational scientists conducting foundational research essential to the design of a drug-delivery system that could reduce medication side effects while increasing treatment efficacy. Their research expands on work proposing the use of magnetic particles to guide cell-like drug carriers toward a specific target, like a tumor.

This work, which was published in Physical Review Letters, reveals how tiny particles moving inside microscopic drug carriers can gradually stress and eventually rupture the enclosing membrane. These findings could help engineers design smarter drug-delivery systems to protect therapeutic cargo during transport and release it on demand at the desired location.

Under the Hood of Vibe Coding: A Deep Dive Into the Holographic Mathematic Miracle

If you’ve been in tech circles lately, you’ve probably heard of “Vibecoding.” Most people treat it like an industry joke—lazy developers throwing sloppy prompts at a screen until an app magically pops out. To traditional gatekeepers, it looks like dangerous, uncompilable chaos.

The “vibe” isn’t a loose, careless emotion. It’s data. Specifically, it is the human-facing interface for what advanced computer science calls Intent Orchestration.

I just published a definitive deep dive into the actual math, physics, and mechanics under the hood of this movement. We break down exactly why the traditional “Filing Cabinet” architecture of multi-agent AI is fundamentally broken, and how Holographic AI Frameworks are the solution.

We are stepping into an era of Decentralized Coherence that liberates creators from traditional development bottlenecks, transforming your role from a low-level syntax translator into a High-Dimensional Intent Architect.

The era of manual syntax is drawing to a close. The computer has finally spent enough time engineering its systems to understand our language.

But make no mistake—if your structural thinking is sloppy, your application will still fail.

The quality of your thinking is the new syntax.

Chinese AI Models Undercut OpenAI and Anthropic by Up to 9x in Pricing

And don’t assume “cheaper” means “worse.” On the SWE-bench Pro—the gold standard benchmark for coding agent capabilities—Zhipu’s GLM 5.2 scored a 62.1, beating OpenAI’s GPT-5.5 at 58.6.


Running the same AI workload through Anthropic’s Claude costs $4,811. Running it through Zhipu’s GLM model costs $544. That’s nearly a 9x price difference for equivalent work, and enterprise customers have started doing the math.

Chinese AI companies are undercutting OpenAI and Anthropic so aggressively on price that the two most prominent US AI firms are now scrambling to respond. OpenAI is reportedly considering major token price cuts, and Anthropic is expected to follow. The timing could not be worse: both companies are preparing for public market debuts.

A comparison of workload costs across leading AI models paints a stark picture. Anthropic’s Claude rings in at $4,811 per workload. OpenAI’s ChatGPT comes in lower at $3,357, but still far above the Chinese alternatives. DeepSeek prices the same workload at $1,071. Moonshot’s Kimi model does it for $948. And Zhipu’s GLM sits at just $544.

Germany’s New Photonic NPU Just Made NVIDIA’s Billion Dollar GPUs Look Like TRASH!

Photonic chips are no longer just a lab experiment, and in this video, we break down why a new photonic NPU could become one of the biggest shifts in AI hardware, data centers, and supercomputing. Instead of using electricity and transistors like a traditional GPU, this new class of processor uses light to perform computation, opening the door to dramatically faster matrix math, far lower energy use, and almost no on-chip heat. From the growing power crisis in AI infrastructure to the limits of silicon, Moore’s Law, and the memory wall, this story explores why photonic computing is suddenly becoming one of the most important technologies to watch. If you’re interested in photonic chips, optical computing, AI chips, NPUs, GPUs, data center efficiency, and the future of semiconductor technology, this video gives you the full picture. We also explore what makes these chips different from conventional silicon. The video covers photons instead of electrons, wavelength-division multiplexing, optical interference, thin-film lithium niobate, and why companies like Q.ANT are now deploying photonic processors in real supercomputing environments instead of just talking about them on research slides. We look at Q.ANT’s Native Processing Unit at the Leibniz Supercomputing Centre in Germany, the jump from first-generation to second-generation performance, and why benchmarks showing huge gains in throughput, AI inference, and energy efficiency are making people take photonic hardware much more seriously. More importantly, this is not just another faster chip story. It is about whether the AI industry can keep scaling without running straight into an energy wall. With GPUs demanding more power, more cooling, and more data movement every year, photonic co-processors may be the first real alternative that changes the economics of compute itself. The technology still has serious challenges, especially memory and optical-electrical conversion, but this may be the moment when computing with light stopped sounding like science fiction and started becoming real infrastructure.

Google Just Released What Comes After AGI — A Million Times More Powerful Than AGI!

Google DeepMind just revealed what could come after AGI, and it may be far more powerful than most people realize. In its new paper “From AGI to ASI,” DeepMind explains why human-level AI may not be the finish line, but the starting point for artificial superintelligence. In this video, we break down what AGI and ASI really mean, why Shane Legg and Marcus Hutter’s involvement matters, and how DeepMind defines superintelligence as something that can outperform massive organizations of top human experts across nearly every domain. We also explore the four possible roads from AGI to ASI: scaling, new AI architectures, recursive self-improvement, and multi-agent AI collectives. One of the most shocking ideas is that you may not need an AI smarter than a human. 100 million human-level AI agents working together could already become something far beyond us. But even superintelligence has limits. Physics, computation, mathematics, uncertainty, data, energy, and regulation could all shape what happens next. Is AGI really the end goal, or just the beginning?

#GoogleDeepMind #AGI #ASI #ArtificialIntelligence #Superintelligence #AI

Ben Goertzel Just Revealed When AGI Will ACTUALLY Happen | Ep. 38

Ben Goertzel, the godfather of AGI research and CEO of SingularityNet, just dropped some mind-blowing insights about artificial general intelligence that will change how you think about AI forever. This isn’t your typical AI hype this is raw truth from someone who’s been building AGI for decades.

In this deep dive conversation, Ben reveals the shocking reality behind current AI limitations, why decentralized AI infrastructure is crucial for humanity’s future, and his honest timeline for when we’ll actually achieve AGI. Plus, he shares what it’s like running a global AI empire while living on a remote island accessible only by ferry.

Key Topics Covered:
The real timeline for AGI development.
Why current AI models aren’t actually intelligent.
How SingularityNet is building decentralized AI infrastructure.
The ASI Alliance and the future of artificial superintelligence.
Ben’s daily routine managing hundreds of AI researchers globally.
Why math and music drive breakthrough AI thinking.

⏰ Timestamps:
0:00 — Introduction to Ben Goertzel.
2:30 — Daily life of an AGI pioneer.
8:45 — Managing a global AI empire.
15:20 — The truth about current AI limitations.
25:10 — SingularityNet and decentralized AI
35:40 — When will AGI actually happen?
45:30 — The future of artificial superintelligence.
58:15 — Closing thoughts.

-️ Subscribe for more exclusive interviews with crypto and AI pioneers who are building the future. Hit the notification bell so you never miss our latest episodes!

💬 What do you think about Ben’s AGI timeline? Drop your thoughts in the comments below!

/* */