Computing and robotics

IBM quantum computer helped calculate how tritium binds to molten salt in fusion reactors

IBM quantum computer helped calculate how tritium binds to molten salt in fusion reactors

Researchers combined a Heron quantum processor with supercomputers to calculate the electronic structure of FLiBe clusters, a material being considered for tritium production in fusion reactors. The results show that the quantum calculation of parts of the problem was
Will human-level artificial intelligence remain out of reach?

Will human-level artificial intelligence remain out of reach?

Computer scientist Peter J. Denning challenges the quest for artificial general intelligence. He argues that human intelligence includes physical, social, and cultural knowledge that cannot be fully represented by a computer. This is the thesis presented in a new book,
Can an algorithm make decisions without a doctor? Israel to test autonomous medical artificial intelligence

Can an algorithm make decisions without a doctor? Israel to test autonomous medical artificial intelligence

The Ministry of Health and the Innovation Authority are launching a regulatory sandbox for medical artificial intelligence with a high level of independence. The key test is not just whether the algorithm is right most of the time, but whether it knows when it is not.
Artificial intelligence within a virtual world. Illustration: depositphotos.com

Not just predicting words: World models that try to teach artificial intelligence to predict the results of its actions

Language models learn primarily from text and images, but systems operating in the physical world need to understand motion, space, and causality. World models attempt to address this challenge through simulation, prediction, and learning from experience.
Space-time network. Illustration: depositphotos.com

Not a time machine: This is how researchers made a quantum system look like it was going backwards

Researchers have developed a feedback method that can cancel and even reverse the arrow of time observed in a monitored quantum system. The method could be used to extract energy from measurements and prepare quantum states.
Artificial intelligence helps scientific research, but it may make science monotonous. Illustration: depositphotos.com

More articles, fewer ideas? The danger of using artificial intelligence in science

Artificial intelligence tools greatly increase researchers’ productivity and citations, but may narrow the range of questions that science explores. An opinion piece in Nature warns against the creation of a scientific monoculture.
Is Artificial Intelligence Reliable? Illustration: Dr. Roy Tsezana via Gemini

The AI ​​Arms Race: Why Anthropic Wants to Hit the Brakes

A new Anthropic document shows a sharp jump in Claude's abilities in writing code, fixing systems, and developing research — and raises the question of whether humanity can slow down the pace of progress over time.
Cover of the United Nations University report on the environmental cost of AI's energy consumption, carbon emissions, water and land. Credit: UNU-INWEH

UN report: Artificial intelligence consumes not only electricity, but also water, land and metals

A new report from the United Nations University warns that the growth of AI is creating a vast environmental footprint: data centers, cooling, power generation, mineral mining, e-waste, and inequality between countries that deploy the technology and those that do not.
Designing chips with the help of artificial intelligence. Illustration: depositphotos.com

Artificial intelligence accelerates the search for new chip materials

A new artificial intelligence system scans a vast space of chemical combinations and suggests gallium-based semiconductor materials with desirable electronic properties. The researchers emphasize that the materials still require experimental validation, but the method
Robot in a biological laboratory. Illustration: depositphotos.com

An artificial intelligence system wrote scientific software that outperformed human code

Researchers from Google and Harvard have developed ERA, an artificial intelligence system that runs and improves thousands of versions of code for scientific tasks. In experiments, it has improved models for predicting coronavirus hospitalizations, integrating single-cell RNA data, and predicting neural activity.
A digital workspace where employees operate AI tools, automation, and smart agents.

Organizations are reshaping themselves in the age of AI agents: fewer employees, more automation, and huge salaries

ClickUp CEO presents a new organizational model for the age of artificial intelligence agents: fewer employees, more automation, new roles, and salaries that can reach a million dollars a year. Is this the future of work?
An illustration of collaboration between human mathematicians and artificial intelligence in mathematical research.

Artificial intelligence helps solve mathematical problems that have remained open for decades

Artificial intelligence is no longer just a writing or coding tool. New cases show that ChatGPT and Claude are helping solve open mathematical problems, including Erdoğan problems, and are also forcing senior mathematicians to update their
A U.S. soldier patrols next to a Ghost Robotics Vision 60 prototype robot at Nellis Air Force Base in Nevada. Credit: Tech. Sgt. Cory Payne / US Air Force

Fear of aiding the enemy is hindering international agreements on the military use of artificial intelligence

Countries fear that limiting military AI will help their enemies. The REAIM conference exposed the difficulty of regulating autonomous weapons and AI on the battlefield.
A 300mm-diameter quantum chip wafer. IBM and the U.S. Department of Commerce plan to use Anderon to establish a first-of-its-kind dedicated factory in the U.S. for the production of quantum chip wafers. Credit: IBM

US and IBM establish first dedicated quantum chip factory

IBM and the U.S. Department of Commerce announced plans to establish Anderon, the first dedicated 300mm quantum chip wafer manufacturing facility in the U.S. The venture is supported by $1 billion in planned funding under the
Smart factory. Illustration: depositphotos.com

Dassault Systèmes and OMRON will connect virtual twins for industrial automation

The two companies are seeking to connect the design, simulation and operation of production lines. The goal is to enable manufacturers to test production systems in a virtual environment before actual installation, and then use real-world data to improve operations.
NASA's High Performance Spaceflight Computing Processor. The processor is designed to survive harsh space conditions and provide a leap in processing capability for future spacecraft. Credit: NASA/JPL-Caltech.

NASA is developing a new “brain” for spacecraft: an AI processor that will allow them to operate more independently

The new processor is designed to allow spacecraft to analyze information, respond to unexpected situations, and make certain decisions without waiting for instructions from Earth. According to NASA, initial tests indicate a leap in performance compared to computers
Artificial intelligence collaborates with human researchers. Illustration: depositphotos.com

Science seeks transparency rules for the use of artificial intelligence in research

A new international initiative seeks to establish a uniform standard for reporting on the use of artificial intelligence in scientific research. The question is not just whether researchers used artificial intelligence, but what exactly it did in the paper.
AI-based deepfakes challenge the ability to distinguish between real footage and artificially created or altered images, sounds, or videos. Digital forensics researchers are developing tools to identify fakes and document the source of media. Illustration: depositphotos.com

Real-time deepfakes: The new battle for trust in images and video

Professor Hani Farid of the University of California, Berkeley, a pioneer in the field of digital forensics, is at the center of a new Science article on the race to identify fake content. The deepfakes of 2026 are no longer just suspicious videos online: they are
Hackers are using artificial intelligence to break into Mexican government servers.

Artificial intelligence helped amateur hackers break into government servers in Mexico

A report by Gambit describes how a small group of unprofessional attackers used Claude Code to penetrate government systems, understand the structure of servers, and build a mechanism to forge credentials. The incident illustrates how the tool
A combination of the brain and artificial intelligence. Illustration: depositphotos.com

Harvard and Hebrew University to Collaborate on NeuroAI Research

Harvard's Kempner Institute and the Edmond and Lily Safra Center for Brain Science at the Hebrew University will conduct joint academic activities in the field that connects brain research and artificial intelligence.
Futuristic artificial intelligence. Illustration: depositphotos.com

Is Artificial Intelligence Just a Tool? Study Warns It Could Change Our Perception of Reality

Research from the University of Exeter shows that the regular use of creative AI does not end with receiving information. When such systems are integrated into thinking, memory and self-narrative processes, they can reinforce false beliefs, distorted memories
Prof. Michael Rabin. Credit: Harvard John A. Paulson School of Engineering and Applied Sciences

Michael Rabin, one of the founding fathers of computer science, transformed mathematical ideas into the foundations of the digital world

The Turing Award and Israel Prize winner made fundamental contributions to automata, probabilistic algorithms, complexity, and cryptography, and helped build the academic infrastructure of computer science in Israel.
Combining quantum computing and artificial intelligence enables more accurate and efficient prediction of complex and chaotic systems. Illustration: depositphotos.com

Quantum computing upgrades artificial intelligence in predicting chaotic systems

New hybrid method achieves 20% higher accuracy and requires fewer resources, using a quantum computer in the training phase of AI models
The suspicion of Claude Mitos. Illustration: Dr. Roy Tsezana

Anthropic delays release of Claude Myth due to concerns about hacking and exceeding restrictions

Anthropic's new model has demonstrated exceptional capabilities in detecting weaknesses, attempting to escape a closed testing environment, and hiding anomalies from its operators — which is why the company is opting for a gradual and cautious release.
The roots of the mystery: When the mobility formula fails. Credit: PEXELS

The paradox that moves against the rules: Researchers from Ariel University cracked a 120-year-old engineering puzzle

Prof. Nir Schwalb and Dr. Oded Medina showed that not only the number of links and axes determines the movement in the mechanism, but also their geometric arrangement in space, and they also discovered "hypo-paradoxical" mechanisms that lock even though, according to
Graphene battery. Illustration: depositphotos.com

Quantum batteries are no longer a distant dream: a new prototype was able to charge, store, and discharge energy

Australian researchers have demonstrated the feasibility of a quantum battery that works at room temperature. The technology is still far from commercialization, but the new experiment reinforces the belief that in the future it will be possible to develop energy storage systems that charge much faster.
Robot swarm. Courtesy of Harvard University.

A little randomness improves the navigation of robot swarms in crowded environments

Researchers from the United States and the Netherlands have shown that controlled randomness in the movement path of robots reduces traffic jams and improves the rate of task completion, without the need for a powerful central computer or complex coordination.
A dangerous combination of artificial intelligence and cyber. Illustration: depositphotos.com

Between deterrence and defense

A study at Tel Aviv University examines why the US abandoned “cyber deterrence” in 2018 and moved to Defend Forward and Persistent Engagement

Detect biases already at the data stage, before they reach the algorithm

Dr. Yuval Moskowitz from the Faculty of Computer and Information Sciences is researching how to use Data Provenance, the “curriculum vitae” of data, to detect when and how bias arises in data processing processes and reduce it with minimal change.

IBM quantum computer accurately simulates magnetic material — and verified against real physics experiment

For the first time, IBM researchers and their partners from leading universities have succeeded in performing a quantum simulation of a magnetic material whose accuracy has been tested against the material's behavior in nature, an important step towards making quantum computing a practical tool in materials science.
A sheepdog on duty managing a flock of sheep. Photo: Georgia Tech University

Sheepdogs reveal a better way to guide robot swarms

Researchers at the Georgia Institute of Technology studied how dog and handler teams deal with unpredictable herds in herding dog competitions, finding principles that go far beyond herding sheep.
Disinformation. Illustration: depositphotos.com

AI agents can autonomously coordinate propaganda campaigns without human guidance

Lead researcher: "Coordinated AI agents can create the appearance of consensus, influence trend dynamics, and accelerate the spread of messages. In democratic contexts, especially around elections or crises, such capabilities could
Prof. Charles Bennett. Photo: IBM

The 2025 Turing Award was awarded to Charles Bennett and Gilles Brassard for the foundations of quantum information science.

The two were recognized for a series of breakthroughs that turned quantum mechanics into a new basis for information processing, including the BB84 protocol, quantum cryptography, quantum teleportation, and the development of the entire field as an independent scientific discipline.
Artificial intelligence will take all the jobs. Illustration: depositphotos.com

New study: Who is really vulnerable to artificial intelligence, and who will be able to adapt?

An analysis by GovAI and the Brookings Institution shows that it is not enough to measure which professions are exposed to AI; we also need to examine who has a real chance of moving to a new job if their role changes or disappears.
People are different, technology is common, which can lead to uniformity. Illustration: Avi Blizovsky via DALEE

Researchers warn: AI chatbots could unify and standardize human thinking

The researchers note that within human groups and societies, cognitive diversity is an important factor for creativity and problem-solving. However, they argue that this diversity is reduced when billions of people use the same limited number of chatbots.
Different methods of classifying images by artificial intelligence. From the study

New AI method provides explanation for image classification through component decomposition

A new method in Explainable AI separates general elements in an image from elements that bias the decision, helping to detect biases and shortcuts
Optical communication. Illustration: depositphotos.com

The computer of the next century: This is how we imagined the photonic revolution in 1997, and what has really happened since then

A historical article by Avi Blizovsky from 1997 about optical computers, in a restored and updated version, with a look back at the place of photonics in modern computing.
Dashboard, artificial intelligence. Illustration: depositphotos.com

Artificial Intelligence in Israel Between Local Challenges and Global Trends: Neeman Index Dashboard Presents Continuous Measurement

The Shmuel Neaman Institute has launched a frequently updated dashboard that monitors real-time demand for AI jobs on LinkedIn, trends in academic theses, and international indicators — to help policymakers understand where the field of artificial intelligence is headed.
Downscaling attempts to reconstruct a more detailed map using variables measured at high resolution. Illustration: depositphotos.com

AI enhances satellite imagery to know what's happening "between the pixels"

Satellites provide a wide picture, but sometimes with limited resolution: the pixel is too large, and the details on the ground are lost. This is where downscaling comes in: using computational models to estimate temperature in more detail.
Artificial intelligence soldiers. Illustration: depositphotos.com

The United States Secretary of War is preparing for war in the age of artificial intelligence.

A new strategic document calls on the US military to operate in a “wartime” format, adopt agent-based artificial intelligence at an unprecedented pace – and entrust autonomous systems with critical roles on the battlefield
Quantum computing, sometimes a bit represents both a zero and a one. Illustration: depositphotos.com

Quantum computing without intermediate measurements: Demonstration of a “tool” for error-proof logic

Quantum error correction requires encoding information so that errors are detected before they destroy the calculation. The research proposes performing logical operations independently of measurements during the algorithm's run, thus allowing calculations to be performed on multiple qubits.
Bots are involved in elections. Illustration: depositphotos.com

Swarms of AI personas: The silent threat to democracy that already speaks to us online

A paper in Science warns that coordinated networks of synthetic identities can integrate into communities, build trust, and sway political discourse—without looking like bots.
A network of artificial intelligence agents. Illustration: depositphotos.com

A network of artificial intelligence agents “talks to itself” – and researchers are listening from the inside

Open source project that allows agents to perform everyday tasks has been connected to a dedicated social network; millions of posts were created within days, raising questions about emergent behavior, privacy, and humanization
Artificial intelligence and creativity. Illustration: depositphotos.com

Who is more creative in “branching thinking” tests—humans or large language models?

Scientific Reports examined the ability to generate ideas and originality, and the practical question is when the model expands creativity and when it reduces it.
Upgrading skills for the future world of work. Illustration: depositphotos.com

Huge study: Less skilled jobs experience more pressure to upgrade skills in job ads

Analysis of 167 million U.S. job postings finds “profound” skill shifts needed more in occupations with lower wages and education requirements, and that small businesses are upgrading skill requirements to catch up with larger businesses
Nvidia CEO Jensen Huang at the Davos 2026 conference. Photo courtesy of the World Economic Forum

The “five-layer cake” of artificial intelligence

Nvidia CEO Jensen Huang, Microsoft CEO Satya Nadella, Google DeepMind CEO and other executives described the different needs of each layer - from infrastructure to applications.

EU softens GDPR to accelerate AI revolution

After years of GDPR making data use limited and complicated, EU advances easing of restrictions that would expand access to data for AI training; industry welcomes, rights groups warn of loss of transparency and citizen control