All posts tagged: researchers

Why do self-driving cars crash? King’s College London researchers think they have the answer

Why do self-driving cars crash? King’s College London researchers think they have the answer

Published by Jenni Sidey

A self-driving car can make a mistake in seconds, but the reason it happened may stretch far back through a long chain of decisions. That is part of what makes autonomous vehicle crashes so hard to explain, and so hard to prevent. A team at King’s College London says it has developed a new way to tackle that problem. Instead of only estimating how likely a failure is to happen again, the approach is designed to work backward through a crash and identify why a specific failure occurred. That distinction matters as autonomous vehicles appear more often on public roads, including in cities such as London and San Francisco. Collisions and serious road safety breaches have sharpened pressure on manufacturers to explain what went wrong when these systems fail. Current methods can offer only limited answers. They tend to rely on failure statistics, which are useful for measuring risk but weaker at explaining one concrete event. Autonomous vehicles appear more often on public roads, including in cities such as London and San Francisco. (CREDIT: Zoox) …

Ditch the niceties in AI prompts to save energy use, say researchers

Ditch the niceties in AI prompts to save energy use, say researchers

Published by skeptic

ChatGPT now processes around 2.5 billion queries every day Alina Vytiuk / Alamy Stock Photo UN researchers are urging people to be less polite to artificial intelligences after a report found that cutting words from prompts could reduce ChatGPT’s energy consumption by up to 25 per cent. Removing “please”, “thank you” and other unnecessary words from AI prompts could save 87 to 98 gigawatt-hours of electricity per year, the report from the UN University Institute for Water, Environment and Health (UNU-INWEH) found. That is the equivalent of the annual residential electricity use of up to 760,000 people in sub-Saharan Africa. To reduce their energy consumption and carbon footprint, people should write concise prompts, avoid getting sucked into conversation loops and refrain from starting relationships with AI, the researchers said. “We are not saying be rude to your AI. But don’t fall into the interaction trap and don’t go falling in love with it either,” says Kaveh Madani at UNU-INWEH. The large language models behind AI chatbots process text in small units known as tokens. Madani …

Researchers may have discovered the key to understanding human consciousness

Researchers may have discovered the key to understanding human consciousness

Published by Jenni Sidey

For centuries, people have wondered what separates being awake from being asleep, dreaming, or unconscious. Scientists have searched for clues throughout the brain, hoping to identify the signals that help create conscious experience. Now, researchers at Ludwig Maximilian University of Munich have uncovered a previously unknown brain rhythm that may offer an important piece of the puzzle. The discovery centers on the thalamus, a small structure buried deep within the brain. Often described as a relay hub, the thalamus helps route information between different brain regions. It also plays a critical role in attention, awareness, and perception. While scientists have long suspected that it helps regulate consciousness, direct evidence has been difficult to obtain. In a new study, researchers identified a distinctive pattern of activity in the human thalamus that appears only during wakefulness and rapid eye movement, or REM, sleep. The signal disappears completely during non-REM sleep, a state associated with reduced awareness and little conscious experience. The findings suggest that this newly discovered rhythm could serve as a measurable biological marker of conscious …

Columbia University researchers discover new clues to Alzheimer’s origins

Columbia University researchers discover new clues to Alzheimer’s origins

Published by Jenni Sidey

A fragile cleanup system sitting on the surface of brain cells may help explain one of Alzheimer’s disease’s oldest mysteries. It may reveal how ordinary tau protein first turns into the twisted filaments tied to memory loss and cognitive decline. That is the central finding from a Columbia University team that traced the earliest stages of tau damage to a neuron-specific protein disposal system called the neuroproteasome. When that system was disrupted, tau rapidly misfolded into paired helical filaments. This is the same broad kind of abnormal structure seen in the brains of people with Alzheimer’s disease. The work points to a possible starting point for tau pathology. Additionally, it connects two of the disease’s biggest risk factors: aging and the APOE4 gene variant. “These prior studies could not capture how tau misfolds in the first place in Alzheimer’s disease but understanding how tau aggregation begins is critical if we want to create therapies that prevent neurodegeneration before it starts,” says the study’s senior author, Kapil Ramachandran, assistant professor of neurological sciences at Columbia University. …

Researchers find evidence of Zwan-Wolf effect on Mars

Researchers find evidence of Zwan-Wolf effect on Mars

Published by skeptic

Using data from the NASA spacecraft MAVEN, a team at West Virginia University have found evidence of the atmosphere of planets being protected from solar winds, even without strong magnetic fields, also known as the Zwan-Wolf effect. The Zwan-Wolf effect was first described in 1976 and had only been observed in planetary magnetospheres, not in atmospheres. But findings reported in Nature Communications have observed the effect in Mars’ atmosphere, bringing new understanding to how our Sun interacts with planetary bodies in our solar system. The sun emits a continuous flow of plasma, known as solar wind. When this plasma flow encounters large objects like planets or comets, it is deflected around them, similar to the flow of water around a rock in a stream. “However,” said Christopher Fowler, planetary scientist at West Virginia University, “because the water in that stream is relatively dense, physical collisions between water molecules bumping into each other and the rock determine how the water is diverted. In contrast, the environment in space is so tenuous that solar wind particles do …

Researchers automated LLM reasoning strategy design and cut token usage by 69.5%

Researchers automated LLM reasoning strategy design and cut token usage by 69.5%

Published by skeptic

Test-time scaling (TTS) has emerged as a proven method to improve the performance of large language models in real-world applications by giving them extra compute cycles at inference time. However, TTS strategies have historically been handcrafted, relying heavily on human intuition to dictate the rules of the model’s reasoning.  To address this bottleneck, researchers from Meta, Google, and several universities have introduced AutoTTS, a framework that automatically discovers optimal TTS strategies. This automated approach allows enterprise organizations to dynamically optimize compute allocation without manually tuning heuristics.  By implementing the optimal strategies discovered by AutoTTS, organizations can directly reduce the token usage and operational costs of deploying advanced reasoning models in production environments. In experimental trials, AutoTTS managed inference budgets efficiently, successfully reducing token consumption by up to 69.5% without sacrificing accuracy. The manual bottleneck in test-time scaling Test-time scaling enhances LLMs by granting them extra compute when generating answers. This extra compute allows the model to generate multiple reasoning paths or evaluate its intermediate steps before arriving at a final response.  The primary challenge for …

Mathematical AI helps researchers crack 50-year-old problem

Mathematical AI helps researchers crack 50-year-old problem

Published by skeptic

Paul Erdős made many conjectures about numbers in his life Oliver Helbig/Getty Images Just a week after an AI disproved an 80-year-old conjecture and astonished mathematicians, another conjecture that had stood for half a century has fallen, inspired by the same techniques, but this time written entirely by humans. Last week, an unreleased AI model from OpenAI disproved an important conjecture first posed by Hungarian mathematician Paul Erdős, called the unit distance problem. The puzzle, which Erdős considered his “most striking contribution to geometry” and which many mathematicians had failed to unravel, concerns the number of similar-sized connections you can make between dots arranged on a flat surface. Erdős had set an upper ceiling on this number, which many experts had assumed was correct. But the AI model showed that this number could in fact be much larger, using an obscure trick from algebraic number theory to make complex structures with extremely high dimensions, which could then be used to arrange the dots in a very different arrangement than humans had considered. The result took …

These Ebola Researchers Are Stuck in US Due to Trump’s Funding Cuts

These Ebola Researchers Are Stuck in US Due to Trump’s Funding Cuts

Published by skeptic

As the world struggles to contain the rapidly growing Ebola outbreak in the Democratic Republic of Congo’s Ituri province, a vital network of research centers has been unable to help on the ground. The reason: The Trump administration slashed its funding last year, in part due to conspiracy theories about the origins of Covid-19. Established in 2020 by the National Institutes of Health, the Centers for Research in Emerging Infectious Diseases (CREID) Network was conducting research into viruses that emerge from wildlife and spill over to people, including the family of viruses that Ebola belongs to. The network operated 10 sites around the world where these types of disease outbreaks are likely to occur, including in Central and East Africa. (The network was also researching hantavirus, a disease that saw a recent rare outbreak on a cruise ship.) NIH provided CREID with approximately $82 million in funding over five years, and its funding was up for renewal in 2025. But last June, the centers received a stop-work order stating that their research had been deemed …

Researchers discover DNA’s hidden defense against UV radiation

Researchers discover DNA’s hidden defense against UV radiation

Published by Jenni Sidey

DNA sits in sunlight every day, absorbing ultraviolet radiation that can set off the kind of chemical changes linked to mutations, aging, and cancer. Yet most of the time, the molecule avoids catastrophe. New simulations suggest that this resilience depends not on one built-in escape route, but on a crowded, ultrafast network of molecular reactions. These reactions can dump harmful energy almost as soon as it appears. The research focused on guanine and cytosine, two of DNA’s four chemical bases, arranged in short stacked segments that mimic key features of the double helix. By tracking their behavior at the atomic scale, the team found that once UV energy enters the system, the excited state quickly shifts into charge-transfer arrangements. Then, it relaxes through several competing pathways, many involving tightly linked movements of electrons and protons. That picture is more complicated than the simpler models that have often dominated the field. Instead of following a single preferred track, the excited DNA fragments explored multiple routes back to stability. Often, this happened within femtoseconds. Each one of …

Former Google and Apple Researchers Launch a Startup to Build AI’s Missing Feedback Loop

Former Google and Apple Researchers Launch a Startup to Build AI’s Missing Feedback Loop

Published by skeptic

A group of AI researchers who previously worked at Google DeepMind, Apple, OpenAI, and Meta Superintelligence Labs announced on Wednesday they’re launching a new startup called Trajectory, which aims to help companies regularly improve their AI products by training on real-world user interactions. Trajectory wants to build a platform for AI that can learn continuously, a capability that researchers have long held up as a major barrier to further AI progress. OpenAI, Google, and Anthropic have found success training increasingly capable versions of AI models, especially for domains such as coding, math, and science. However, these systems stop getting smarter after their training is done. While there have been some recent breakthroughs in continual learning, tech companies have generally struggled to make AI products that learn from their errors in real time. In December 2025 at NeurIPS, one of the largest annual AI research conferences, Turing award winner Richard Sutton argued that continual learning is essential for building superintelligent agents. Trajectory has raised a $15 million seed round at a $115 million post-money valuation, led …