A self-improving superintelligence could trigger an acceleration in technological capabilities that would signal the end of the human era, writes Anthony Cuthbertson. According to some AI tech leaders, that moment has already arrived
Category: robotics/AI – Page 19
Smart sensor identifies present molecules by remembering the past
Most sensors are designed to do only one thing: detect what passes through them. But what if a sensor could do more? To create a new generation of technology, researchers have looked to living systems for inspiration. If a sensor could detect molecules, could it also remember previous interactions and selectively respond to them?
Researchers have developed a device that does exactly that on a tiny scale, laying the groundwork for a new generation of smart molecular sensors.
In an article recently published in ACS Nano, researchers from SANKEN at the University of Osaka and collaborating institutions created an autonomous solid-state nanopore that can sense molecules, generate electrical signals and retain memories of recent events without external control. Unlike conventional nanopores, which act as passive channels, the new device continuously changes its own structure through chemical reactions, creating a dynamic sensing environment that responds to molecules passing through it.
Johns Hopkins’ AI-powered, voice-controlled robot performs autonomous surgery
Researchers at Johns Hopkins University have shown that an artificial intelligence-powered robot—trained on videos of previous surgeries—could learn how to perform the procedure itself, without help from a human operator.
According to the university, the autonomous system was able to adapt to the natural variations in anatomy between simulated patients, and react correctly to unplanned events in real-time—such as the introduction of blood-like dyes that obscured the surrounding tissue. It was also able to recover on its own from initially missed instrument placements.
PRX Intelligence publishes its first papers
The American Physical Society’s newest highly selective, open access journal, PRX Intelligence, has published its inaugural papers. The studies demonstrate how artificial intelligence and machine learning methods can be used to advance scientific knowledge and capabilities across the physical sciences — from neural networks that streamline molecular simulations to data-driven learning schemes that accelerate quantum embedding workflows.
As AI and machine learning transform the physical sciences, PRX Intelligence is designed to provide a multidisciplinary platform for research pioneering the development and application of these approaches. Building on the foundation of Physical Review X, the journal publishes open access articles expected to have substantial and lasting impact. It welcomes studies that apply AI and machine learning across theory, simulation, and experimentation in physics and related fields — including computer science, mathematics, engineering, materials science, chemistry, biology, and earth and environmental sciences. Relevant topics include discovery and synthesis, physics-informed learning, data-driven approaches, machine learning pipelines for observational platforms, and more. Articles have flexible formats and lengths and may include research papers, perspectives, roadmaps, tutorials, and more.
PRX Intelligence will waive article publication charges for manuscripts submitted or transferred before Jan. 1, 2027. And like all other APS journals, it will always waive these charges for researchers in low-and middle-income countries. Sign up for email updates to keep up with the latest news from the journal.
Zoox gets the green light to offer paid rides
The Department of Transportation has approved a request from the driverless vehicle company Zoox to start offering rides to paying customers. A California-based company owned by Amazon, Zoox has been testing its driverless electric vehicle with free rides. Now, it has approval to commercially deploy up to 5,000 vehicles over the next two years, starting in Las Vegas. Unlike Google’s Waymo, which adds driverless technology to existing vehicles, Zoox’s vehicles look like a box on wheels and don’t have steering wheels or pedals. Find “NPR News Now” wherever you listen to podcasts. Host: Jeanine Herbst/NPR Reporter: Camila Domonoske/NPR Producer: Michael Zamora/NPR
Follow NPR elsewhere, too: • Instagram: https://www.instagram.com/npr/ • TikTok: https://www.tiktok.com/@npr • Facebook: https://www.facebook.com/NPR
How Generative AI Should Transform Clinical Decision Support
Consider a primary care physician seeing a patient aged 58 years for routine follow-up. The electronic health record (EHR) alerts that the patient is eligible for statin therapy. The physician overrides it, as clinicians do for the vast majority of alerts. The system verifies what is computationally easy (eg, age, lipid values, and risk score) but ignores what the clinician needs: has this patient been offered statins before? Did they decline, and if so, why (cost concerns, fear of adverse effects, preference for lifestyle modification)? Have they tried statins previously and experienced muscle pain? If they were prescribed a statin, did they ever pick it up from the pharmacy? What did they write in that patient portal message 2 months ago when they mentioned reading online that statins cause memory problems? The answers are scattered across notes, dispensing records, and portal messages. The alert identifies eligibility but not the patient’s decision state or the barriers to action.
Consistent with established definitions, clinical decision support (CDS) includes tools that provide knowledge and patient-specific information to support health decisions and is not limited to guideline adherence.1,2 This Perspective focuses on clinician-facing CDS organized around a defined decision; generic note drafting, inbox management, and open-ended chart summarization are excluded unless they directly support that decision. A prior reason for declining statin therapy is relevant because it changes the next action, not eligibility. Deterministic methods remain preferable when criteria and outputs are explicit; large language models (LLMs) may extend them through flexible synthesis and adaptive presentation.
Early medical LLM applications have focused on drafting replies and summarizing charts.3,4 The larger opportunity is to revisit a long-standing trade-off between clinical fidelity and computational tractability. Health information technology has historically represented complex narratives and knowledge through structured fields and rules because they were computable.5 LLMs do not provide the first access to narrative text; their incremental value is the flexibility to extract, synthesize, and communicate across heterogeneous sources.
AI tech proactively prevents subway door entrapment accidents
A research team led by Professor Jo Woon Chong of the School of Electronic and Electrical Engineering at Sungkyunkwan University (SKKU), in collaboration with researchers from KAIST and Texas Tech University in the United States, has developed the Passenger Movement Estimation System (PMES). This AI-based system predicts passenger movements using CCTV footage to prevent subway door entrapment accidents before they occur.
Overcoming the limitations of conventional reactive methods, in which sensors trigger only after a passenger has entered the danger zone, the system proactively identifies risks before passengers reach the boarding area. The findings are published in IEEE Transactions on Intelligent Transportation Systems.
Chong, who has led research at Sungkyunkwan University on human-centered AI, multimodal signal processing and AI-embedded systems, oversaw the study. Hee Jo, the first author and a Ph.D. student, led the data analysis and AI model design.