AI Research News

RSS

AI-Powered Chemistry Delivers a Breakthrough for Photodynamic Therapy

Researchers developed an AI-driven discovery pipeline called AAPSI that designed and experimentally validated a new photosensitizer, HB4Ph, with strong singlet oxygen generation and red-shifted light absorption for photodynamic therapy. The study also introduced a large public database of photosensitizer-solvent pairs and showed that AI can speed multi-objective molecular design from years to days.

20 Mar 2026

Metasurface AI Breakthrough Brings Faster Medical Imaging

A new hybrid optical-digital AI system uses a massive fixed metasurface and a tiny trainable backend to match top digital models across vision tasks, while delivering striking speed gains in demanding applications such as cancer slide analysis.

19 Mar 2026

How Could VR and Brain Signals Personalise Arachnophobia Treatment?

Researchers at TU Graz developed a prototype VR system called VRSpi that uses EEG and heart rate data to measure anxiety in real time and automatically adjust spider exposure intensity for people with arachnophobia. The approach could make virtual exposure therapy more personalised and effective by replacing subjective therapist judgments with objective neurophysiological signals.

19 Mar 2026

How Could AI Help Social Entrepreneurs Win The Funding They Need?

A new research chapter argues that artificial intelligence could help social entrepreneurs overcome persistent funding barriers by improving how investors and lenders assess financial risk, social impact, and business potential. The authors suggest that AI-driven fintech tools, impact data, and alternative funding models may create more inclusive and transparent pathways to finance for mission-led enterprises.

19 Mar 2026

How is One Scientist Using Human Cognition to Build Better AI?

Assistant Professor Gaurav Malhotra studies how human cognition, perception, and decision-making can both inform and be explained by artificial intelligence, blending psychology, neuroscience, and computational modeling. At UAlbany, his research explores how humans adapt to complex environments, manage cognitive effort, and represent the world efficiently, with implications for safer and more energy-efficient AI systems.

19 Mar 2026

How Could AI Help DUNE Unlock The Secrets of Neutrinos?

Researchers at Rice University brought together international experts to explore how artificial intelligence and machine learning could help the Deep Underground Neutrino Experiment analyze vast datasets, detect rare signals, and improve detector operations. The workshop highlighted how AI could accelerate neutrino research by strengthening simulation, monitoring, and data management across one of the world’s most ambitious physics collaborations.

19 Mar 2026

Vision Enabled AI Medical Scribes Boost Accuracy to 98 Percent and Transform Clinical Documentation

A study shows that vision-enabled AI medical scribes combining audio and visual data significantly improve clinical documentation accuracy, capturing key medication details often missed by audio-only systems. The findings suggest multimodal AI could reduce clinician workload and errors, while still requiring human oversight for safe implementation.

19 Mar 2026

AI Decodes Ancient Roman Board Game And Reveals Lost Rules From 2,000 Years Ago

Researchers used AI simulations to reconstruct the likely rules of an ancient Roman board game from a limestone artefact, identifying it as a strategic “blocking” game. The study demonstrates how AI can decode archaeological mysteries by linking wear patterns to plausible gameplay mechanics.

19 Mar 2026

Why Human AI Teams Fail and How Cognitive Alignment Unlocks Better Collaboration

A new study finds that failures in human–AI collaboration often stem from misalignment in how tasks, roles, and expectations are shared, rather than limitations in AI capability alone. The research highlights “hybrid cognitive alignment” as a dynamic process essential for effective teamwork, requiring ongoing adaptation, training, and calibrated trust.

18 Mar 2026

AI Coding Tools Still Struggle With Basic Tasks as Study Reveals Reliability Gaps in Structured Outputs

A new benchmarking study shows that even advanced AI models achieve only moderate accuracy when generating structured code outputs, with performance dropping further in complex multimodal tasks. The findings highlight persistent limitations in reliability and consistency, reinforcing the need for human oversight in AI-assisted software development.

18 Mar 2026

Over Half of U.S. Teens Use AI to Create Sexualized Images, Study Reveals

More than half of U.S. adolescents reported using AI “nudification” tools, with substantial proportions also experiencing non-consensual creation and sharing of sexualized images. The findings highlight widespread normalization of AI-driven sexual image manipulation among teens, raising significant concerns around consent, exploitation, and digital harm.

18 Mar 2026

ChatGPT’s Scientific Judgement Questioned As Study Exposes Reliability Gaps

A large-scale study of over 700 scientific hypotheses found that ChatGPT achieved moderate accuracy but showed significant inconsistency, particularly struggling to correctly identify false claims. Even as performance improved slightly over time, results remained only modestly better than chance when accounting for guessing.

17 Mar 2026

Sheepdog Strategies Inspire New Algorithm That Improves Swarm Control Under Uncertainty

Researchers analyzing sheepdog trials identified how timing, pressure, and controlled “noise” enable handlers to guide unpredictable flocks. These principles were translated into computational models showing improved coordination strategies for robot swarms and autonomous systems under uncertainty.

17 Mar 2026

Bird Flocking Algorithm Reduces AI Hallucinations and Improves Summary Accuracy

A novel preprocessing framework inspired by bird flocking organizes sentences into structured clusters, enabling large language models to generate more accurate and less redundant summaries. By reducing noise and preserving key information, the approach significantly improves factual fidelity across thousands of documents.

17 Mar 2026

How AI-Driven Science Is Revolutionizing Environmental Decision-Making Worldwide

Artificial intelligence is reshaping environmental science by integrating large, complex datasets to enable predictive, system-level insights across water, soil, air, and waste systems. It is evolving from a supporting analytical tool into an active partner in scientific discovery, enabling real-time monitoring and more precise environmental management.

17 Mar 2026

Researchers Warn AI Talking Toys May Misread Emotions And Need Stronger Safety Rules

Researchers found that AI-powered conversational toys may support language development but can misinterpret emotions, disrupt social play, and encourage children to form potentially misleading emotional bonds.

16 Mar 2026

AI Weather Models Predict Hurricane Paths Well But Struggle With Storm Physics

Researchers evaluated leading AI weather models and found they can accurately predict tropical cyclone tracks but often struggle to reproduce the physical wind structures that determine storm impacts. The findings highlight both the promise of rapid AI forecasting and the need for continued scientific oversight to ensure physically realistic storm simulations.

16 Mar 2026

AI Chatbots May Standardize Human Thinking And Threaten Cognitive Diversity, Researchers Warn

esearchers warn that widespread use of AI chatbots could standardize how people write, reason, and express ideas, potentially reducing global cognitive diversity. They argue that incorporating more linguistic, cultural, and reasoning diversity into large language model training could both preserve human creativity and improve AI reasoning.

16 Mar 2026

AI Defence System Stops Sophisticated 5G Cyber-Attacks In Under A Tenth Of A Second

By combining reinforcement learning with a real-time digital twin of live mobile infrastructure, researchers have created an AI defence system capable of detecting and shutting down evolving cyber-attacks on 5G networks in milliseconds, offering a glimpse of how future 6G systems may defend themselves.

16 Mar 2026

AI Weather Agent Zephyrus Turns Natural Language Questions Into Instant Forecast Analysis

Researchers developed Zephyrus, an AI weather agent that converts natural language questions into code to analyze complex outputs from AI-driven weather and climate models. The system allows users to query meteorological datasets in plain language and receive interpretable explanations, potentially accelerating research and improving access to climate data.

15 Mar 2026

AI Research News

AI-Powered Chemistry Delivers a Breakthrough for Photodynamic Therapy

Metasurface AI Breakthrough Brings Faster Medical Imaging

How Could VR and Brain Signals Personalise Arachnophobia Treatment?

How Could AI Help Social Entrepreneurs Win The Funding They Need?

How is One Scientist Using Human Cognition to Build Better AI?

How Could AI Help DUNE Unlock The Secrets of Neutrinos?

Vision Enabled AI Medical Scribes Boost Accuracy to 98 Percent and Transform Clinical Documentation

AI Decodes Ancient Roman Board Game And Reveals Lost Rules From 2,000 Years Ago

Why Human AI Teams Fail and How Cognitive Alignment Unlocks Better Collaboration

AI Coding Tools Still Struggle With Basic Tasks as Study Reveals Reliability Gaps in Structured Outputs

Over Half of U.S. Teens Use AI to Create Sexualized Images, Study Reveals

ChatGPT’s Scientific Judgement Questioned As Study Exposes Reliability Gaps

Sheepdog Strategies Inspire New Algorithm That Improves Swarm Control Under Uncertainty

Bird Flocking Algorithm Reduces AI Hallucinations and Improves Summary Accuracy

How AI-Driven Science Is Revolutionizing Environmental Decision-Making Worldwide

Researchers Warn AI Talking Toys May Misread Emotions And Need Stronger Safety Rules

AI Weather Models Predict Hurricane Paths Well But Struggle With Storm Physics

AI Chatbots May Standardize Human Thinking And Threaten Cognitive Diversity, Researchers Warn

AI Defence System Stops Sophisticated 5G Cyber-Attacks In Under A Tenth Of A Second

AI Weather Agent Zephyrus Turns Natural Language Questions Into Instant Forecast Analysis

Trending Stories

Latest AI News