Large Language Models News and Research

RSS

AI Tools Play Distinct Roles in Creativity: Study Maps How LLMs, T2I, and T2-3D Boost Design Innovation

This study systematically compares how different generative AI tools—LLMs, text-to-image (T2I), and text-to-3D (T2-3D) models—support distinct stages of the creative process. The findings reveal a practical framework for aligning each AI type with ideation, visualization, or prototyping tasks.

17 Jun 2025

AI’s Blind Spot: Machines Fail to Capture the Full Bloom of Human Experience

A new study finds that artificial intelligence, even with advanced training, cannot represent human concepts grounded in sensory and motor experiences, such as “flower,” with the same richness as humans do. AI language models excel at abstract terms but fall short when concepts require multisensory and embodied understanding.

5 Jun 2025

AI Chatbots Struggle With Mental Health Drug Side Effects, Georgia Tech Study Finds

Researchers at Georgia Tech developed a new framework to evaluate how well AI chatbots detect and respond to adverse psychiatric medication reactions in user conversations. The study found that while LLMs mimic human empathy, they often struggle to provide clinically accurate, actionable advice for mental health side effects.

3 Jun 2025

VizTrust Reveals How Users Build And Lose Trust In AI Chatbots In Real Time

Researchers at Binghamton University have developed VizTrust, a novel analytics tool that tracks and visualizes the dynamic nature of user trust during real-time interactions with AI chatbots. The system uses machine learning to analyze emotional and behavioral cues, offering actionable insights for improving conversational AI.

3 Jun 2025

Aaron Courville Champions Collaborative AI To Solve Society’s Biggest Challenges

Aaron Courville, incoming director of IVADO and co-founder of Mila, champions an optimistic and pragmatic vision for artificial intelligence, emphasizing its potential to solve societal challenges. Rather than fearing AI, Courville advocates for responsible, collaborative development to maximize its benefits across fields from climate action to public policy.

28 May 2025

University of Idaho Secures $4.5M Grant to Revolutionize Research Administration With Generative AI

The University of Idaho has received a $4.5 million NSF grant to develop generative AI tools that streamline and automate research administration. This initiative aims to reduce administrative burdens, improve compliance, and share solutions across the research community.

26 May 2025

Experts Say AI Now Meets Criteria for Free Will and Moral Agency

A new study by Frank Martela in AI and Ethics argues that advanced generative AI agents meet all three philosophical conditions of free will, raising the prospect of AI moral responsibility. As AI autonomy increases, urgent ethical questions emerge regarding how developers 'parent' and program these systems.

26 May 2025

AI Grades Science Homework Fast, But Misses the Mark Without Human Guidance

A University of Georgia study tested how large language models like Mixtral grade middle school science responses. While fast, these AIs often use keyword shortcuts that undermine accuracy without human-designed rubrics.

8 May 2025

Scientists Pinpoint When AI Turns From Helpful to Harmful

Researchers at George Washington University have derived an exact, reproducible formula that predicts when an AI language model's output will shift from accurate to misleading. The tipping point occurs when the model’s internal attention dynamics snap under pressure.

5 May 2025

Experts Warn AI-Enhanced Scams Are Getting Harder to Spot

Artificial intelligence tools like ChatGPT 4o are revolutionizing scams by enabling the creation of hyper-realistic images, voices, and documents. Experts warn that traditional detection methods may no longer suffice and call for education, vigilance, and technologies like blockchain to counter deception.

5 May 2025

AI Jam Session Unites Scientists To Tackle Scientific Challenges With Frontier Models

Researchers from Los Alamos and Sandia National Laboratories joined nearly 1,000 scientists nationwide in an AI jam session to test the capabilities of cutting-edge AI models in solving real-world scientific problems. The event assessed AI tools from OpenAI and Anthropic across disciplines from physics to geospatial analytics.

25 Apr 2025

AI And Supercomputing Team Up To Target Cancer’s Toughest Proteins

Researchers at Argonne National Laboratory are harnessing the Aurora exascale supercomputer and advanced AI to accelerate cancer drug discovery and tackle “undruggable” proteins. Their efforts combine structural biology, molecular simulation, and experimental validation to identify novel cancer therapies.

23 Apr 2025

ChatGPT Earns a B in Engineering Course but Fails at Critical Thinking

ChatGPT earned a B in a semester-long undergraduate aerospace engineering course, acing structured math problems but struggling with open-ended reasoning tasks. The study reveals its limitations in critical thinking and suggests a need for course redesign to promote deeper learning.

23 Apr 2025

ChatGPT Can Pass Engineering Homework, But Struggles With Deeper Thinking

ChatGPT outperformed on simple math tasks but faltered on open-ended questions in an undergraduate control systems course, earning a low B overall. The study warns that relying solely on AI may lead to passing grades without deep learning.

18 Apr 2025

Binghamton Launches AI and Society Institute to Harness Supercomputing Power for Public Good

Binghamton University will lead a new Institute for AI and Society, backed by $5M in New York state funding and powered by Empire AI, the nation’s top academic AI supercomputer. Researchers will use this boost to tackle online antisemitism, climate tech, and accessible robotics.

17 Apr 2025

Scientists Stress Test OpenAI Models to Accelerate Scientific Discovery Across U.S. National Labs

More than 1,600 scientists across nine U.S. national labs took part in a first-of-its-kind AI Jam Session to test OpenAI’s cutting-edge models on complex scientific tasks. Brookhaven researchers explored AI’s potential in spectroscopy, accelerator physics, nanomaterials, and more.

17 Apr 2025

Penn Engineers Unleash Programmable Photonic Chip That Trains AI at the Speed of Light

Penn Engineers have unveiled the first programmable chip capable of training nonlinear neural networks using only light, revolutionizing photonic AI. This innovation promises rapid, energy-efficient AI training and marks a significant step toward fully light-powered computing.

16 Apr 2025

MIT AI Breakthrough Automates Molecular Design and Synthesis with Unmatched Accuracy

MIT researchers developed Llamole, a multimodal AI system combining large language models with graph-based AI to automate molecular design and synthesis. It significantly boosts success rates and simplifies drug discovery by interpreting natural language queries and generating viable synthesis plans.

10 Apr 2025

How AI Coding Tools Open the Door to Hackers Through Fake Packages

UTSA researchers revealed that large language models (LLMs) frequently hallucinate non-existent software packages in generated code, posing a serious cybersecurity threat. Their analysis found that up to 1 in 5 packages suggested by open-source LLMs could lead to malicious code injection.

9 Apr 2025

Can AI Be Fair in Medicine? Study Finds Socioeconomic Bias in Generative Models

A large-scale analysis by Mount Sinai researchers reveals that generative AI models can alter medical recommendations based solely on a patient’s socioeconomic and demographic background. The study calls for early AI assurance protocols to prevent bias and ensure equitable healthcare delivery.

7 Apr 2025