AI Research
50 articles
Nvidia Director Mark Stevens Donates $200 Million to USC for AI Research
Mark Stevens, a director at Nvidia, and his wife Mary are donating $200 million to the University of Southern California. The funds will support advancements in...
MIT Professor Advances AI Through Game Theory and Strategic Reasoning
Gabriele Farina, an assistant professor at MIT, combines game theory with machine learning to enhance decision-making algorithms. His research includes developi...
Mark Stevens Donates $200 Million to USC for AI Research and Education
Mark Stevens, a billionaire and director at Nvidia, along with his wife Mary, has pledged $200 million to the University of Southern California. This donation a...
Quality of Data is Crucial for Advancing Physical AI and World Models
The effectiveness of AI systems, particularly in physical applications, is increasingly hindered by the prevalence of junk data. As AI companies seek vast amoun...
%22%2F%3E%3Crect%20width%3D%22800%22%20height%3D%22450%22%20fill%3D%22url(%23glow)%22%2F%3E%3Ctext%20x%3D%22400%22%20y%3D%22275%22%20text-anchor%3D%22middle%22%20fill%3D%22%2360a5fa%22%20font-family%3D%22system-ui%2C-apple-system%2Csans-serif%22%20font-size%3D%22132%22%20font-weight%3D%22800%22%20letter-spacing%3D%22-6%22%20opacity%3D%220.9%22%3Ecn%3C%2Ftext%3E%3Ctext%20x%3D%22400%22%20y%3D%22318%22%20text-anchor%3D%22middle%22%20fill%3D%22%2360a5fa%22%20font-family%3D%22system-ui%2C-apple-system%2Csans-serif%22%20font-size%3D%2211%22%20font-weight%3D%22600%22%20letter-spacing%3D%223%22%20opacity%3D%220.38%22%3ECONTEXT%20NEWS%3C%2Ftext%3E%3C%2Fsvg%3E)
Elon Musk confirms xAI used OpenAI models to train Grok during court testimony
Elon Musk testified that his company xAI employed distillation techniques on OpenAI models to develop its AI, Grok. This admission comes amid Musk's lawsuit aga...
NSA Evaluates Anthropic's AI Model for Cybersecurity Vulnerabilities in Microsoft Software
The National Security Agency is assessing Anthropic's Mythos AI model to identify cybersecurity flaws in widely used software, including Microsoft products. Off...
MIT President Highlights Importance of Curiosity-Driven Science and Research Funding
Sally Kornbluth, President of MIT, emphasized the critical role of curiosity-driven science for the future of the U.S. during a podcast discussion. She expresse...
Exploring the Origins of Goblins in Folklore
The article delves into the historical and cultural origins of goblins in folklore. It examines how these mythical creatures have been portrayed across differen...
MIT and IBM establish new lab to advance AI and quantum computing research
MIT and IBM have launched the MIT-IBM Computing Research Lab to enhance their collaboration in AI and quantum computing. The lab aims to develop new computation...
AI Detects Pancreatic Cancer Early in CT Scans, Study Shows
A study indicates that an AI system can identify pancreatic cancer in CT scans nearly 475 days before a formal diagnosis. This advancement could lead to earlier...
Meta introduces Muse Spark AI model as part of strategy to enhance advertising revenue
Meta has launched its new AI model, Muse Spark, aiming to strengthen its position in the competitive AI market. The company plans to monetize this technology by...
Ineffable Intelligence secures $1.1 billion to develop AI that learns independently
Ineffable Intelligence, a new AI lab founded by David Silver, has raised $1.1 billion to create a 'superlearner' that utilizes reinforcement learning to acquire...
Discussion on AI Collaboration and Complex Task Management
The Odd Lots podcast features a discussion with METR's president Chris Painter and technical staff member Joel Becker about the organization's efforts to assess...
Divided Opinions on the Future of Quantum Computing in Finance
Quantum computing is seen as a potential game-changer in various fields, including finance, drug discovery, and machine learning. However, experts believe that ...
SWE-bench Verified Discontinues Measurement of Advanced Coding Skills
SWE-bench Verified has ceased its evaluation of frontier coding capabilities due to limitations in accurately assessing complex programming skills. This decisio...
Exploring AI Model Evaluation and Complex Task Performance with METR
METR, or Model Evaluation and Threat Research, focuses on assessing AI models' capabilities in performing complex tasks autonomously. The organization emphasize...
DeepSeek's New AI Model V4 Does Not Challenge US Dominance in AI
DeepSeek has introduced its new AI model V4, which is designed to be cost-effective. However, it does not significantly reduce the competitive gap between the U...
arXivLabs Introduces Collaborative Framework for New Features
arXivLabs is a new initiative that enables individuals and organizations to collaborate on developing features for the arXiv platform. The project emphasizes va...
MIT and Partners Launch Comprehensive Dataset of Olympiad-Level Math Problems
Researchers from MIT, KAUST, and HUMAIN have created MathNet, the largest dataset of Olympiad-level math problems, featuring over 30,000 expert-authored problem...
DeepSeek introduces new AI models aiming to compete with leading technologies
DeepSeek has unveiled two versions of its latest large language model, DeepSeek V4, which features significant architectural improvements and a mixture-of-exper...
Goldman Sachs Report Highlights Importance of 'World Models' in Advancing AI Technology
Goldman Sachs emphasizes the need for 'world models' in artificial intelligence, which could lead to significant advancements beyond current large language mode...
NASA's upcoming space telescope and AI advancements intensify demand for GPUs
NASA plans to launch the Nancy Grace Roman space telescope in September 2026, which will generate significant data for astronomers. This increase in data from m...
Tencent Launches Advanced AI Model Amid Competitive Landscape
Tencent Holdings has introduced its latest AI model, Hy3, which is designed to enhance complex reasoning and coding capabilities. This launch follows the compan...
MIT researchers develop method to improve AI confidence estimates
Researchers at MIT's CSAIL have created a technique called RLCR that enhances AI models' ability to express uncertainty in their answers. This method significan...
NeoCognition secures $40 million seed funding to develop self-learning AI agents
NeoCognition, a startup focused on creating self-learning AI agents, has raised $40 million in seed funding. Led by Yu Su, the company aims to enhance the relia...
Jacob Andreas and Brett McGuire receive 2026 Edgerton Faculty Achievement Award
Jacob Andreas and Brett McGuire, both associate professors at MIT, have been awarded the 2026 Harold E. Edgerton Faculty Achievement Award for their exceptional...
OpenAI Launches AI Model for Accelerating Drug Discovery
OpenAI has introduced GPT-Rosalind, an AI model designed to enhance drug discovery processes. This initiative reflects the growing interest among tech companies...
Automated Agent Exposes Vulnerabilities in Major AI Benchmarks
An automated scanning agent has revealed that several prominent AI benchmarks can be easily exploited to achieve high scores without actually completing tasks. ...
Research Shows Smaller AI Models Can Identify Cybersecurity Vulnerabilities Effectively
Recent tests indicate that smaller AI models can outperform larger counterparts in identifying cybersecurity vulnerabilities. The findings suggest that while la...
New AI Lab from Harvard Aims to Raise $100 Million for Memory Technology
Engramme, a new AI lab founded by Gabriel Kreiman and Spandan Madan, is seeking $100 million in funding. The lab aims to develop technology that enhances human ...
New AI Training Method Enhances Model Efficiency by Compressing During Learning
Researchers have developed a technique called CompreSSM that allows AI models to be compressed during training, improving efficiency without sacrificing perform...
Meta launches Muse Spark, its first AI model from new Superintelligence Labs
Meta has introduced Muse Spark, its first AI model developed by the newly established Meta Superintelligence Labs. The model shows competitive performance in be...
Meta Launches Muse Spark, Its First AI Model from New Superintelligence Team
Meta Platforms has introduced Muse Spark, its first AI model developed by the newly formed Superintelligence Labs. This model will be used to enhance the Meta A...
Databricks co-founder honored with ACM Prize, discusses future of AI
Matei Zaharia, co-founder and CTO of Databricks, has been awarded the ACM Prize in Computing for his contributions to big data technology, particularly through ...
MIT researchers develop system to enhance data center storage performance
Researchers at MIT have created a software system called Sandook that improves the performance of storage devices in data centers by addressing three sources of...
MIT study finds AI workers often perform at minimal standards in various tasks
Research from MIT indicates that while AI technology is improving, it often meets only basic performance standards in workplace tasks. The study analyzed 41 AI ...
Google plans to build AI data centers in space by 2027, aiming for solar energy efficiency
Google's CEO Sundar Pichai announced plans to construct AI data centers in space, starting in 2027, as part of Project Suncatcher. This initiative aims to harne...
Introduction of Qwen3.6-Plus Aims to Enhance Real-World Agent Capabilities
Qwen3.6-Plus has been launched with the goal of improving the functionality of real-world agents. This advancement is significant as it seeks to bridge the gap ...
MIT develops AI model to identify and quantify atomic defects in materials
Researchers at MIT have created an AI model that can classify and quantify atomic defects in materials using noninvasive neutron-scattering data. This advanceme...
A Reflection on the Initial 40 Months of Advancements in AI Technology
The article discusses the significant developments in artificial intelligence over the past 40 months. It highlights the transformative impact of AI on various ...
Anthropic accidentally exposes unreleased model details and internal data due to security lapse
Anthropic has revealed sensitive information about an upcoming AI model and an exclusive CEO retreat due to a security oversight that left nearly 3,000 internal...
MIT engineers develop AI model to design proteins based on motion patterns
MIT engineers have created an AI model called VibeGen that designs proteins by specifying desired motion patterns rather than just their shapes. This innovative...
Memory Chip Stocks Decline Following Google's AI Storage Algorithm Announcement
Memory chip stocks continued to fall after Google revealed a new algorithm aimed at improving storage efficiency for artificial intelligence applications. This ...
Technical Report Released for ARC-AGI-3 Project
A technical report detailing the ARC-AGI-3 project has been published, providing insights into its objectives and methodologies. This report is part of ongoing ...
Data Centers Shift from AC Power to DC Power Systems
Data centers are increasingly adopting direct current (DC) power systems instead of alternating current (AC). This transition is driven by the potential for imp...
AI Agents Show Improved Capabilities but Struggle with Reliability
Recent evaluations highlight that while AI agents are becoming more capable, their reliability remains a significant concern. A study assessing various AI model...
MIT Postdoc Sojun Park Reflects on Research and Community Impact
Sojun Park, a postdoctoral fellow at MIT's Center for International Studies, has focused on the global diffusion of AI technologies and their political implicat...
MIT Professor Discusses Impact of Operations Research and AI on Healthcare and Education
Dimitris Bertsimas, a professor at MIT, delivered the 54th annual James R. Killian Faculty Achievement Award Lecture, highlighting his contributions to operatio...
Experts Discuss the Future of AI at MIT Conference
A recent conference at MIT focused on the implications and future direction of artificial intelligence. Keynote speakers Karen Hao and Paola Ricaurte emphasized...
MIT and Hasso Plattner Institute launch AI and Creativity Hub for interdisciplinary research
MIT and the Hasso Plattner Institute have established the AI and Creativity Hub to enhance collaboration between computing and design. This 10-year initiative a...