AI Research

Twitter

Facebook

Copy Link

Key Facts

FocusScaling, Reasoning, Safety, Multimodality

Key OrgsGoogle DeepMind, Stanford HAI, MILA, MIT CSAIL, Allen AI

Key EventsNeurIPS, ICML, ICLR

Top News

AI Co-Pilot for Bionic Hands Transforms How Amputees Control Prosthetics with Intuitive Grasping

University of Utah researchers developed an AI co-pilot for bionic hands that dramatically improves control for amputees. The system uses proximity and pressure sensors combined with artificial intelligence to enable natural, intuitive grasping. Success rates jumped from 10-20% to 80-90% in tests involving fragile objects, while significantly reducing the cognitive burden on users.

Health4 sources

AI Co-Pilot for Bionic Hands Transforms How Amputees Control Prosthetics with Intuitive Grasping

Health4 sources

26,000 AI researchers admit they can't explain how models work as interpretability crisis deepens

At this year's NeurIPS conference, a record 26,000 AI researchers confronted an unsettling reality: no one fully understands how today's most advanced AI systems actually work. As companies like Google and OpenAI pursue diverging approaches to AI interpretability, the field grapples with the black box problem and an evaluation crisis that threatens trust in increasingly powerful systems.

Science2 sources

26,000 AI researchers admit they can't explain how models work as interpretability crisis deepens

Science2 sources

Unconventional AI raises $475M to build brain-inspired chips that solve AI's energy crisis

Former Databricks VP Naveen Rao has launched Unconventional AI with a massive $475 million seed round, backed by Jeff Bezos and top venture firms. The startup aims to build analog AI chips inspired by biological systems to tackle AI's growing energy problem. With AI demand set to outpace global energy capacity within 3-4 years, the company is exploring neuromorphic and analog computing approaches that could deliver 1,000 times better efficiency than current silicon.

Startups6 sources

Unconventional AI raises $475M to build brain-inspired chips that solve AI's energy crisis

Startups6 sources

AI models achieve breakthrough accuracy in diagnosing brain tumors from MRI scans

Two groundbreaking studies reveal AI's potential to transform brain tumor diagnosis. York University researchers developed an AI model achieving over 85% accuracy in distinguishing tumor progression from radiation necrosis, while Thomas Jefferson University's automated machine learning system reached 97% accuracy in differentiating pituitary macroadenomas from parasellar meningiomas using preoperative MRI scans.

Health2 sources

AI models achieve breakthrough accuracy in diagnosing brain tumors from MRI scans

Health2 sources

Anthropic launches AI Interviewer tool to collect user feedback on how people actually use AI

Anthropic unveiled its Interviewer tool, an AI-powered chatbot that conducts adaptive interviews to understand how people integrate AI into their work. The research pilot program runs for a week, with early findings showing 86% of professionals believe AI saves time, though concerns about job identity and data security persist among creative and scientific professionals.

Technology4 sources

Anthropic launches AI Interviewer tool to collect user feedback on how people actually use AI

Technology4 sources

Google launches Gemini 3 Deep Think, its most advanced reasoning AI for complex problem-solving

Google has rolled out Gemini 3 Deep Think exclusively to AI Ultra subscribers at $250 per month. The advanced thinking mode uses parallel reasoning to tackle complex math, science, and logic problems, scoring 41% on Humanity's Last Exam benchmark. The feature explores multiple hypotheses simultaneously to deliver refined solutions.

Technology8 sources

Google launches Gemini 3 Deep Think, its most advanced reasoning AI for complex problem-solving

Technology8 sources

Zanskar uses AI to discover hidden geothermal energy site, first industry find in decades

Geothermal startup Zanskar has identified a commercially viable hidden geothermal system in Nevada using AI technology, marking the first such discovery by the energy industry in decades. The site, called Big Blind, reaches 250°F at 2,700 feet below the surface and could provide constant clean power without greenhouse gas emissions as electricity demands rise globally.

Technology2 sources

Zanskar uses AI to discover hidden geothermal energy site, first industry find in decades

Technology2 sources

RBC targets up to C$1 billion in enterprise value from AI by 2027 as deployment accelerates

Royal Bank of Canada is on track to generate between C$700 million and C$1 billion in incremental enterprise value from AI by 2027. The bank has deployed AI across multiple functions, from its proprietary ATOM foundation model for credit adjudication to RBC Assist reaching over 30,000 employees, positioning itself among global banking leaders in AI research.

Business2 sources

RBC targets up to C$1 billion in enterprise value from AI by 2027 as deployment accelerates

Business2 sources

Google funds $10M chair at University of Toronto to honor Nobel Prize-winner Geoffrey Hinton

Google is investing $10 million CAD to establish the Hinton Chair in Artificial Intelligence at the University of Toronto, honoring Nobel laureate Geoffrey Hinton's foundational work on neural networks. The university will match the donation, creating a $20 million initiative to recruit visionary scholars and drive AI innovations across medicine, engineering, and scientific discovery.

Business2 sources

Google funds $10M chair at University of Toronto to honor Nobel Prize-winner Geoffrey Hinton

Business2 sources

OpenAI Acquires Neptune to Enhance AI Model Training Workflows and Research Capabilities

OpenAI has entered a definitive agreement to acquire Neptune, a Poland-based startup that provides tools for analyzing AI model training progress. The stock transaction, reportedly valued under $400 million, will integrate Neptune's experiment tracking and monitoring platform into OpenAI's training infrastructure. Neptune, used by over 60,000 AI researchers, will wind down external services by March 4, 2025.

Technology6 sources

OpenAI Acquires Neptune to Enhance AI Model Training Workflows and Research Capabilities

Technology6 sources

Anthropic Engineers Reveal AI Productivity Gains Come With Deep Concerns About Job Security

Anthropic conducted an internal study of 132 engineers to understand how AI is transforming work at the company. While employees report significant productivity gains and can delegate up to 20% of tasks to Claude Code, they're also expressing concerns about losing technical competence, reduced collaboration with colleagues, and fears of becoming irrelevant as AI capabilities advance.

Business3 sources

Anthropic Engineers Reveal AI Productivity Gains Come With Deep Concerns About Job Security

Business3 sources

Gradium Emerges from Stealth with $70M to Revolutionize Voice AI with Ultra-Low Latency Models

Paris-based AI voice startup Gradium launches with $70 million seed funding to develop audio language models that deliver ultra-realistic voice interactions with dramatically reduced latency. The company aims to make voice the primary interface between humans and machines.

Startups3 sources

Gradium Emerges from Stealth with $70M to Revolutionize Voice AI with Ultra-Low Latency Models

Startups3 sources

Anthropic's AI agents autonomously generate $4.6M in smart contract exploits, raising alarm

Anthropic revealed that its Claude AI models and OpenAI's GPT-5 autonomously identified and exploited smart contract vulnerabilities worth $4.6 million in simulated tests. The AI agents discovered zero-day vulnerabilities in recently deployed contracts, with exploitation costs falling to just $1.22 per scan. The findings underscore how rapidly advancing AI capabilities are outpacing blockchain security defenses.

Technology4 sources

Anthropic's AI agents autonomously generate $4.6M in smart contract exploits, raising alarm

Technology4 sources

NVIDIA Unveils Alpamayo-R1: First Open-Source Reasoning AI Model for Autonomous Driving Research

NVIDIA announced Alpamayo-R1, the world's first open reasoning vision-language-action model for autonomous driving research at NeurIPS 2025. The model combines chain-of-thought reasoning with path planning to help vehicles navigate complex scenarios with human-like judgment, marking a significant step toward Level 4 autonomous driving capabilities.

Technology6 sources

NVIDIA Unveils Alpamayo-R1: First Open-Source Reasoning AI Model for Autonomous Driving Research

Technology6 sources

AI Research Tool Shows Social Media Algorithms Can Reduce Political Polarization by 3 Years in Days

Stanford-led researchers developed an AI-powered browser extension that reranks X feeds without platform permission, demonstrating that downranking polarizing content reduced partisan animosity equivalent to three years of attitude change in just 10 days. The breakthrough tool enables independent field experiments on how social media algorithms shape political views during the 2024 election.

Science5 sources

AI Research Tool Shows Social Media Algorithms Can Reduce Political Polarization by 3 Years in Days

Science5 sources

Prev 1…212223…31 Next