🔬 AI Research

Academic papers, research breakthroughs, and scientific advances in machine learning and AI.

585 articles

Could AI let us program matter? Hassabis weighs in

In a 2025 conversation with Daniela Rus, Demis Hassabis discusses the possibility that matter could one day be programmed the way computers are. The talk explores that idea and its implications. Source: youtube.com

Hassabis asks whether AI could program the physical world

Google DeepMind cofounder and Nobel laureate Demis Hassabis explores whether AI could let us program the physical world much as we program computers. The coverage looks at what such a capability might involve. Source: youtube.com

Why Zuckerberg let Yann LeCun walk away from Meta AI

With Alexandr Wang's arrival, open source dropped down Meta's priority list, and Yann LeCun left to start his own venture. The coverage examines why Zuckerberg let him go. Source: youtube.com

LeCun's AMI Labs raises $1.03B seed at $3.5B valuation

AMI Labs, co-founded by Yann LeCun after leaving Meta, raised $1.03 billion in a seed round at a $3.5 billion pre-money valuation. The startup is building AI world models. Source: stockanalysis.com

Kernel-Tuned ML Pushes Network Intrusion Detection to Nearly 99% Accuracy

A new study reports that Random Forest and Support Vector Machines with Gaussian and Sigmoid kernels, paired with systematic data preprocessing, reached accuracy of up to 99.87% on benchmark data. The results suggest careful kernel tuning can significantly boost network intrusion detection performance. Source: bioengineer.org

Google unveils Gemini 4 Carbon, its new flagship frontier model

Google has announced Gemini 4 Carbon, describing it as its most capable model to date. The company says it delivers state-of-the-art performance in cybersecurity and long-horizon tasks. Source: deepmind.google

MIT lecture on LLM deployment techniques features Nvidia's Jim Fan

The listing is for Lecture 13 of MIT's EfficientML.ai course (6.5940, Fall 2026) on LLM deployment techniques. It also references a talk on the end game of robotics by Nvidia's Jim Fan. Source: youtube.com

UC Berkeley's free AI agents course draws top speakers

UC Berkeley has put a full course on AI agents online, with every lecture free to watch. Speakers include Denny Zhou on reasoning, Shunyu Yao on the history of agents, and Jim Fan. Source: threads.com

Cyber Valley Days 2024 open a new chapter for AI in Europe

A highlight of the event was the ELLIS Institute Symposium, where leading AI researchers including Yann LeCun shared their views on the next generation of AI. Source: cyber-valley.de

LeCun: LLMs still struggle to reason and plan in the physical world

At the Evident AI Symposium 2026, AI pioneer Yann LeCun told Evident's Alexandra Mousavizadeh that large language models remain limited when it comes to reasoning and planning in the physical world. The appearance is tied to his AMI Labs venture. Source: youtube.com

Hassabis: AI should push education beyond rote learning

Nobel laureate Demis Hassabis argues that AI tools are reshaping the 50-year-old rote-learning model of education. He suggests that memorization should take a back seat as AI changes how people learn. Source: youtube.com

The Download: Anthropic's IPO, smarter power grids and OpenAI's science wins

This edition of MIT Technology Review's newsletter covers Anthropic's IPO and smarter power grids. It also notes that OpenAI has announced three new scientific discoveries across different fields. Source: technologyreview.com

Genie 4 brings real-time interactive world simulation

Genie 4 is a new generation of world simulation that runs interactively in real time. It adds native audio, advanced physics and natural human motion, with steering powered by Gemini models. Source: deepmind.google

Faking a scientific paper with AI: an experiment

The author generated two AI-written scientific papers and submitted them to a publisher's screening process. The experiment reveals how well, or poorly, such checks catch fabricated research. Source: spectrum.ieee.org

AI engineering event features OpenAI, Hugging Face, Andrew Ng and Yann LeCun

The event program includes an AI Engineering Skills Map and a fireside chat with Yann LeCun on world models. Speakers also come from Vocal Bridge and the NYC Tech Talent Pipeline, among others. Source: youtube.com

Trossen tackles physical AI's data bottleneck as Nvidia's Jim Fan weighs in on robotics

The coverage looks at how Trossen is addressing the shortage of training data for physical AI. It also points to a Sequoia Capital talk, 'Robotics' End Game', featuring Nvidia's Jim Fan. Source: youtube.com

Giving LLMs a 'sleep phase' to keep what they learn

Researchers propose a sleep framework for continual learning, adding a dedicated sleep phase that lets large language models absorb new facts over time. The approach also aims to strengthen reasoning while limiting the loss of previously learned knowledge. Source: research.google

Hassabis Revises AGI Timeline, Suggesting It Could Come Sooner

Google DeepMind chairman Demis Hassabis has noticeably shifted his assessment of when artificial general intelligence (AGI) might arrive. The change was voiced in a recent interview. Source: finance.biggo.com

Hassabis Aims to Turbocharge Science and Drug Discovery With AI

A Wall Street Journal profile published October 21 examines Demis Hassabis and his drive to speed up scientific discovery and drug development using AI. The piece raises the question of who will ultimately benefit from those advances. Source: bostonglobalforum.org

NVIDIA Isaac Lab 3.0 Promises Large-Scale Robot Training

The snippet highlights NVIDIA Isaac Lab 3.0 as a tool for training AI-powered robots at scale. The surrounding text is mostly a listing of related videos, including a Sequoia Capital talk with NVIDIA's Jim Fan on robotics. Source: youtube.com