🔬 AI Research

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

585 articles

ML and Deep Learning Speed Up Antimicrobial Resistance Prediction

A scientific review describes how machine learning and deep learning models can rapidly predict antimicrobial resistance using multi-omic datasets. The approach promises faster insight into resistance patterns than traditional methods. Source: bioengineer.org

New Research Probes Theoretical Limits of LM Alignment

Language model (LM) alignment aims to steer model outputs toward human preferences without sacrificing the base model's capabilities. The paper examines the theoretical boundaries of how far this approach can go. Source: machinelearning.apple.com

Hassabis Cautions of an AI 'Oil Spill' Unless the Industry Shifts Direction

DeepMind co-founder Demis Hassabis warns that a disaster comparable to an oil spill could hit the AI field if the industry does not change course. He points out that AI systems are now being used to build better AI. The coverage is based on a brief snippet, so further details are limited. Source: inc.com

ML Model Flags Autism-Linked Variants That Disrupt Splicing

A new study suggests rare genetic variants tied to autism may interfere with the regulation of tiny DNA segments known as microexons. A machine-learning model was used to identify variants that scramble splicing. Source: thetransmitter.org

ML Models Predict Suicide Risk Among U.S. Army Soldiers, Study Finds

A study reports that machine-learning models can estimate the risk of suicide and nonfatal suicide attempts among regular U.S. Army soldiers. The analysis covered more than 580,000 soldier person-historical assessments. Source: usmedicine.com

Review Surveys Quantum Machine Learning for Medical Imaging

Quantum machine learning is being explored as a way to ease the growing computational demands that modern deep learning models place on radiology. A new review examines the architectures proposed for this purpose. Source: quantumzeitgeist.com

Weighted State Souping Boosts Long-Context Retrieval in State Space Models

A paper accepted at the PALM workshop at NeurIPS 2026 examines how to improve long-context retrieval in state space models. The work proposes a weighted 'souping' approach to combining model states for more efficient recall. Source: machinelearning.apple.com

Simple 11-Variable Model Predicts Hospital Death Risk After Traumatic Brain Injury

A simplified random forest model using 11 variables predicted in-hospital mortality among ICU patients who survived the first 24 hours after traumatic brain injury. The model achieved a strong AUC despite its compact input set. Source: bioengineer.org

Can LLM Judges Reliably Assess Personalized Writing Style?

A paper accepted at the 'Can We Trust AI Evaluation?' workshop at NeurIPS 2026 investigates how well LLM judges evaluate personalized writing style. The work probes the reliability of using LLMs as evaluators. Source: machinelearning.apple.com

AlphaGenome 2 sheds light on how cells read DNA

The next-generation DNA sequence model links single-cell biology with human genetics. It predicts how genetic variants affect hundreds of different human cell types. Source: deepmind.google

Hassabis's Talk With Reid Hoffman Offers Two Lessons for Entrepreneurs

In a conversation with Reid Hoffman, Google DeepMind's Demis Hassabis lays out two realities that entrepreneurs should grasp. The first is that artificial intelligence represents a major shift. The snippet is cut off, so the second point is not covered. Source: thewantrepreneurshow.com

Gemini 4 Carbon launches as Google's top model yet

Carbon is billed as the company's most capable model, delivering frontier-level results in cybersecurity and in long-horizon agentic workflows. It targets complex knowledge work and software tasks, and is now available to users. Source: deepmind.google

Daily AI Agents Brief: New Models From Anthropic, OpenAI, Meta and Google

The September 6, 2026 edition of a daily AI agents roundup tracks fresh model releases from Anthropic, OpenAI, Meta, and Google. It also covers progress in AI-assisted software development and head-to-head comparisons of the leading models. Source: aiagentsdirectory.com

NVIDIA Shows GR00T 2 and MANUS Data Gloves in Jim Fan's CoRL 2026 Keynote

In his CoRL 2026 keynote, Jim Fan presented NVIDIA's GR00T 2 and demonstrated how human demonstrations can teach robots new dexterous skills. MANUS data gloves featured in the showcase. Source: youtube.com

Gemini Robotics 2.5 brings better dexterity and safety to real-world robots

Google introduces Gemini Robotics 2.5, its most capable vision-language-action model to date. It is designed to improve dexterity, reliability, and safety for robots operating in real-world environments. Source: deepmind.google

AI Release Analytics: Cadence, Trends and Benchmarks Across Major Labs

An analytics resource tracks how quickly AI models ship, including releases per month, each lab's cadence, and benchmark trajectories. It covers OpenAI, Anthropic, Google, Meta, SpaceXSI, and others. Source: aireleasetracker.com

Everything Google Announced About Gemini at I/O 2026

AI was the clear focus of Google I/O 2026, with Gemini news taking center stage. Mashable rounds up all the Gemini announcements from the event. Source: mashable.com

Google's September 2026 AI Recap: Gemini 4 Argon, Connected Apps, WeatherNext 3

Google's monthly roundup highlights Gemini 4 Argon, new Connected Apps in the Gemini App, and WeatherNext 3. The company notes it has invested in machine learning and AI research, tools, and infrastructure for more than 20 years. Source: blog.google

Google releases Gemini 4 Carbon with sharper knowledge work and coding

Gemini 4 Carbon is now available, promising higher-precision knowledge work, artifact generation, and data analysis. Google also highlights improvements in agentic coding and full-stack development. Source: deepmind.google

New recipe distills Transformer knowledge into Mamba models

State Space Models like Mamba are gaining ground as an alternative to Transformers thanks to their lower memory consumption. The paper presents a recipe for cross-architecture distillation, transferring knowledge from attention-based models into Mamba. Source: machinelearning.apple.com