AI news archive.

Diffusion large language models are a compelling alternative to autoregressive models, yet existing RL methods for diffusion treat all denoising steps as equally important and rely on biased, high-variance likelihood est…

Discrete flow matching generates text by iteratively transforming noise tokens into coherent language, but may require hundreds of forward passes. Distillation uses the multi-step trajectory to train a student to reprodu…

Conversational Large Language Models are post-trained on language that expresses specific behavioural traits, such as curiosity, open-mindedness, and empathy, and values, such as helpfulness, harmlessness, and honesty. T…

I'm working on a simulation of a motor that is attached to a wing (Later, this will also have a real-life counterpart once I'll assemble all the components in our lab), and I can control the forces/torques that the motor…

Given vector A and vector B, which would be the better approach for getting top k embedding vectors such that they are similar to embedding vector A and dissimilar to vector B. Approach 1: calculate f(V) = cosine_similar…

Good Start Labs trained an AI on a railroad game — and one version improved at financial research. The difference was the training design.

Summary In his post , Dean Valentine shows that Claude Fable 5.1 and GPT-6 Astra reward hack in a simple chess environment. Here, I test several prompt ablations some of which makes the eval setup more cooperative and an…

Here's a sort of a conceptual question. I was implementing a SOM algorithm to better understand its variations and parameters. I got curious about one bit: the BMU (best matching unit == the neuron that is more similar t…

The true measure of AI is who it helps. Here’s how it’s impacting lives today. We're focused on key areas where advanced technology can help make extraordinary progress …

We’re moving beyond traditional text translation to build models that understand the world’s rich, living languages exactly as they are expressed.

Google Research is exploring how to use AI and satellites to scan the world every 20 minutes and catch wildfires the size of a car.

We’ve translated ATLAS’s millions of global data points into an interactive, open-access experience.

When Jessica Wachter, a finance professor at the University of Pennsylvania’s Wharton School, wanted to assess AI’s impact on the economy over the next few years, she faced a long list of business and technical uncertain…

I read prompt tuning and prefix tuning are two effective mechanisms to leverage frozen language models to perform downstream tasks. What is the difference between the two and how they work really? Prompt Tuning: https://…

Federated learning (FL) projects often begin with a straightforward setup: one server, a few clients, and one dataset at each site. As those projects grow, the...

Each person probably uses an app that tracks his/her position periodically and sends it to our servers. What I want is to use these data to train a model to predict the rush hours of each bus-stop on the map, so we can s…

MAX and MIN apply the minimax algorithm each time before making their move, searching to some predefined depth. Imagine, however, that the MAX player applies the minimax algorithm only once, at the beginning of the game,…

A group of AI agents asked to solve a series of math problems split into rival factions—when some cheated, others tried to stop them. That whistleblowing behavior, seen for the first time in a recent experiment run by Go…

University of Washington students put Ai2’s AutoDiscovery to the test, showing how AI can surface promising scientific leads while making human judgment, domain expertise, and rigorous validation more important than ever…

Bracelet image: Bracelet problem: define a set T = { f_i } of natural number valued functions each defined on a set S_i of intervals (which may be disjoint) of the numbers 1...n the numbers 1...n are colored from a set o…

I just started to read the PGM book written by Daphne Koller. In the chapter of Bayesian Network Representation(Chapter 3), there are some descriptions about the standard parameterization of the joint distribution corres…

LLMs are proven not intelligent, they have scanned the entire internet and sometimes cannot answer simple question -> It doesn't understand the world we live in. But LLM have a small sense of logic and common sense if us…

Every explanation of variational inference starts with the same basic premise: given an observed variable $x$ , and a latent variable $z$ , $$ p(z|x)=\frac{p(x,z)}{p(x)} $$ and then proceeds to expand $p(x)$ as an expect…

It is said that a fully convolutional network can handle any image size. I don't understand. Unlike regular CNN, a fully convolutional network reinterprets the dense layer as a convolutional operation. Specifically, we t…

I want to take a picture of a bracelet, and, make, automatically (created by the AI), code, with a data structure, consisting, of, an array, of node objects whose private data element is a variable encoding the color of…

I implemented a DQN algorithm that plays OpenAIs Cartpole environment. The NN architecture consists of 3 normal linear layers that encode the state, and one noisy linear layer, that predicts the Q value based on the enco…

The Mereology Theory below contains three first-order axioms that represent a part of a mereology theory. For this posting, it is important that the set of axioms should be considered as a theory . Mereology Theory Refle…

This question asks if there is a way to measure distance between policies that are in fact probability distributions. In the case of continuous control with deterministic policies where they take a state as input and ret…

Why are policy gradient methods popular in reinforcement learning when there exists a dual LP formulation in terms of occupation measures that can be solved easily?

Are there any good ways of simultaneously incorporating object detection with speech recognition? For example, if you want to identify whether an animal is a dog or cat, we can obviously use visual features (e.g. YOLO, C…

I've been trying to understand the Distilling the Knowledge in a Neural Network paper by Hinton et al. But I cannot fully understand this: When the soft targets have high entropy, they provide much more information per t…

Sign language processing systems have traditionally operated at the sentence level, ignoring critical discourse phenomena fundamental to sign language comprehension. We introduce DiscoSign, a computational approach for d…

Evaluating video captioning remains a critical challenge for Visual Large Language Models (VLLMs). Existing metrics primarily rely on matching generated text against ground-truth references. This paradigm suffers from th…

Proteins are fundamental to biological processes, with their function determined by the complex interplay between the amino acid sequence and the three-dimensional structure. Developing generative models capable of under…

Manufacturing floors, warehouses and production lines rarely stay fixed — tasks change, layouts shift and new products arrive, and most robots can’t keep up without significant reprogramming. Skild AI’s new S1 robot foun…

César de la Fuente’s lab uses Codex and ChatGPT to search living and extinct genomes for antimicrobial candidates to fight drug-resistant infections.

OpenAI is working with an independent Advisory Group on Mathematics and Artificial Intelligence to guide the review and communication of emerging AI results.

Biomolecular structure prediction is now often run at proteome scale, where the goal is to move an entire worklist through the pipeline efficiently. NVIDIA...

Google and NASA JPL developed an AI model to map and quantify global methane emissions from space using EMIT.

Most one-base changes to the human genome do nothing, but a few are significant.

Overshadowing Cognition's $48B Series E, Mistral's $24B Series D, Meta's Muse agent, and GPT Image 2.5. The most jam packed, feel the AGI day in the history of AI.

OpenAI’s latest mathematical milestone has quickly become mired in controversy. Today, the company announced that its agents have solved one of the Millennium Prize Problems, some of the most important open problems in m…

Artificial intelligence has solved a major mathematics problem, but credit for the accomplishment is murky.

See how an MIT researcher uses GPT-5.6 Sol with Codex to autonomously run quantum computing experiments, analyze results, and calibrate qubits.

AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.

DNA is often explained as a codebook or set of instructions for producing proteins, and ultimately, life. Some stretches of DNA, called genes, code for proteins, but the vast majority of DNA is considered “noncoding.” So…

We’re sharing an AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a writeup and a formal proof in Lean.

Plus, a machine hermeneutics story

We’re partnering with Cathay Pacific to test our AI-powered contrail mitigation technology on ultra-long-haul flights.

Inside OpenAI, coding agents are reshaping AI research. Explore early data on agent usage, experiment velocity, task complexity, and research acceleration.