Elon Musk's artificial intelligence company is stirring controversy in Memphis, Tennessee. That's where he's building a massive supercomputer to power his company xAI. Community residents and environmental activists say that since the supercomputer was fired up last summer it has become one of the biggest air polluters in the county. But some local officials have championed the billionaire, saying he's investing in Memphis. The first public hearing with the health department is scheduled for Friday, where county officials will hear from all sides of the debate.

AI for education is a new but rapidly expanding field. Can it support student outcomes and help teachers avoid burnout? On Wednesday, AI education company Kira launched a "fully AI-native learning platform" for K-12 education, complete with agents to assist teachers with repetitive tasks. The platform hosts assignments, analyzes progress data, offers administrative assistance, helps build lesson plans and quizzes, and more. Also: Google's One AI Premium plan with Gemini Advanced is now free for students - for an entire year "Unlike traditional tools that merely layer AI onto existing platforms, Kira integrates artificial intelligence directly into every educational workflow -- from lesson planning and instruction to grading, intervention, and reporting," the release explains.

Elon Musk's xAI artificial intelligence firm has acquired Musk's X – the social media platform formerly known as Twitter – for 33bn, marking the latest twist in the billionaire's rapid consolidation of power. The all-stock deal announced on Friday combines two of Musk's multiple portfolio companies, which also include automaker Tesla and SpaceX, and potentially eases Musk's ability to train his AI model known as Grok. Musk announced the transaction in a post on X, saying: "The combination values xAI at 80bn and X at 33bn ( 45B less 12B debt)." "xAI and X's futures are intertwined," he wrote. "Today, we officially take the step to combine the data, models, compute, distribution and talent."

Elon Musk's artificial intelligence firm xAI has acquired his social media platform X in an all-stock transaction that values the company at 33 billion, including 12 billion worth of debt, the centibillionaire announced Friday. The sale comes just weeks after Musk reportedly raised an additional roughly 1 billion in debt financing for X that valued the company at 44 billion--the same price Musk paid for it three years ago. "xAI and X's futures are intertwined," Musk wrote in an X post. "Today, we officially take the step to combine the data, models, compute, distribution and talent. This combination will unlock immense potential by blending xAI's advanced AI capability and expertise with X's massive reach."

When trying to fit a deep neural network (DNN) to a G-invariant target function with G a group, it only makes sense to constrain the DNN to be G-invariant as well. However, there can be many different ways to do this, thus raising the problem of "G-invariant neural architecture design": What is the optimal G-invariant architecture for a given problem? Before we can consider the optimization problem itself, we must understand the search space, the architectures in it, and how they relate to one another. In this paper, we take a first step towards this goal; we prove a theorem that gives a classification of all G-invariant single-hidden-layer or "shallow" neural network (G-SNN) architectures with ReLU activation for any finite orthogonal group G, and we prove a second theorem that characterizes the inclusion maps or "network morphisms" between the architectures that can be leveraged during neural architecture search (NAS). The proof is based on a correspondence of every G-SNN to a signed permutation representation of G acting on the hidden neurons; the classification is equivalently given in terms of the first cohomology classes of G, thus admitting a topological interpretation. The G-SNN architectures corresponding to nontrivial cohomology classes have, to our knowledge, never been explicitly identified in the literature previously. Using a code implementation, we enumerate the G-SNN architectures for some example groups G and visualize their structure. Finally, we prove that architectures corresponding to inequivalent cohomology classes coincide in function space only when their weight matrices are zero, and we discuss the implications of this for NAS.

Reward models trained on human preference data have been proven to effectively align Large Language Models (LLMs) with human intent within the framework of reinforcement learning from human feedback (RLHF). However, current reward models have limited generalization capabilities to unseen prompts and responses, which can lead to an unexpected phenomenon known as reward over-optimization, resulting in a decline in actual performance due to excessive optimization of rewards. While previous research has advocated for constraining policy optimization, our study introduces a novel approach to enhance the reward model's generalization ability against distribution shifts by regularizing the hidden states. Specifically, we retain the base model's language model head and incorporate a suite of text-generation losses to preserve the hidden states' text-generation capabilities, while concurrently learning a reward head behind the same hidden states.

Computational intensiveness of deep learning has motivated low-precision arithmetic designs. However, the current quantized/binarized training approaches are limited by: (1) significant performance loss due to arbitrary approximations of the latent weight gradient through its discretization/binarization function, and (2) training computational intensiveness due to the reliance on full-precision latent weights. This paper proposes a novel mathematical principle by introducing the notion of Boolean variation such that neurons made of Boolean weights and/or activations can be trained --for the first time-- natively in Boolean domain instead of latent-weight gradient descent and real arithmetic. We explore its convergence, conduct extensively experimental benchmarking, and provide consistent complexity evaluation by considering chip architecture, memory hierarchy, dataflow, and arithmetic precision. Our approach achieves baseline full-precision accuracy in ImageNet classification and surpasses state-of-the-art results in semantic segmentation, with notable performance in image super-resolution, and natural language understanding with transformer-based models. Moreover, it significantly reduces energy consumption during both training and inference.

While visiting HP's Omen gaming exhibit at the company's Amplify Conference in Nashville Tennessee this week, I realized something: I've been optimizing my PC's performance for Counter-Strike 2 all wrong! What I'm doing is painstakingly combing through my hardware settings, OS settings, and game settings in a confusing and sometimes panic-ridden mind muddle in the hope I'll achieve an actual uplift in FPS. That's where HP's Omen AI comes in. Originally unveiled at CES 2025 Las Vegas last January, Omen AI made another appearance at the conference in Nashville, this time showing off some impressive FPS gains on both laptops and desktop PCs. In one of the promotional videos, Omen AI boosted a laptop's performance from 82 FPS to 111 FPS.

Two high school graduates who say they can't read or write are suing their respective public school systems, arguing they were not given the free public education to which they are entitled. Cornell Law School Professor William A. Jacobson, director of the Securities Law Clinic, told Fox News Digital the lawsuits signify a "much deeper problem" with the American public school system. "I think these cases reflect a deeper problem in education. For each of these cases, there are probably tens of thousands of students who never got a proper education -- they get pushed along the system," Jacobson said. "Unfortunately … we've created incentives, particularly for public school systems, to just push students along and not to hold them accountable."

Unexpected stimuli induce "error" or "surprise" signals in the brain. The theory of predictive coding promises to explain these observations in terms of Bayesian inference by suggesting that the cortex implements variational inference in a probabilistic graphical model. However, when applied to machine learning tasks, this family of algorithms has yet to perform on par with other variational approaches in high-dimensional, structured inference problems. To address this, we introduce a novel predictive coding algorithm for structured generative models, that we call divide-and-conquer predictive coding (DCPC); it differs from other formulations of predictive coding, as it respects the correlation structure of the generative model and provably performs maximum-likelihood updates of model parameters, all without sacrificing biological plausibility. Empirically, DCPC achieves better numerical performance than competing algorithms and provides accurate inference in a number of problems not previously addressed with predictive coding. We provide an open implementation of DCPC in Pyro on Github.