Transformer-based large language models (LLMs) use the key-value (KV) cache to significantly accelerate inference by storing the key and value embeddings of past tokens. However, this cache consumes significant GPU memory. In this work, we introduce HashEvict, an algorithm that uses locality-sensitive hashing (LSH) to compress the KV cache. HashEvict quickly locates tokens in the cache that are cosine dissimilar to the current query token. This is achieved by computing the Hamming distance between binarized Gaussian projections of the current token query and cached token keys, with a projection length much smaller than the embedding dimension. We maintain a lightweight binary structure in GPU memory to facilitate these calculations. Unlike existing compression strategies that compute attention to determine token retention, HashEvict makes these decisions pre-attention, thereby reducing computational costs. Additionally, HashEvict is dynamic - at every decoding step, the key and value of the current token replace the embeddings of a token expected to produce the lowest attention score. We demonstrate that HashEvict can compress the KV cache by 30%-70% while maintaining high performance across reasoning, multiple-choice, long-context retrieval and summarization tasks.

Detecting and analyzing complex patterns in multivariate time-series data is crucial for decision-making in urban and environmental system operations. However, challenges arise from the high dimensionality, intricate complexity, and interconnected nature of complex patterns, which hinder the understanding of their underlying physical processes. Existing AI methods often face limitations in interpretability, computational efficiency, and scalability, reducing their applicability in real-world scenarios. This paper proposes a novel visual analytics framework that integrates two generative AI models, Temporal Fusion Transformer (TFT) and Variational Autoencoders (VAEs), to reduce complex patterns into lower-dimensional latent spaces and visualize them in 2D using dimensionality reduction techniques such as PCA, t-SNE, and UMAP with DBSCAN. These visualizations, presented through coordinated and interactive views and tailored glyphs, enable intuitive exploration of complex multivariate temporal patterns, identifying patterns' similarities and uncover their potential correlations for a better interpretability of the AI outputs. The framework is demonstrated through a case study on power grid signal data, where it identifies multi-label grid event signatures, including faults and anomalies with diverse root causes. Additionally, novel metrics and visualizations are introduced to validate the models and evaluate the performance, efficiency, and consistency of latent maps generated by TFT and VAE under different configurations. These analyses provide actionable insights for model parameter tuning and reliability improvements. Comparative results highlight that TFT achieves shorter run times and superior scalability to diverse time-series data shapes compared to VAE. This work advances fault diagnosis in multivariate time series, fostering explainable AI to support critical system operations.

Social media firms could be ordered to use facial recognition technology to check children's ages. Millions of children could have their online profiles banned by the tech giants under plans to be set out by online regulator Ofcom next spring. Social media executives have been warned they could face huge fines and even prison sentences if they fail to follow guidance designed to ensure their users are not underage. John Higham, Ofcom's head of online safety policy, said platforms would be expected to remove children's accounts from their sites by using'highly accurate and effective' AI age checks. The regulator estimates that up to 60 per cent of eight to 11-year-olds have social media profiles, despite sites such as Facebook, Instagram, TikTok and Snapchat having minimum age limits of 13.

Video shows the moment drones started falling from the sky during a drone show at Eola Lake in Orlando, Florida on Dec. 21, 2024. A 7-year-old Florida boy who was injured when drones collided and fell into a crowd at a holiday airshow over the weekend underwent open heart surgery, his parents said. Adriana Edgerton and Jessica Lumsden, parents of Alexander, said one of the red and green-lit drones struck him and knocked him out upon impact, causing a chest injury, Fox Orlando reported. Hundreds of drones being used as part of a Saturday night aerial light show in Lake Eola Park in downtown Orlando appeared to be flying into position before several started falling from the sky before slamming to the ground, according to videos posted online. Alexander, a 7-year-old boy, has undergone heart surgery after he was struck by a falling drone during a holiday airshow in Orlando, his parents said.

The headlines seem to write themselves (if that cliché is allowed anymore in the age of ChatGPT and generative AI). But that is a metaphor, right? When I first saw Michael Saylor's Twitter account, I wasn't sure. Saylor is an entrepreneur, tech executive, and former billionaire. Once reportedly the richest man in the Washington, DC, area, he lost most of his 7 billion net worth in 2000 when, in his mid-30s, he reached a settlement with the US Securities and Exchange Commission after it brought charges against him and two of his colleagues at a company called MicroStrategy for inaccurate reporting of their financial results.

AI tools rapidly infiltrated peoples' lives in 2024, but AI lawmaking in the U.S. moved much more slowly. While dozens of AI-related bills were introduced this Congress--either to fund its research or mitigate its harms--most got stuck in partisan gridlock or buried under other priorities. In California, a bill aiming to hold AI companies liable for harms easily passed the state legislature, but was vetoed by Governor Gavin Newsom. This inaction has some AI skeptics increasingly worried. "We're seeing a replication of what we've seen in privacy and social media: of not setting up guardrails from the start to protect folks and drive real innovation," Ben Winters, the director of AI and data privacy at the Consumer Federation of America, tells TIME.

Reveal has been a weekly investigative podcast for nearly 10 years now, so we've produced hundreds of hours of investigative journalism over the years designed to inspire, inform, or infuriate you (and occasionally, all three at the same time). We've curated some of our favorite Reveal series and serials to take you through your holiday travel time--episodes that will resonate today and into 2025. You can find the link to each episode on your preferred podcast platform below. Mississippi Goddam (seven-part series): Billey Joe Johnson Jr. dreamed of graduating high school, going to college, and one day playing pro football. On a cold December morning in 2008, that future was shattered.

As investor enthusiasm for artificial intelligence, and lately for a Trump Presidency, has been driving the stock market to record highs this year, Jeremy Grantham has been having flashbacks. At the end of the nineteen-nineties, the veteran value investor--one that looks for undervalued stocks--shied away from soaring Internet and technology stocks, believing that their prices had departed from financial reality, and that the market was heading for a crash. Far from thanking him for sounding the alarm, many clients of G.M.O., a Boston-based investment-management firm that Grantham had co-founded, held it responsible for making them miss out on a vertiginous rise in the Nasdaq, which went up by about a hundred and sixty per cent between 1998 and 1999. Some withdrew their money from the company. "We started off in a good position, and in two years we lost almost half of our business," Grantham recalled.

The year 2024 saw the global economy stabilise following the fallout of the COVID-19 pandemic, even as growth in many countries lagged pre-2020 levels. Amid a patchy recovery, more than 2 billion people were eligible to vote this year, and economic issues, particularly rising living costs, were a top concern for voters around the world. Meanwhile, governments grappled with how to regulate potentially transformational technology such as artificial intelligence, and Donald Trump's victory in the United States' presidential election heralded a sharp turn towards protectionism. Trump has indicated that he will pursue an even more aggressive version of the "America First" protectionism that fuelled his rise to power during his second stint in the White House. On the campaign trail, Trump pledged to impose tariffs of 60 percent or higher on Chinese goods and a blanket 20 percent tariff on all other imports.

Video shows the moment drones started falling from the sky during a drone show at Eola Lake in Orlando, Florida on Dec. 21, 2024. A child was hospitalized on Saturday after being hit by a drone that was part of an Orlando, Florida holiday drone show. According to the Orlando Fire Department, a 7-year-old boy was transported to the hospital because of injuries sustained from the falling drones, FOX 35 in Orlando reported. In a video posted online by X user MosquitoCoFl, hundreds of drones being used as part of an aerial light show appeared to be flying into position before several started falling from the sky before slamming to the ground. A man could be heard saying to children nearby, "Oh no! I don't believe they're supposed to be falling."