call_end

    • Ar chevron_right

      DeepSeek tests “sparse attention” to slash AI processing costs

      news.movim.eu / ArsTechnica • 30 September • 1 minute

    Ever wonder why ChatGPT slows down during long conversations? The culprit is a fundamental mathematical challenge: processing long sequences of text requires massive computational resources, even with the efficiency tricks that companies have already deployed. While US tech giants can afford to throw more hardware at the problem, Chinese AI company DeepSeek , which is cut off from a steady supply of some advanced AI chips by export restrictions , has extra motivation to squeeze more performance from less silicon.

    On Monday, DeepSeek released an experimental version of its latest simulated reasoning language model, DeepSeek-V3.2-Exp, which introduces what it calls "DeepSeek Sparse Attention" (DSA). It's the company's implementation of a computational technique likely already used in some of the world's most prominent AI models. OpenAI pioneered sparse transformers in 2019 and used the technique to build GPT-3, while Google Research published work on " Reformer " models using similar concepts in 2020. (The full extent to which Western AI companies currently use sparse attention in their latest models remains undisclosed.)

    Despite sparse attention being a known approach for years, DeepSeek claims its version achieves "fine-grained sparse attention for the first time" and has cut API prices by 50 percent to demonstrate the efficiency gains. But to understand more about what makes DeepSeek v3.2 notable, it's useful to refresh yourself on a little AI history.

    Read full article

    Comments

    • Ar chevron_right

      After threatening ABC over Kimmel, FCC chair may eliminate TV ownership caps

      news.movim.eu / ArsTechnica • 30 September

    Fresh off his crusade against Jimmy Kimmel and ABC, the chair of the Federal Communications Commission may eliminate TV station ownership limits in a potential gift for station owners like Sinclair and Nexstar.

    When FCC Chairman Brendan Carr threatened ABC affiliates with license revocations for carrying Jimmy Kimmel's show, he said that national networks exert too much control over local TV stations and that he's trying "to empower local TV stations to serve the needs of the local communities." Taking a cue from Carr, Sinclair and Nexstar continued blocking Jimmy Kimmel Live! on their ABC affiliates even after ABC and its owner Disney ended Kimmel's suspension.

    Within days, Sinclair and Nexstar decided to put Kimmel back on the air . Pressure from viewers and advertisers likely played a major role in the reversal. But for Carr, the episode might reinforce his belief that station groups should have more influence over national programming.

    Read full article

    Comments

    • Ar chevron_right

      With new agent mode for Excel and Word, Microsoft touts “vibe working”

      news.movim.eu / ArsTechnica • 30 September

    With a new set of Microsoft 365 features , knowledge workers will be able to generate complex Word documents or Excel spreadsheets using only text prompts to Microsoft's chat bot. Two distinct products were announced, each using different models and accessed from within different tools—though the similar names Microsoft chose make it confusing to parse what's what.

    Driven by OpenAI's GPT-5 large language model, Agent Mode is built into Word and Excel, and it allows the creation of complex documents and spreadsheets from user prompts. It's called "agent" mode because it doesn't just work from the prompt in a single step; rather, it plans multi-step work and runs a validation loop in the hopes of ensuring quality.

    It's only available in the web versions of Word and Excel at present, but the plan is to bring it to native desktop applications later.

    Read full article

    Comments

    • Ar chevron_right

      YouTuber unboxes what seems to be a pre-release version of an M5 iPad Pro

      news.movim.eu / ArsTechnica • 30 September • 1 minute

    Apple's biggest product event of the year happens in September, when the company puts out a new batch of iPhones and Apple Watches and other odds and ends. But in most years, Apple either has another smaller event or just a handful of additional product announcements later in the fall in October or November—usually the focus is on the Mac, the iPad, or both.

    It seems like a new iPad Pro could be one of the announcements on tap. Russian YouTube channel Wylsacom has posted an unboxing video and early tour of what appears to be a retail boxed version of a new 256GB 13-inch iPad Pro, as well as an M5 processor that we haven't seen in any other Apple product yet. This would be the first new iPad Pro since May of 2024, when Apple introduced the current M4 version .

    The same channel also got ahold of the M4 MacBook Pro early , so it seems likely that this is genuine. And while the video is mostly dedicated to complaining about packaging, the wattage of the included power adapter, and how boring it is that Apple doesn't introduce dramatic design changes every generation, it does also give us some early performance numbers for the new M5.

    Read full article

    Comments

    • wifi_tethering open_in_new

      This post is public

      arstechnica.com /gadgets/2025/09/youtuber-unboxes-what-seems-to-be-a-pre-release-version-of-an-m5-ipad-pro/

    • Pictures 1 image

    • visibility
    • Ar chevron_right

      SpaceX has a few tricks up its sleeve for the last Starship flight of the year

      news.movim.eu / ArsTechnica • 30 September

    On its surface, the flight plan for SpaceX's next Starship flight looks a lot like the last one.

    The rocket's Super Heavy booster will again splash down in the Gulf of Mexico just offshore from SpaceX's launch site in South Texas. And Starship, the rocket's upper stage, will fly on a suborbital arc before reentering the atmosphere over the Indian Ocean for a water landing northwest of Australia.

    SpaceX will again test the rocket's satellite deployer and reignite one of the ship's Raptor engines in space to adjust the vehicle's path for reentry. These demonstrations will pave the way for future Starship flights into low-Earth orbit. All of the rocket's ascents to date have, by design, ended before reaching orbital velocity.

    Read full article

    Comments

    • Ar chevron_right

      iOS 26.0.1, macOS 26.0.1 updates fix install bugs, new phone problems, and more

      news.movim.eu / ArsTechnica • 30 September

    Now that iOS 26 , macOS 26 Tahoe , and Apple's other big software updates for the year are out in public, Apple's efforts for the next few months will shift to fixing bugs and adding individual new features. The first of those bug fix updates has arrived this week in the form of iOS 26.0.1, macOS 26.0.1, iPadOS 26.0.1, and equivalent updates for most of the devices across Apple's ecosystem.

    The release notes for most of the updates focus on device- and platform-specific early adopter problems, particularly for buyers of the new iPhone 17, iPhone 17 Pro, and iPhone Air.

    The iOS 26.0.1 update fixes a bug that could prevent phones from connecting to cellular networks, a bug that could cause app icons to appear blank, and the VoiceOver feature becoming disabled on devices that have it on. Camera, Wi-Fi, and Bluetooth bugs with the new iPhones have also been patched. The iPadOS update also fixes a bug that was causing the floating software keyboard to move around.

    Read full article

    Comments

    • Ar chevron_right

      California’s newly signed AI law just gave Big Tech exactly what it wanted

      news.movim.eu / ArsTechnica • 30 September

    On Monday, California Governor Gavin Newsom signed the Transparency in Frontier Artificial Intelligence Act into law, requiring AI companies to disclose their safety practices while stopping short of mandating actual safety testing. The law requires companies with annual revenues of at least $500 million to publish safety protocols on their websites and report incidents to state authorities, but it lacks the stronger enforcement teeth of the bill Newsom vetoed last year after tech companies lobbied heavily against it.

    The legislation, S.B. 53 , replaces Senator Scott Wiener's previous attempt at AI regulation, known as S.B. 1047 , that would have required safety testing and "kill switches" for AI systems. Instead, the new law asks companies to describe how they incorporate "national standards, international standards, and industry-consensus best practices" into their AI development, without specifying what those standards are or requiring independent verification.

    "California has proven that we can establish regulations to protect our communities while also ensuring that the growing AI industry continues to thrive," Newsom said in a statement, though the law's actual protective measures remain largely voluntary beyond basic reporting requirements.

    Read full article

    Comments

    • Ar chevron_right

      Behind the scenes with the most beautiful car in racing: The Ferrari 499P

      news.movim.eu / ArsTechnica • 30 September • 1 minute

    Ferrari provided flights from Washington, DC, to Austin, Texas, and accommodation so Ars could attend the Lone Star Le Mans. Ars does not accept paid editorial content.

    By the time you read this, the World Endurance Championship will be 100 races old. Once centered on the mega-expensive, mega-fast LMP1 hybrids , it's all about the Hypercars now. These are purpose-built, closed-top race cars, with some of the most complex hybrid systems you'll find outside of Formula 1, clad in bodywork that could give the Batmobile a run for its money. These Hypercars are designed to last the distance. During the 24 hours of Le Mans this year, the winning car covered 3,276 miles (5,273 km); for context, an F1 race is usually 190 miles (305 km), and at Monza this year , the race lasted little more than an hour. Here's another difference with F1: When it comes to endurance racing, Ferrari has been winning a lot.

    In fact, it's taken victory at Le Mans for three years in a row, scoring a hat trick after 50 years away from this corner of the sport. This year has been even better: Ferrari leads the manufacturer's championship and the driver's championship with the #51 factory car. Its closest rival for the driver's title is another 83 Ferrari, this one entered as a privateer car by the AF Corsa team. When the invite arrived to join the team for its race at the Circuit of the Americas in Texas, it seemed like a perfect opportunity to watch it in action and find out the key to that success.

    "It has been an amazing challenge for us because after 50 years, it was not simple to restart in the pinnacle of motorsport," said Antonello Coletta, head of Ferrari's endurance racing program. Eighteen months after the car was greenlit, it was racing at Sebring in early 2023, a notoriously bumpy WWII bomber base in Florida that is as hard a test on a race car as any. Later that year, the 499P won Le Mans on its first try .

    Read full article

    Comments

    • Ar chevron_right

      Is the “million-year-old” skull from China a Denisovan or something else?

      news.movim.eu / ArsTechnica • 30 September

    A fossil skull from China that made headlines last week may or may not be a million years old, but it's probably closely related to Denisovans.

    The fossil skull, dubbed Yunxian 2, is one of three unearthed from a terrace alongside the Han River, in central China, in a layer of river sediment somewhere between 600,000 and 1 million years old. Archaeologists originally identified them as Homo erectus , but Hanjiang Normal University paleoanthropologist Xiaobo Feng and his colleagues’ recent digital reconstruction of Yunxian 2 suggests the skulls may actually have belonged to someone a lot more similar to us: a hominin group defined as a species called Homo longi or a Denisovan, depending on who’s doing the naming.

    The recent paper adds fuel—and a new twist—to that debate. And the whole thing may hinge on a third skull from the same site, still waiting to be published.

    Read full article

    Comments