• chevron_right

      Can a technology called RAG keep AI models from making stuff up?

      news.movim.eu / ArsTechnica · Thursday, 6 June - 11:00 · 1 minute

    Can a technology called RAG keep AI models from making stuff up?

    Enlarge (credit: Aurich Lawson | Getty Images)

    We’ve been living through the generative AI boom for nearly a year and a half now, following the late 2022 release of OpenAI’s ChatGPT. But despite transformative effects on companies’ share prices, generative AI tools powered by large language models (LLMs) still have major drawbacks that have kept them from being as useful as many would like them to be. Retrieval augmented generation, or RAG, aims to fix some of those drawbacks.

    Perhaps the most prominent drawback of LLMs is their tendency toward confabulation (also called “hallucination”), which is a creative gap-filling technique AI language models use when they encounter holes in their knowledge that weren’t present in their training data. They generate plausible-sounding text that can veer toward accuracy when the training data is solid but otherwise may just be completely made up.

    Relying on confabulating AI models gets people and companies in trouble, as we’ve covered in the past. In 2023, we saw two instances of lawyers citing legal cases, confabulated by AI, that didn’t exist. We’ve covered claims against OpenAI in which ChatGPT confabulated and accused innocent people of doing terrible things. In February, we wrote about Air Canada’s customer service chatbot inventing a refund policy , and in March, a New York City chatbot was caught confabulating city regulations.

    Read 30 remaining paragraphs | Comments

    • chevron_right

      Windows Recall demands an extraordinary level of trust that Microsoft hasn’t earned

      news.movim.eu / ArsTechnica · Tuesday, 4 June - 17:15 · 1 minute

    The Recall feature as it currently exists in Windows 11 24H2 preview builds.

    Enlarge / The Recall feature as it currently exists in Windows 11 24H2 preview builds. (credit: Andrew Cunningham)

    Microsoft’s Windows 11 Copilot+ PCs come with quite a few new AI and machine learning-driven features, but the tentpole is Recall . Described by Microsoft as a comprehensive record of everything you do on your PC, the feature is pitched as a way to help users remember where they’ve been and to provide Windows extra contextual information that can help it better understand requests from and meet the needs of individual users.

    This, as many users in infosec communities on social media immediately pointed out , sounds like a potential security nightmare. That’s doubly true because Microsoft says that by default, Recall’s screenshots take no pains to redact sensitive information, from usernames and passwords to health care information to NSFW site visits. By default, on a PC with 256GB of storage, Recall can store a couple dozen gigabytes of data across three months of PC usage, a huge amount of personal data.

    The line between “potential security nightmare” and “actual security nightmare” is at least partly about the implementation, and Microsoft has been saying things that are at least superficially reassuring . Copilot+ PCs are required to have a fast neural processing unit (NPU) so that processing can be performed locally rather than sending data to the cloud; local snapshots are protected at rest by Windows’ disk encryption technologies, which are generally on by default if you’ve signed into a Microsoft account; neither Microsoft nor other users on the PC are supposed to be able to access any particular user’s Recall snapshots; and users can choose to exclude apps or (in most browsers) individual websites to exclude from Recall’s snapshots.

    Read 18 remaining paragraphs | Comments

    • chevron_right

      No physics? No problem. AI weather forecasting is already making huge strides.

      news.movim.eu / ArsTechnica · Monday, 3 June - 11:00

    AI weather models are arriving just in time for the 2024 Atlantic hurricane season.

    Enlarge / AI weather models are arriving just in time for the 2024 Atlantic hurricane season. (credit: Aurich Lawson | Getty Images)

    Much like the invigorating passage of a strong cold front, major changes are afoot in the weather forecasting community. And the end game is nothing short of revolutionary: an entirely new way to forecast weather based on artificial intelligence that can run on a desktop computer.

    Today's artificial intelligence systems require one resource more than any other to operate—data. For example, large language models such as ChatGPT voraciously consume data to improve answers to queries. The more and higher quality data, the better their training, and the sharper the results.

    However, there is a finite limit to quality data, even on the Internet. These large language models have hoovered up so much data that they're being sued widely for copyright infringement . And as they're running out of data, the operators of these AI models are turning to ideas such as synthetic data to keep feeding the beast and produce ever more capable results for users.

    Read 41 remaining paragraphs | Comments

    • chevron_right

      Driverless racing is real, terrible, and strangely exciting

      news.movim.eu / ArsTechnica · Friday, 31 May - 11:00

    Several brightly colored race cars are parked at a race course

    Enlarge / No one's entirely sure if driverless racing will be any good to watch, but before we find that out, people have to actually develop driverless race cars. A2RL in Abu Dhabi is the latest step down that path. (credit: A2RL)

    A2RL provided flights from London to Abu Dhabi and accommodation so Ars could attend the autonomous race event. Ars does not accept paid editorial content.

    ABU DHABI—We live in a weird time for autonomous vehicles. Ambitions come and go, but genuinely autonomous cars are further off than solid-state vehicle batteries. Part of the problem with developing autonomous cars is that teaching road cars to take risks is unacceptable.

    A race track, though, is a decent place to potentially crash a car. You can take risks there, with every brutal crunch becoming a learning exercise. (You’d be hard-pressed to find a top racing driver without a few wrecks smoldering in their junior career records.)

    That's why 10,000 people descended on the Yas Marina race track in Abu Dhabi to watch the first four-car driverless race.

    Read 49 remaining paragraphs | Comments

    • chevron_right

      Small, cheap, and weird: A history of the microcar

      news.movim.eu / ArsTechnica · Monday, 27 May - 11:00

    Small, cheap, and weird: A history of the microcar

    Enlarge (credit: Aurich Lawson)

    European car manufacturers are currently tripping over themselves to figure out how personal transport and "last mile" solutions will look in the years to come. The solutions are always electric, and they're also tiny. What most companies (bar Citroen, Renault, and Fiat) seem to have forgotten is that we've had an answer to this problem all along: the microcar.

    The microcar is a singular little thing —its job is to frugally take one person (or maybe two people) where they need to go while taking up as little space as possible. A few have broken their way into the public consciousness— Top Gear made a global megastar of Peel's cars, BMW's Isetta remains a design icon, and the Messerschmitt KR200 is just plain cool—but where did these tiny wonders come from? And do they have a future?

    Well, without the microcar's predecessors, we may not have the modern motorcar as we know it. Sort of.

    Read 29 remaining paragraphs | Comments

    • chevron_right

      Google’s “AI Overview” can give false, misleading, and dangerous answers

      news.movim.eu / ArsTechnica · Friday, 24 May - 11:00

    This is fine.

    Enlarge / This is fine. (credit: Getty Images)

    If you use Google regularly, you may have noticed the company's new AI Overviews providing summarized answers to some of your questions in recent days. If you use social media regularly, you may have come across many examples of those AI Overviews being hilariously or even dangerously wrong .

    Factual errors can pop up in existing LLM chatbots as well, of course. But the potential damage that can be caused by AI inaccuracy gets multiplied when those errors appear atop the ultra-valuable web real estate of the Google search results page.

    "The examples we've seen are generally very uncommon queries and aren’t representative of most people’s experiences," a Google spokesperson told Ars. "The vast majority of AI Overviews provide high quality information, with links to dig deeper on the web."

    Read 18 remaining paragraphs | Comments

    • chevron_right

      On self-driving, Waymo is playing chess while Tesla plays checkers

      news.movim.eu / ArsTechnica · Wednesday, 22 May - 11:00

    A Waymo autonomous taxi in San Francisco.

    Enlarge / A Waymo autonomous taxi in San Francisco. (credit: David Paul Morris/Bloomberg via Getty Images)

    Tesla fans—and CEO Elon Musk himself—are excited about the prospects for Tesla’s Full Self Driving (FSD) software. Tesla released a major upgrade—version 12.3—of the software in March. Then, last month, Musk announced that Tesla would unveil a purpose-built robotaxi on August 8. Last week, Musk announced that a new version of FSD—12.4—will come out in the coming days and will have a “5X to 10X improvement in miles per intervention.”

    But I think fans expecting Tesla to launch a driverless taxi service in the near future will be disappointed.

    During a late March trip to San Francisco, I had a chance to try the latest self-driving technology from both Tesla and Google’s Waymo.

    Read 74 remaining paragraphs | Comments

    • chevron_right

      Municipal broadband advocates fight off attacks from “dark money” groups

      news.movim.eu / ArsTechnica · Tuesday, 21 May - 13:37

    Illustration of shadowy figures and a light bulb over a map of the United States with lines depicting broadband networks.

    Enlarge (credit: Aurich Lawson | Getty Images)

    Cities and towns that build their own broadband networks often say they only considered the do-it-yourself option because private Internet service providers didn't meet their communities' needs. When a cable or phone company's Internet service is too slow, too expensive, not deployed widely enough, or all of the above, local government officials sometimes decide to take matters into their own hands.

    Hundreds of municipal broadband networks have been built around the US as a result, including dozens that have started operating since 2021. The rise of public broadband hasn't happened without a fight, though. Private ISPs that would rather face no government-funded competition have tried to convince voters that public networks are doomed to become boondoggles.

    Opponents of public broadband don't always attach their names to these campaigns, but it often seems likely that private ISPs are behind the anti-municipal broadband lobbying. Public broadband advocates say that over the past few years, they've seen a noticeable increase in "dark money" groups attacking public network projects.

    Read 85 remaining paragraphs | Comments

    • chevron_right

      We take a stab at decoding SpaceX’s ever-changing plans for Starship in Florida

      news.movim.eu / ArsTechnica · Monday, 20 May - 12:04

    SpaceX's Starship tower (left) at Launch Complex 39A dwarfs the launch pad for the Falcon 9 rocket (right).

    Enlarge / SpaceX's Starship tower (left) at Launch Complex 39A dwarfs the launch pad for the Falcon 9 rocket (right). (credit: NASA/Aubrey Gemignani )

    There are a couple of ways to read the announcement from the Federal Aviation Administration that it's kicking off a new environmental review of SpaceX's plan to launch the most powerful rocket in the world from Florida.

    The FAA said on May 10 that it plans to develop an Environmental Impact Statement (EIS) for SpaceX's proposal to launch Starships from NASA's Kennedy Space Center in Florida. The FAA ordered this review after SpaceX updated the regulatory agency on the projected Starship launch rate and the design of the ground infrastructure needed at Launch Complex 39A (LC-39A), the historic launch pad once used for Apollo and Space Shuttle missions.

    Dual environmental reviews

    At the same time, the US Space Force is overseeing a similar EIS for SpaceX's proposal to take over a launch pad at Cape Canaveral Space Force Station, a few miles south of LC-39A. This launch pad, designated Space Launch Complex 37 (SLC-37 ), is available for use after United Launch Alliance's last Delta rocket lifted off there in April.

    Read 56 remaining paragraphs | Comments