An anonymous reader quotes a report from Ars Technica: Today Google search works by building a huge index of the web, and when you search for something, those index entries gets scanned and ranked and categorized, with the most relevant entries showing up in your search results. Google's results page actually tells you how long all of this takes when you search for something, and it's usually less than a second. A ChatGPT-style search engine would involve firing up a huge neural network modeled on the human brain every time you run a search, generating a bunch of text and probably also querying that big search index for factual information. The back-and-forth nature of ChatGPT also means you'll probably be interacting with it for a lot longer than a fraction of a second.

All that extra processing is going to cost a lot more money. After speaking to Alphabet Chairman John Hennessy (Alphabet is Google's parent company) and several analysts, Reuters writes that "an exchange with AI known as a large language model likely costs 10 times more than a standard keyword search" and that it could represent "several billion dollars of extra costs."

Exactly how many billions of Google's $60 billion in yearly net income will be sucked up by a chatbot is up for debate. One estimate in the Reuters report is from Morgan Stanley, which tacks on a $6 billion yearly cost increase for Google if a "ChatGPT-like AI were to handle half the queries it receives with 50-word answers." Another estimate from consulting firm SemiAnalysis claims it would cost $3 billion. [...] Alphabet's Hennessy told Reuters that Google is looking into driving down costs, calling it "a couple year problem at worst."

AI voice clones are already being deployed in podcasts and video games, but how long until they can be harnessed directly by the general public? From a report: Probably sooner than you think, with Samsung today announcing a feature for its Bixby mobile assistant that lets users clone their voice to answer phone calls. The idea is that if someone calls you but you can't answer aloud you can type out a response and it'll be read in a simulacrum of your voice. Some caveats here: this feature is only currently available in Korean as the Bixby Custom Voice Creator app for a small number of Samsung handsets (the new Galaxy S23, S23+ and S23 Ultra), which means we've been unable to test it ourselves. The voice quality might be abysmal and response time too slow to be useful. But cloning voices to answer calls is well within the scope of current technology, with AI tools able to create realistic copies of voices from just a few minutes of audio.

The steep decline in Amazon's stock over the past year is roiling the technology company's stock-heavy compensation plan, resulting in employee pay coming in significantly lower than target compensation, WSJ reported, citing people familiar with the matter. From the report: Amazon pays its corporate employees a large chunk of their annual salaries in restricted stock units, and a prolonged slump in the company's shares is causing pay for 2023 to be between 15% and 50% lower than the projected targets Amazon gave to employees, some of the people said. Amazon has historically given less base-pay compensation to employees than its big-tech peers but made up the difference with stock awards that vest over several years. Employees say the longer an Amazon employee stays with the company, the more their compensation can depend on stock awards, with stocks making up 50% or more of total income for some.

Regulators have told major Chinese tech companies not to offer ChatGPT services to the public amid growing alarm in Beijing over the AI-powered chatbot's uncensored replies to user queries. From a report: Tencent Holdings and Ant Group, the fintech affiliate of Alibaba Group Holding, have been instructed not to offer access to ChatGPT services on their platforms, either directly or via third parties, people with direct knowledge of the matter told Nikkei Asia. Tech companies will also need to report to regulators before they launch their own ChatGPT-like services, the sources added.

ChatGPT, developed by Microsoft-backed startup OpenAI, is not officially available in China but some internet users have been able to access it using a virtual private network (VPN). There have also been dozens of "mini programs" released by third-party developers on Tencent's WeChat social media app that claim to offer services from ChatGPT. Under regulatory pressure, Tencent has suspended several such third-party services regardless of whether they were connected to ChatGPT or were in fact copycats, people familiar with the matter told Nikkei. This is not the first time that China has blocked foreign websites or applications. Beijing has banned dozens of prominent U.S. websites and apps. Between 2009 and 2010, it moved to block Google, Facebook, YouTube, and Twitter. Between 2018 and 2019, it instituted bans on Reddit and Wikipedia.

Tech privacy advocates frustrated by failures on Capitol Hill are looking to mine state capitals for legislative victories. From a report: A broad bipartisan federal privacy bill that died in Congress last year has quickly become the template for a statehouse-by-statehouse campaign to enact tough new restrictions on how Americans' personal data can be mined and shared. Lawmakers in Massachusetts and Illinois are already proposing privacy measures modeled on the federal bill, and Democrats in Indiana are using it as inspiration to strengthen legislation that's already been proposed. Four other states have already passed their own data-privacy laws in the past two years -- raising anxiety levels among tech companies about a national "patchwork" of hard-to-navigate data rules -- but encouraging advocates who see an appetite for broader consumer protections.

"We were wondering if there would be something passed federally. It would definitely guide what we would be doing for the state," Democratic Indiana state Sen. Shelli Yoder said in an interview. "Because that failed, it put us in a position of needing to do something." The new statehouse focus by privacy advocates isn't necessarily designed to sweep across all 50 states but rather tighten regulations just enough in just enough places to force the industry into a de facto national standard. They're hoping to enact state-level privacy proposals that align closely with what Congress attempted to pass with the American Data and Privacy Protection Act: regulations that would limit what data companies can collect and share, create a data broker registry and establish new rights for Americans to delete data about themselves. But they're playing catch-up to an industry-led campaign that's made significant headway in several states, including Virginia and Utah, where weaker laws were enacted over the past two years.

Google has claimed a breakthrough in correcting for the errors that are inherent in today's quantum computers, marking an early but potentially significant step in overcoming the biggest technical barrier to a revolutionary new form of computing. From a report: The internet company's findings, which have been published in the journal Nature, mark a "milestone on our journey to build a useful quantum computer," said Hartmut Neven, head of Google's quantum efforts. He called error correction "a necessary rite of passage that any quantum computing technology has to go through."

Quantum computers struggle to produce useful results because the quantum bits, or qubits, they are based on only hold their quantum states for a tiny fraction of a second. That means information encoded in a quantum system is lost before the machine can complete its calculations. Finding a way to correct for the errors this causes is the hardest technical challenge the industry faces. [...] Google's researchers said they had found a way to spread the information being processed in a quantum computer across a number of qubits in a way that meant the system as a whole could retain enough to complete a calculation, even as individual qubits fell out of their quantum states. The research published in Nature pointed to a reduction of only 4 per cent in the error rate as Google scaled up its technique to run on a larger quantum system. However, the researchers said this was the first time that increasing the size of the computer had not also led to a rise in the error rate.

The U.S. Supreme Court, hearing a case that could reshape the internet, considered on Tuesday whether Google bears liability for user-generated content when its algorithms recommend videos to users. From a news writeup: In the case, Gonzalez vs, Google, the family of a terrorist attack victim contends that YouTube violated the federal Anti-Terrorism Act because its algorithm recommended ISIS videos to users, helping to spread their message. Nohemi Gonzalez was an American student killed in a 2015 ISIS attack in Paris, and his family's lawsuit challenges the broad legal immunity that tech platforms enjoy for third party content posted on their sites. Section 230 of the Communications Decency Act, passed in 1996, protects platforms from legal action over user-generated content, and it also protects them if they choose to remove content. Section 230 has withstood court challenges for the past three decades even as the internet exploded.

The attorney for Gonzalez's family claimed that YouTube's recommendations fall outside the scope of Section 230, as it is the algorithms, not the third party, that actively pick and choose where and how to present content. In this case, the attorney said, it enhanced the ISIS message. "Third parties that post on YouTube don't direct their videos to specific users," said the Gonzalez's attorney Eric Schnapper. Instead, he said, those are choices made by the platform. Justice Neil Gorsuch said he was '"not sure any algorithm is neutral. Most these days are designed to maximize profit." [...] Internet firms swear that removing or limiting 230 protections would destroy the medium. Would it? Chief Justice John Roberts asked Google's attorney Lisa Blatt. "Would Google collapse and the internet be destroyed if Google was prevented from posting what it knows is defamatory?" She said, "Not Google," but other, smaller websites, yes. She said if the plaintiffs were victorious, the internet would become a zone of extremes -- either The Truman Show, where things are moderated into nothing, or like "a horror show," where nothing is.

Facebook parent company Meta is preparing for a fresh round of job cuts, deputizing human resources, lawyers, financial experts and top executives to draw up plans to deflate the company's hierarchy, in a reorganization and downsizing effort that could affect thousands of workers. From a report: Meta plans to push some leaders into lower-level roles without direct reports, flattening the layers of management between Meta CEO Mark Zuckerberg and the company's interns, according to a person familiar with the matter who spoke on the condition of anonymity because they were not authorized to speak on about internal matters. Other managers may end up overseeing a higher number of employees as their teams grow bigger. Some inside Meta expect employees whose jobs have been converted to eventually quit, trimming the company's workforce by default. In addition to targeting managers, the company is also considering more traditional cuts, including slashing some projects and jobs, the person said. These efforts, which are targeted at divisions across the company and around the world, may not happen on a single day, but will likely roll out across the company in the coming months.

Subway said on Tuesday it plans to add charging parks to select restaurants. "Dubbed Subway Oasis, the EV parks will be outfitted with 'charging canopies with multiple ports, picnic tables, Wi-Fi, restrooms, green space, and even playgrounds,'" reports Insider. From the report: Subway is working with EV tech startups GenZ EV Solution and RED E Charging to open these parks. Additionally, the company said that Subway is opening smaller fast-charging EV stations at new or newly remodeled restaurants across the US this year. "On average, the smaller-format, fast EV chargers will offer a 120-mile charge in 17 minutes for approximately $20," the company said. Once open, EV customers might also get the added perk of receiving Subway discounts while waiting for their cars to charge, the company said. Subway did not specify how much it would cost consumers to charge their cars at their new charging stations, nor did they mention where and when the first Subway Oasis would be built.

Tumblr's parody of paid verification has already delivered the social network and blogging platform a 125% boost in iOS in-app purchase revenue since November, according to a new analysis of the app's in-app consumer spending. TechCrunch reports: The company, now operated by WordPress.com owner Automattic following its 2019 acquisition, launched its response to Twitter's paid verification hustle with the addition of its own purely cosmetic double blue checks -- a sort of tongue-in-cheek rebuttal to the idea that subscription-based verification had any real value. As it turns out, at least some Tumblr users were willing to pay -- though perhaps not for clout, but because in-jokes have proven to be a more successful monetization strategy for the blogging network than some of its more legitimate attempts to make money, such as its creator-focused subscription, Post+. After being met with community backlash, at one point Post+ was being outperformed from a monetization perspective by crabs -- a goofy paid feature that let users send animated dancing crabs to each other's dashboards.

According to new data from app intelligence firm Sensor Tower provided to TechCrunch, consumer spending on Tumblr's iOS app increased since November 2022's double-blue check launch, now totaling $263,000 in net revenue. While that's not a significant figure in the grand scheme of things by any means, it still represents a 125% jump in spending compared with the prior three-month total of August through October 2022. When looking at more long-term trends, Tumblr's revenue remains up -- but not by as much. Sensor Tower says the in-app purchase revenue on iOS is up 19%, compared with the prior ten months ahead of the blue check's launch (January through October 2022).

Starry, an ISP that launched in 2016 with a focus on delivering home internet with wireless antennas instead of cables, has declared bankruptcy. The Verge reports: In a press release (PDF), the company says that it intends to quickly restructure and that it'll continue providing internet service in its "five core operating markets." Those are Boston, New York City, Los Angeles, Denver, and Washington, DC. The ISP has clearly been struggling over the past few months. In October 2022, it announced that it was laying off around 500 people, which amounted to about half of its staff. A few months later, Starry announced it was leaving Columbus, Ohio, in a bid to focus more on its five "core" markets. All the while, it was burning millions of dollars in cash, and its stock was dropping after a special purpose acquisition company-backed IPO in March -- it started at around $10 a share but is now worth $0.012, down from last week when it was approximately $0.02 per share.

The company also defaulted on its Rural Digital Opportunity Fund bids after it won awards from the FCC to work on providing internet to underserved areas in the US. Had it completed the work, it stood to receive almost $269 million, according to Light Reading. Starry has asked the United States Bankruptcy Court for the District of Delaware to approve a plan that would give it $43 million in funding from lenders, which it says would provide "the necessary liquidity to continue its normal business operations and meet its post-filing obligations to its employees, customers and vendors." "With the support of our lenders, we feel confident in our ability to successfully exit this process as a stronger company, well-positioned to continue" providing internet to customers, said Starry CEO Chet Kanojia in the company's press release.

The European Union is (finally) coming to grips with the dysfunctionalities of its most famous tech law of all: the General Data Protection Regulation. From a report: The European Commission will propose a new law before the summer that's aimed at improving how EU countries' privacy regulators enforce the GDPR, a newly published page on its website showed. Adopted in 2016, the privacy rulebook was a watershed moment in global tech regulation, forcing companies to abide by new standards such as asking for consent to collect people's data online against threats of hefty fines of up to 4 percent of global annual turnover. The law effectively became European officials' poster child of powerful legislation coming out of Brussels. But five years after EU data protection authorities started their job, as GDPR entered into force, activists, experts and some national privacy watchdogs have become frustrated at what they see as an inefficient system to tackle major cases, especially from Big Tech companies.

Most notably, critics have lamented the powerful role that the Irish Data Protection Commission has under the so-called one-stop shop rule, which directs most major investigations to run through the Irish system because tech companies like Meta, Google, Apple and others have set up their European homes there. Under the GDPR, tech companies are overseen by the national regulator in the EU country where they are headquartered. Ireland and, to a lesser extent, Luxembourg, where Amazon's EU headquarters is based, have faced mounting criticism in recent years for lax enforcement, which they deny. The Irish data authority in recent months imposed some major multimillion-euro fines to sanction GDPR infringements from Meta, the parent company of Instagram and Facebook. Now, a new EU regulation that is expected in the second quarter of 2023 wants to set clear procedural rules for national data protection authorities dealing with cross-border investigations and infringements. The law "will harmonize some aspects of the administrative procedure" in cross-border cases and " support a smooth functioning of the GDPR cooperation and dispute resolution mechanisms," the Commission wrote.

Crypto exchange Coinbase has confirmed that it was briefly compromised by the same attackers that targeted Twilio, Cloudflare, DoorDash, and more than a hundred other organizations last year. From a report: In a post-mortem of the incident published over the weekend, Coinbase said that the so-called '0ktapus' hackers stole the login credentials of one of its employees in an attempt to remotely gain access to the company's systems. 0ktapus is a hacking group that has targeted more than 130 organizations in 2022 as part of an ongoing effort to steal the credentials of thousands of employees, often by impersonating Okta log-in pages. That figure of 130 organizations is now likely much higher, as a leaked Crowdstrike report seen by TechCrunch claims that the gang is now targeting several tech and video game companies.

Microsoft has partnered with web3 infrastructure provider Ankr to offer a node service for enterprises in need of blockchain data access. From a report: The two firms will work together on a new node hosting service in Microsoft's Azure cloud marketplace, with tailored memory and bandwidth specifications for blockchain nodes. The enterprise node deployment service would enable web3 projects or developers to deploy smart contracts, relay transactions and read or write blockchain data, according to a company release. "Our partnership with Ankr will enable developers and organizations to access blockchain data in a reliable and secure way as they explore how web3 can address real-world business challenges," said Rashmi Misra, Microsoft's general manager for AI and emerging technologies in the release. "Together, we are building a robust web3 infrastructure layer."

Steven Vaughan-Nichols, writing for ZDNet: Linux 6.2 was released yesterday, and Linus Torvalds described the latest Linux kernel release as, "Maybe it's not a sexy LTS release like 6.1 ended up being, but all those regular pedestrian kernels want some test love too." For once, I disagree with Torvalds. By adding upstream support for the Apple M1 Pro, M1 Max, and M1 Ultra chips, newer Mac owners can look forward to running Linux on their M1-powered machines. And, for techies, that's sexy. Getting Linux to run on the M1 family wasn't easy.

When these high-powered ARM chips first arrived, Torvalds told me in an exclusive interview that he'd like to run Linux on these next-generation Macs. But, while he'd been "waiting for an ARM laptop that can run Linux for a long time," he worried, saying, "The main problem with the M1 for me is the GPU and other devices around it because that's likely what would hold me off using it because it wouldn't have any Linux support unless Apple opens up."
UPDATE (2/26/2023): Asahi Linux called ZDNet's report "misleading and borderline false," posting on Twitter that "You will not be able to run Ubuntu nor any other standard distro with 6.2 on any M1 Mac. Please don't get your hopes up." We are continuously upstreaming kernel features, and 6.2 notably adds device trees and basic boot support for M1 Pro/Max/Ultra machines. However, there is still a long road before upstream kernels are usable on laptops. There is no trackpad/keyboard support upstream yet.

While you can boot an upstream 6.2 kernel on desktops (M1 Mac Mini, M1 Max/Ultra Mac Studio) and do useful things with it, that is only the case for 16K page size kernel builds. No generic ARM64 distro ships 16K kernels today, to our knowledge.

Our goal is to upstream everything, but that doesn't mean distros instantly get Apple Silicon support. As with many other platforms, there is some integration work required. Distros need to package our userspace tooling and, at this time, offer 16K kernels. In the future, once 4K kernel builds are somewhat usable, you can expect zero-integration distros to somewhat work on these machines (i.e. some hardware will work, but not all, or only partially). This should be sufficient to add a third-party repo with the integration packages.

But for out-of-the-box hardware support, distros will need to work with us to get everything right. We are already working with some, and we expect to announce official Apple Silicon support for a mainstream distro in the near future. Just not quite yet!

Key members of China's most influential scientific body have outlined the country's plan to circumvent US chip sanctions for the first time, codifying Beijing's view of how it could win a crucial technological conflict with Washington. From a report: Two of the country's senior academics wrote that Beijing should amass a portfolio of patents that govern the next generation of chipmaking, from novel materials to new techniques. That should propel its semiconductor ambitions while giving China the clout to push back against US sanctions designed to hamstring its semiconductor sector, Luo Junwei and Li Shushen wrote in the bulletin of the Chinese Academy of Sciences.

The article, published to a social media account affiliated with the academy, offers a rare glimpse into how Beijing thinks about and might react to the Biden administration's escalating hostilities over semiconductors. The academy advises China's top decision makers and the article echoes remarks by President Xi Jinping calling for victory in developing core technologies. It comes as the country's new technology overseer outlined his vision for moving past American sanctions, stressing the need to modernize and rectify weak links in its supply chain. China has a plan to develop next-generation chip materials that it put in place in 2020 as a reaction to Trump-era restrictions. Yet that national strategy has yet to yield a technological edge on the world's leading chipmakers. Washington has implemented a series of measures limiting exports of technology such as chipmaking gear and artificial intelligence processors to China, part of a broader set of technology sanctions.

Google has released optimization features designed to improve battery life and memory usage on machines running the latest version of its Chrome desktop web browser. From a report: Chrome's new Energy Saver and Memory Saver modes were first announced in December last year alongside the release of Chrome 108, and now as noted by Android Police, the two optimization utilities are starting to roll out globally onto Chrome 110 desktops for Mac, Windows, and Chromebooks.

Memory Saver mode essentially snoozes Chrome tabs that aren't currently in use to free up RAM for more intensive tasks and create a smoother browsing experience. Don't worry if you're a tab hoarder though, as these inactive tabs are still visible and can be reloaded at any time to pick up where you left off. Your most used websites can also be marked as exempt from Memory Saver to ensure they're always running at the maximum possible performance.

New submitter calicuse writes: Microsoft's Outlook spam filters appear to be broken for many users today. I woke up to more than 20 junk messages in my Focused Inbox in Outlook this morning, and spam emails have kept breaking through on an hourly basis today. Many Outlook users in Europe have also spotted the same thing, with some heading to Twitter to complain about waking up to an inbox full of spam messages. Most of the messages that are making it into Outlook users' inboxes are very clearly spam. Today's issues are particularly bad, after weeks of the Outlook spam filter progressively deteriorating for me personally.

An anonymous reader shares a report: Did you force your PC to install Windows 11 despite it not meeting the official requirements? Microsoft might start nagging you for doing that -- or at least reminding you that what you've done is against the intended use of its operating system. The January 2023 Windows 11 update is pestering folks who forced the update on their PCs with a persistent watermark on the desktop warning that system requirements haven't been met. The story is circulating among Windows blogs, though I found a couple of instances of folks complaining about the watermark on the official Microsoft support forums.

The watermark says "system requirements not met" and is emblazoned on the desktop's lower right hand corner if the operating system notices that it's running on hardware that doesn't meet the minimum requirements. It's possible the culprit is the dedicated security processor, or TPM 2.0 (Trusted Platform Module) chip, used by services like BitLocker and Windows Hello. Microsoft requires this module before upgrading. It's why many PCs were rendered un-upgradeable when Windows 11 was announced. Most new CPUs and motherboards have capability for it built into them, but the feature wasn't a guaranteed inclusion prior to the Windows 11 launch.

A hedge fund that was one of the highest-profile victims of the FTX scandal when half its assets were trapped on the collapsed cryptocurrency exchange has decided to close and return its remaining money to investors. From a report: Galois Capital, which last year had been managing about $200mn in assets and was one of the biggest crypto-focused quantitative funds, told investors that it had halted all trading and unwound all its positions as it was no longer viable, according to documents seen by the Financial Times. "Given the severity of the FTX situation, we do not think it is tenable to continue operating the fund both financially and culturally," wrote co-founder Kevin Zhou. "Once again I'm terribly sorry about the current situation we find ourselves in."

The FT revealed in November that Galois, despite pulling out some money, still had about half its assets stuck on FTX when the exchange collapsed. In a situation reminiscent of Lehman Brothers in 2008, hedge funds were left with billions of dollars trapped on the exchange, with many having viewed it as one of the more reputable trading platforms in an often lightly regulated or unregulated industry. As many as 1mn creditors have been identified in FTX's Delaware bankruptcy. Its founder, Sam Bankman-Fried, is due to face trial in October on fraud charges, to which he has pleaded not guilty. Since sending the letter, Galois has sold its claim for approximately 16 cents on the dollar.