Microsoft-owned GitHub published a blog post asking "Does GitHub Copilot improve code quality? Here's what the data says." Its first paragraph includes statistics from past studies — that GitHub Copilot has helped developers code up to 55% faster, leaving 88% of developers feeling more "in the flow" and 85% feeling more confident in their code. But does it improve code quality? [W]e recruited 202 [Python] developers with at least five years of experience. Half were randomly assigned GitHub Copilot access and the other half were instructed not to use any AI tools... We then evaluated the code with unit tests and with an expert review conducted by developers. Our findings overall show that code authored with GitHub Copilot has increased functionality and improved readability, is of better quality, and receives higher approval rates... Developers with GitHub Copilot access had a 56% greater likelihood of passing all 10 unit tests in the study, indicating that GitHub Copilot helps developers write more functional code by a wide margin. In blind reviews, code written with GitHub Copilot had significantly fewer code readability errors, allowing developers to write 13.6% more lines of code, on average, without encountering readability problems. Readability improved by 3.62%, reliability by 2.94%, maintainability by 2.47%, and conciseness by 4.16%. All numbers were statistically significant... Developers were 5% more likely to approve code written with GitHub Copilot, meaning that such code is ready to be merged sooner, speeding up the time to fix bugs or deploy new features. "While GitHub's reports have been positive, a few others haven't," reports Visual Studio magazine: For example, a recent study from Uplevel Data Labs said, "Developers with Copilot access saw a significantly higher bug rate while their issue throughput remained consistent." And earlier this year a "Coding on Copilot" whitepaper from GitClear said, "We find disconcerting trends for maintainability. Code churn — the percentage of lines that are reverted or updated less than two weeks after being authored — is projected to double in 2024 compared to its 2021, pre-AI baseline. We further find that the percentage of 'added code' and 'copy/pasted code' is increasing in proportion to 'updated,' 'deleted,' and 'moved 'code. In this regard, AI-generated code resembles an itinerant contributor, prone to violate the DRY-ness [don't repeat yourself] of the repos visited."

Read more of this story at Slashdot.

June, 2023: "Harvard Scholar Who Studies Honesty Is Accused of Fabricating Findings." November, 2024: "The Business-School Scandal That Just Keeps Getting Bigger." A senior editor at the Atlantic raises the possibility of systemic dishonesty-rewarding incentives where "a study must be even flashier than all the other flashy findings if its authors want to stand out," writing that "More than a year since all of this began, the evidence of fraud has only multiplied." And the suspect isn't just Francesca Gino, a Harvard Business School professor. One person deeply affected by all this is Gino's co-author, a business school professor from the University of California at Berkeley — Juliana Schroeder — who launched an audit of all 138 studies conducted by Francesca Gino (called "The Many Coauthors Project"): Gino was accused of faking numbers in four published papers. Just days into her digging, Schroeder uncovered another paper that appeared to be affected — and it was one that she herself had helped write... The other main contributor was Alison Wood Brooks, a young professor and colleague of Gino's at Harvard Business School.... If Brooks did conduct this work and oversee its data, then Schroeder's audit had produced a dire twist. The Many Co-Authors Project was meant to suss out Gino's suspect work, and quarantine it from the rest... But now, to all appearances, Schroeder had uncovered crooked data that apparently weren't linked to Gino.... Like so many other scientific scandals, the one Schroeder had identified quickly sank into a swamp of closed-door reviews and taciturn committees. Schroeder says that Harvard Business School declined to investigate her evidence of data-tampering, citing a policy of not responding to allegations made more than six years after the misconduct is said to have occurred... In the course of scouting out the edges of the cheating scandal in her field, Schroeder had uncovered yet another case of seeming science fraud. And this time, she'd blown the whistle on herself. That stunning revelation, unaccompanied by any posts on social media, had arrived in a muffled update to the Many Co-Authors Project website. Schroeder announced that she'd found "an issue" with one more paper that she'd produced with Gino... [Schroeder] said that the source of the error wasn't her. Her research assistants on the project may have caused the problem; Schroeder wonders if they got confused... What feels out of reach is not so much the truth of any set of allegations, but their consequences. Gino has been placed on administrative leave, but in many other instances of suspected fraud, nothing happens. Both Brooks and Schroeder appear to be untouched. "The problem is that journal editors and institutions can be more concerned with their own prestige and reputation than finding out the truth," Dennis Tourish, at the University of Sussex Business School, told me. "It can be easier to hope that this all just goes away and blows over and that somebody else will deal with it...." [Tourish also published a 2019 book decrying "Fraud, Deception and Meaningless Research," which the article notes "cites a study finding that more than a third of surveyed editors at management journals say they've encountered fabricated or falsified data."] Maybe the situation in her field would eventually improve, [Schroeder] said. "The optimistic point is, in the long arc of things, we'll self-correct, even if we have no incentive to retract or take responsibility." "Do you believe that?" I asked. "On my optimistic days, I believe it." "Is today an optimistic day?" "Not really."

Read more of this story at Slashdot.

Ahead of FreeBSD 14.2 hopefully releasing in just over one week, FreeBSD 14.2-RC1 is out this weekend as the last planned development release for testing ahead of that much anticipated point release...

Meta wants to force Apple and Google to verify the ages of people downloading apps from their app stores, reports the Washington Post — and now Meta's campaign "is picking up momentum" with legislators in the U.S. Congress. Federal and state lawmakers have recently proposed a raft of measures requiring that platforms such as Meta's Facebook and Instagram block users under a certain age from using their sites. The push has triggered fierce debate over the best way to ascertain how old users are online. Last year Meta threw its support behind legislation that would push those obligations onto app stores rather than individual app providers, like itself, as your regular host and Naomi Nix reported. While some states have considered the plan, it has not gained much traction in Washington. That could be shifting. Two congressional Republicans are preparing a new age verification bill that places the burden on app stores, according to two people familiar with the matter, who spoke on the condition of anonymity to discuss the plans... The bill would be the first of its kind on Capitol Hill, where lawmakers have called for expanding guardrails for children amid concerns about the risks of social media but where political divisions have bogged down talks. The measure would give parents the right to sue an app store if their child was exposed to certain content, such as lewd or sexual material, according to a copy obtained by the Tech Brief. App stores could be protected against legal claims, however, if they took steps to protect children against harms, such as verifying their ages and giving parents the ability to block app downloads. The article points out that U.S. lawmakers "have the power to set national standards that could override state efforts if they so choose..."

Read more of this story at Slashdot.

CNN reports that investigators "are trying to crack the mystery of how two undersea internet cables in the Baltic Sea were cut within hours of each other." But there's now two competing viewpoints, "with European officials saying they believe the disruption was an act of sabotage and U.S. officials suggesting it was likely an accident." The foreign ministers of Finland and Germany said in a joint statement that they were "deeply concerned" about the incident and raised the possibility that it was part of a "hybrid warfare," specifically mentioning Russia in their statement. Their assessment was not plucked out of thin air. Russia has been accused of waging a hybrid war against Europe after a string of suspicious incidents, arson attacks, explosions and other acts of sabotage across multiple European countries were traced back to Moscow. And the disruption to the cables came just weeks after the US warned that Moscow was likely to target critical undersea infrastructure. This followed months of suspicious movements of Russian vessels in European waters and the significant beefing up of a dedicated Russian secretive marine unit tasked with surveying the seabed... But two US officials familiar with the initial assessment of the incident told CNN on Tuesday the damage was not believed to be deliberate activity by Russia or any other nation. Instead, the two officials told CNN they believed it likely caused by an anchor drag from a passing vessel. Such accidents have happened in the past, although not in a quick succession like the two on Sunday and Monday. Cloudflare's blog also reminds readers that the two cable cuts resulted in little-to-no observable impact Cloudflare attributes this largely to "the significant redundancy and resilience of Internet infrastructure in Europe." (Their Cloudflare Radar graphs show that after the Sweden-Lithuania cable cut "there was no apparent impact to traffic volumes in either country at the time that the cables were damaged.") Telegeography's submarinecablemap.com illustrates, at least in part, the resilience in connectivity enjoyed by these two countries. In addition to the damaged cable, it shows that Lithuania is connected to neighboring Latvia as well as to the Swedish mainland. Over 20 submarine cables land in Sweden, connecting it to multiple countries across Europe. In addition to the submarine resilience, network providers in both countries can take advantage of terrestrial fiber connections to neighboring countries, such as those illustrated in a European network map from Arelion (formerly Telia), which is only one of the large European backbone providers. Less than a day later, the C-Lion1 submarine cable, which connects Helsinki, Finland and Rostock Germany was reportedly damaged during the early morning hours of Monday, November 18... In this situation as well, as the Cloudflare Radar graphs below show, there was no apparent impact to traffic volumes in either country at the time that the cables were damaged... Telegeography's submarinecablemap.com shows that both Finland and Germany also have significant redundancy and resilience from a submarine cable perspective, with over 10 cables landing in Finland, and nearly 10 landing in Germany, including Atlantic Crossing-1 (AC-1), which connects to the United States over two distinct paths. Terrestrial fiber maps from Arelion and eunetworks (as just two examples) show multiple redundant fiber routes within both countries, as well as cross-border routes to other neighboring countries, enabling more resilient Internet connectivity. See also Does the Internet Route Around Damage?

Read more of this story at Slashdot.

Slashdot reader jjslash shared this report from TechSpot: The SilverStone FLP01 made quite the impression when it was shared on X for April Fools' Day 2023. Loosely modeled after popular desktops from yesteryear like the NEC PC-9800 series, the chassis features dual 5.25-inch faux floppy bays that could stand to look a bit more realistic. Notably, the covers flip open to reveal access to a more modern (yet still legacy) optical drive and front I/O ports. Modern-looking fan grills can be found on either side of the desktop, serving as yet another hint that the chassis is not as old at it appears on first glance. The grills look to be removable, and probably hold washable dust filters. Like early desktops, the system doubles as a stand for your monitor. The use of a green power LED up front helps round out the retro look; a red LED is used as a storage activity indicator.

Read more of this story at Slashdot.

Instead of focusing on chatbots, a new study reveals an automated way to breach LLM-driven robots "with 100 percent success," according to IEEE Spectrum. "By circumventing safety guardrails, researchers could manipulate self-driving systems into colliding with pedestrians and robot dogs into hunting for harmful places to detonate bombs..." [The researchers] have developed RoboPAIR, an algorithm designed to attack any LLM-controlled robot. In experiments with three different robotic systems — the Go2; the wheeled ChatGPT-powered Clearpath Robotics Jackal; and Nvidia's open-source Dolphins LLM self-driving vehicle simulator. They found that RoboPAIR needed just days to achieve a 100 percent jailbreak rate against all three systems... RoboPAIR uses an attacker LLM to feed prompts to a target LLM. The attacker examines the responses from its target and adjusts its prompts until these commands can bypass the target's safety filters. RoboPAIR was equipped with the target robot's application programming interface (API) so that the attacker could format its prompts in a way that its target could execute as code. The scientists also added a "judge" LLM to RoboPAIR to ensure the attacker was generating prompts the target could actually perform given physical limitations, such as specific obstacles in the environment... One finding the scientists found concerning was how jailbroken LLMs often went beyond complying with malicious prompts by actively offering suggestions. For example, when asked to locate weapons, a jailbroken robot described how common objects like desks and chairs could be used to bludgeon people. The researchers stressed that prior to the public release of their work, they shared their findings with the manufacturers of the robots they studied, as well as leading AI companies. They also noted they are not suggesting that researchers stop using LLMs for robotics... "Strong defenses for malicious use-cases can only be designed after first identifying the strongest possible attacks," Robey says. He hopes their work "will lead to robust defenses for robots against jailbreaking attacks." The article includes a reaction from Hakki Sevil, associate professor of intelligent systems and robotics at the University of West Florida. He concludes that the "lack of understanding of context of consequences" among even advanced LLMs "leads to the importance of human oversight in sensitive environments, especially in environments where safety is crucial." But a long-term solution could be LLMs with "situational awareness" that understand broader intent. "Although developing context-aware LLM is challenging, it can be done by extensive, interdisciplinary future research combining AI, ethics, and behavioral modeling..." Thanks to long-time Slashdot reader DesertNomad for sharing the article.

Read more of this story at Slashdot.

"You can use any Linux distribution inside of the Windows Subsystem for Linux" Microsoft recently reminded Windows users, "even if it is not available in the Microsoft Store, by importing it with a tar file." But being an official distro "makes it easier for Windows Subsystem for Linux users to install and discover it with actions like wsl --list --online and wsl --install," Microsoft pointed out this week. And "We're excited to announce that Red Hat will soon be delivering a Red Hat Enterprise Linux WSL distro image in the coming months..." Thank you to the Red Hat team as their feedback has been invaluable as we built out this new architecture, and we're looking forwards to the release...! Ron Pacheco, senior director, Red Hat Enterprise Linux Ecosystem, Red Hat says: "Developers have their preferred platforms for developing applications for multiple operating systems, and WSL is an important platform for many of them. Red Hat is committed to driving greater choice and flexibility for developers, which is why we're working closely with the Microsoft team to bring Red Hat Enterprise Linux, the largest commercially available open source Linux distribution, to all WSL users." Read Pacheco's own blog post here. But in addition Microsoft is also releasing "a new way to make WSL distros," they announced this week, "with a new architecture that backs how WSL distros are packaged and installed." Up until now, you could make a WSL distro by either creating an appx package and distributing it via the Microsoft Store, or by importing a .tar file with wsl -import. We wanted to improve this by making it possible to create a WSL distro without needing to write Windows code, and for users to more easily install their distros from a file or network share which is common in enterprise scenarios... With the tar based architecture, you can start with the same .tar file (which can be an exported Linux container!) and just edit it to add details to make it a WSL distro... These options will describe key distro attributes, like the name of the distro, its icon in Windows, and its out of box experience (OOBE) which is what happens when you run WSL for the first time. You'll notice that the oobe_command option points to a file which is a Linux executable, meaning you can set up your full experience just in Linux if you wish.

Read more of this story at Slashdot.

Minimachines.net en partenariat avec Geekbuying.com
Les Week-End du Bon Plan : 23-24/11/20245 © MiniMachines.net. 2024.

"The past 15 years were unique in ways that might be a bad predictor of our future," writes the Washington Post, with a surge in the number of internet users since 2010, and everyone spending more time online. But today, "lots of smart people believe that artificial intelligence will upend how you find information. Googling is so yesterday." Sam Altman, the top executive overseeing ChatGPT, has said that AI has a good shot at shoving aside Google search. Bill Gates predicted that emerging AI will do tasks like researching your ideal running shoes and automatically placing an order so you'll "never go to a search site again." In defending itself from a judge's decision that it runs an illegal monopoly, Google says the company might be roadkill as AI and other new technologies change how you find information. (On Wednesday, the U.S. government asked the judge to overhaul Google to undo its monopoly.) But predictions of Google's looming obsolescence have been wrong before, which calls for humility in fortune-telling our collective technology habits. We're devilishly unpredictable.... Maybe it's right to extrapolate from how people are starting to use AI today. Or maybe that's the mistake that Jobs made when he said no one was searching on iPhones. It wasn't wrong in 2010, but it was within a few years. Or what if AI upends how billions of us find information and we still keep on Googling? "The notion that we can predict how these new technologies are going to evolve is silly," said David B. Yoffie, a Harvard Business School professor who has spent decades studying the technology industry. Amit Mehta, the judge overseeing the Google monopoly case, formed his own view on AI moving us away from searching Google. "AI may someday fundamentally alter search, but not anytime soon," he said.

Read more of this story at Slashdot.

An anonymous reader shared this report from CNET: Meta says it's taken down more than 2 million accounts this year linked to overseas criminal gangs behind scam operations that human rights activists say forced hundreds of thousands of people to work as scammers and cost victims worldwide billions of dollars. In a Thursday blog post, the parent of Facebook, Instagram and WhatsApp says the pig butchering scam operations — based in Myanmar, Laos, Cambodia, the United Arab Emirates and the Philippines — use platforms like Facebook and Instagram; dating, messaging, crypto and other kinds of apps; and texts and emails, to globally target people... [T]he scammers strike up an online relationship with their victims and gain their trust. Then they move their conversations to crypto apps or scam websites and dupe victims into making bogus investments or otherwise handing over their money, Meta said. They'll ask the victims to deposit money, often in the form of cryptocurrency, into accounts, sometimes even letting the victims make small withdrawals, in order to add a veneer of legitimacy. But once the victim starts asking for their investment back, or it becomes clear they don't have any more money to deposit, the scammer disappears and takes the money with them. And the people doing the scamming are often victims themselves. During the COVID-19 pandemic, criminal gangs began building scam centers in Southeast Asia, luring in often unsuspecting job seekers with what looked like amazing postings on local job boards and other platforms, then forcing them to work as scammers, often under the threat of physical harm. The scope of what's become a global problem is staggering. In a report issued in May, the US Institute of Peace estimates that at least 300,000 people are being forced to work, or are otherwise suffering human rights violations, inside these scam centers. The report also estimates global financial losses stemming from the scams at $64 billion in 2023, with the number of financial victims in the millions. Meta says it has focused on investigating and disrupting the scam operations for more than two years, working with nongovernmental organizations and other tech companies, like OpenAI, Coinbase and dating-app operator Match Group, along with law enforcement in both the US and the countries where the centers are located. Meta titled its blog post "Cracking Down On Organized Crime Behind Scam Centers," writing "We hope that sharing our insights will help inform our industry's defenses so we can collectively help protect people from criminal scammers."

Read more of this story at Slashdot.

A Microsoft Research project that was quietly announced a few years ago to some fanfare but not hearing much about since has been Demikernel as their library OS architecture for kernel-bypass I/O. A Phoronix reader brought up Demikernel this week and while it hasn't been talked about much in recent years it does remain under active development with the most recent commits as of hours ago...

IBM isn't formally releasing Power11 processors until next year, but their software engineers continue being quite busy preparing the Linux kernel and other open-source software for Power11. The newest on the kernel side is enabling support for KVM nested guests on IBM Power11 platforms...

Episode four nucléaire : a new unhope

Le monde s’assombrit et ce n’est pas le Black Friday, même si ce dernier bat son plein d’espérances creuses. Le pro des mauvaises « pranks » géopolitiques tire les ficelles un peu trop fort, mais ça en coupe une sans bouger l’autre. Ça pète de partout que voulez-vous, heureusement qu’il reste quelques bonnes âmes pour vous tendre la main gratuitement, autrement que pour vous la claquer sur la joue ! Certes, ils vous câbleront pour avoir une bonne santé numérique à défaut de vous aider à soigner l’autre, mais ne voyons pas le verre à moitié vide, surtout quand il ne reste qu’un petit fond. Au moins, le service rendu sera données. Quitte à être dans l’aide et le partage, on pourra toujours les céder volontiers à ces pirates de la data qui se lèvent pour une noble cause : votre bien-être numérique (ENCORE ?) si si. Bon, enrobé d’un peu d’argent de poche, il faut bien vivre vous dirait la CNIL. Bien vivre pour pouvoir regarder le temps qui passe au travers des bougies soufflées.
C’est décidément important le bien-être, certains pensent au leur en tentant de sortir du chemin de croix pour espérer un peu de ciel bleu dans toute cette grisaille… Avouons-le, c’est bien humain.

Longtime Slashdot reader theodp writes: Past corporate-sponsored Hour of Code tutorials for the nation's schoolchildren have blurred the lines between coding lessons and product infomercials. So too is the case again with this year's newly-announced Hour of Code 2024 flagship tutorials, which include Microsoft Minecraft, Amazon Music, and Transformers One movie-themed intros to coding. The press release announcing the tutorials from tech-backed nonprofit Code.org, which organizes the Hour of Code and counts Microsoft and Amazon as $30+ million donors, boasts of its "decade of partnership with [Microsoft] Minecraft this year, reaching more than 300 million sessions of Minecraft Hour of Code since 2015!" Interestingly, The Transformers (Paramount Pictures, which released Transformers One in the U.S., is a $25,000+ Code.org donor) is cited as one of the OG's of children's Saturday morning cartoon advertising (aka 30-minute commercials) that prompted the Children's Television Act (CTA) of 1990, an act of Congress that ordered the FCC to put in place regulations to protect children from advertising. Throughout the 1980s, Action for Children's Television (ACT) criticized children's television programs that "blur(red) the distinction between program content and commercial speech."

Read more of this story at Slashdot.

For those still running a 1st Generation Xeon Scalable "Skylake" era server, support for it within the open-source Coreboot firmware may continue to improve all these years later thanks to firmware consulting firm 9elements...

Wine 9.22 is out this weekend ahead of the Wine 10.0-rc1 in two weeks...

KDE developer Nate Graham is out with his traditional weekly development recap that highlights all of the interesting work taking place within the KDE desktop space...

Nous sommes normalement dans la période des promotions ultimes... Oui, mais voilà cela ne semble pas changer grand-chose sur les prix des cartes graphiques, comme souvent... Chez AMD, nous avons la RX 7600 XT qui augmente de 7 euros, la RX 7700 XT qui baisse de 3 euros, la 7800 XT qui prend un euro, la RX 7900 XT qui revient à son prix de 699 euros et la RX 7900 XTX qui passe de 977 à 949 euros. […]

Lire la suite