news aggregator

Infosec academic suggests Beijing’s warning that iThing owners aren’t anonymous deserves attention outside the great firewall, too

In June 2023 China made a typically bombastic announcement: operators of short-distance ad hoc networks must ensure they run according to proper socialist principles, and ensure all users divulge their real-world identities.…

Generative AI "doesn't easily carry over into robotics," write two researchers in IEEE Spectrum, "because the Internet is not full of robotic-interaction data in the same way that it's full of text and images." That's why they're working on a single deep neural network capable of piloting many different types of robots... Robots need robot data to learn from, and this data is typically created slowly and tediously by researchers in laboratory environments for very specific tasks... The most impressive results typically only work in a single laboratory, on a single robot, and often involve only a handful of behaviors... [W]hat if we were to pool together the experiences of many robots, so a new robot could learn from all of them at once? We decided to give it a try. In 2023, our labs at Google and the University of California, Berkeley came together with 32 other robotics laboratories in North America, Europe, and Asia to undertake the RT-X project, with the goal of assembling data, resources, and code to make general-purpose robots a reality... The question is whether a deep neural network trained on data from a sufficiently large number of different robots can learn to "drive" all of them — even robots with very different appearances, physical properties, and capabilities. If so, this approach could potentially unlock the power of large datasets for robotic learning. The scale of this project is very large because it has to be. The RT-X dataset currently contains nearly a million robotic trials for 22 types of robots, including many of the most commonly used robotic arms on the market... Surprisingly, we found that our multirobot data could be used with relatively simple machine-learning methods, provided that we follow the recipe of using large neural-network models with large datasets. Leveraging the same kinds of models used in current LLMs like ChatGPT, we were able to train robot-control algorithms that do not require any special features for cross-embodiment. Much like a person can drive a car or ride a bicycle using the same brain, a model trained on the RT-X dataset can simply recognize what kind of robot it's controlling from what it sees in the robot's own camera observations. If the robot's camera sees a UR10 industrial arm, the model sends commands appropriate to a UR10. If the model instead sees a low-cost WidowX hobbyist arm, the model moves it accordingly. "To test the capabilities of our model, five of the laboratories involved in the RT-X collaboration each tested it in a head-to-head comparison against the best control system they had developed independently for their own robot... Remarkably, the single unified model provided improved performance over each laboratory's own best method, succeeding at the tasks about 50 percent more often on average." And they then used a pre-existing vision-language model to successfully add the ability to output robot actions in response to image-based prompts. "The RT-X project shows what is possible when the robot-learning community acts together... and we hope that RT-X will grow into a collaborative effort to develop data standards, reusable models, and new techniques and algorithms." Thanks to long-time Slashdot reader Futurepower(R) for sharing the article.

Read more of this story at Slashdot.

PLUS: It's difficult to make a bad LLM turn good; ELVIS Act seeks AI voice clone ban

AI in brief OpenAI has changed the policies covering use of its models and removed "disallowed usages” of its models including "the generation of malware", "military and warfare" applications, "multi-level marketing", "plagiarism", "astroturfing", and more.…

ScienceAlert reminds us that plants exude "a fine mist of airborne compounds that they use to communicate and protect themselves." And while they've been detected in over 80 plant species, now researchers have applied real-time imaging techniques "to reveal how plants receive and respond to these aerial alarms." Yuri Aratani and Takuya Uemura, molecular biologists at Saitama University in Japan, and colleagues rigged up a pump to transfer compounds emitted by injured and insect-riddled plants onto their undamaged neighbors, and a fluorescence microscope to watch what happened. Caterpillars (Spodoptera litura) were set upon leaves cut from tomato plants and Arabidopsis thaliana, a common weed in the mustard family, and the researchers imaged the responses of a second, intact, insect-free Arabidopsis plant to those danger cues. These plants weren't any ordinary weeds: they had been genetically altered so their cells contained a biosensor that fluoresced green when an influx of calcium ions was detected... [T]he team visualized how plants responded to being bathed in volatile compounds, which plants release within seconds of wounding. It wasn't a natural set-up; the compounds were concentrated in a plastic bottle and pumped onto the recipient plant at a constant rate, but this allowed the researchers to analyze what compounds were in the pungent mix... [T]he undamaged plants received the messages of their injured neighbors loud and clear, responding with bursts of calcium signaling that rippled across their outstretched leaves... [G]uard cells generated calcium signals within a minute or so, after which mesophyll cells picked up the message... "We have finally unveiled the intricate story of when, where, and how plants respond to airborne 'warning messages' from their threatened neighbors," says Masatsugu Toyota, a molecular biologist at Saitama University in Japan and senior author of the study.

Read more of this story at Slashdot.

An opinion piece in the Los Angeles Times imagines a world after "the largest coordinated deepfake attack in history... a steady flow of new deepfakes, mostly manufactured in Russia, North Korea, China and Iran." The breakthrough actually came in early 2026 from a working group of digital journalists from U.S. and international news organizations. Their goal was to find a way to keep deepfakes out of news reports... Journalism organizations formed the FAC Alliance — "Fact Authenticated Content" — based on a simple insight: There was already far too much AI fakery loose in the world to try to enforce a watermarking system for dis- and misinformation. And even the strictest labeling rules would simply be ignored by bad actors. But it would be possible to watermark pieces of content that deepfakes. And so was born the voluntary FACStamp on May 1, 2026... The newest phones, tablets, cameras, recorders and desktop computers all include software that automatically inserts the FACStamp code into every piece of visual or audio content as it's captured, before any AI modification can be applied. This proves that the image, sound or video was not generated by AI. You can also download the FAC app, which does the same for older equipment... [T]o retain the FACStamp, your computer must be connected to the non-profit FAC Verification Center. The center's computers detect if the editing is minor — such as cropping or even cosmetic face-tuning — and the stamp remains. Any larger manipulation, from swapping faces to faking backgrounds, and the FACStamp vanishes. It turned out that plenty of people could use the FACStamp. Internet retailers embraced FACStamps for videos and images of their products. Individuals soon followed, using FACStamps to sell goods online — when potential buyers are judging a used pickup truck or secondhand sofa, it's reassuring to know that the image wasn't spun out or scrubbed up by AI. The article envisions the world of 2028, with the authentication stamp appearing on everything from social media posts to dating app profiles: Even the AI industry supports the use of FACStamps. During training runs on the internet, if an AI program absorbs excessive amounts of AI-generated rather than authentic data, it may undergo "model collapse" and become wildly inaccurate. So the FACStamp helps AI companies train their models solely on reality. A bipartisan group of senators and House members plans to introduce the Right to Reality Act when the next Congress opens in January 2029. It will mandate the use of FACStamps in multiple sectors, including local government, shopping sites and investment and real estate offerings. Counterfeiting a FACStamp would become a criminal offense. Polling indicates widespread public support for the act, and the FAC Alliance has already begun a branding campaign. But all this leaves Slashdot reader Bruce66423 with a question. "Is it really technically possible to achieve such a clear distinction, or would, in practice, AI be able to replicate the necessary authentication?"

Read more of this story at Slashdot.

PLUS: Microsoft taps 700 million new customers in China; AI comes to Korean DMZ; India’s semiconductor sector surges

Asia in Brief Papua New Guinea (PNG) has implemented a two-week state of emergency after failure to reconfigure the nation’s payroll system for government employees sparked riots that resulted in multiple deaths.…

Last April the Python Software Foundation warned that Europe's proposed Cyber Resilience Act jeopardized their organization and "the health of the open-source software community" with overly broad policies that "will unintentionally harm the users they are intended to protect." They'd worried that the Python Software Foundation could incur financial liabilities just for hosting Python and its PyPI package repository due to the proposed law's attempts to penalize cybersecurity lapses all the way upstream. But a new blog post this week cites some improvements: We asked for increased clarity, specifically: "Language that specifically exempts public software repositories that are offered as a public good for the purpose of facilitating collaboration would make things much clearer. We'd also like to see our community, especially the hobbyists, individuals and other under-resourced entities who host packages on free public repositories like PyPI be exempt." The good news is that CRA text changed a lot between the time the open source community — including the PSF — started expressing our concerns and the Act's final text which was cemented on December 1st. That text introduces the idea of an "open source steward." "'open-source software steward' means any legal person, other than a manufacturer, which has the purpose or objective to systematically provide support on a sustained basis for the development of specific products with digital elements qualifying as free and open-source software that are intended for commercial activities, and ensures the viability of those products;" (p. 76) [...] So are we totally done paying attention to European legislation? Ah, while it would be nice for the Python community to be able to cross a few things off our to-do list, that's not quite how it works. Firstly, the concept of an "open source steward" is a brand new idea in European law. So, we will be monitoring the conversation as this new concept is implemented or interacts with other bits of European law to make sure that the understanding continues to reflect the intent and the realities of open source development. Secondly, there are some other pieces of legislation in the works that may also impact the Python ecosystem so we will be watching the Product Liability Directive and keeping up with the discussion around standard-essential patents to make sure that the effects on Python and open source development are intentional (and hopefully benevolent, or at least benign.)

Read more of this story at Slashdot.

Linux Mint 21.3 is now available to download, reports the blog OMG Obuntu. It's the first version to offer Wayland support in its Cinnamon desktop: Following a successful bout of bug-busting in last month's beta release, Mint devs have gone ahead and rubber-stamped a stable release. Thus, you can reasonably expect to not encounter any major issues when installing or using it... [I]t's based on Ubuntu 22.04 LTS and continues to use the Linux 5.15 kernel by default, but newer kernels are available to install within the OS... In my own testing I find Cinnamon's Wayland support to be well-rounded. It's not perfect but I didn't hit any major snafus that prevented me from working (though admittedly I did only attempt 'basic' tasks like web browsing, playing music, and adding applets). However, Cinnamon's Wayland support is in an early state, is not enabled by default, and Linux Mint devs expect it won't be good enough for everyone until the 23.x series (due 2026) at the earliest. Still, try it out yourself and see if it works for you. Select the 'Cinnamon on Wayland (Experimental)' session from the login screen session selector, and then login as normal... Additionally, the latest version of Mozilla Firefox is pre-installed (as a deb, not a Snap) Among the new features are a whole new category of desktop add-ons — "Actions" — which upgrade the right-clicking context menu. (So for .iso files there's two new choices: "Verify" or "Make bootable USB stick".) The article says there's also "a raft of smaller refinements," plus "a bevvy of buffs and embellishments" for Linux Mint's homegrown apps. Any Linux Mint users reading Slashdot? Share your thoughts or experiences in the comments...

Read more of this story at Slashdot.

With subsidiaries like WebMD and CarsDirect, the digital media company "Internet Brands" has over 5,000 employees — and 20 offices in expensive locations like Seattle, San Francisco, Chicago, and New York City. Their solution? Create a cheery corporate video on the company's Vimeo account announcing a new (non-negotiable) hybrid return-to-office policy. SFGate.com calls it "the return-to-office fight's most bizarre corporate messaging yet." Executives from Internet Brands' internet brands are so wide-eyed and declarative, they appear to be at their breaking point in wanting more workers at the office. "Too big of a group hasn't returned," CEO Bob Brisco complains, near the video's opening. The vehicle to deliver that message has it all: rapid jump cuts, odd sound mixing and executives clearly reading their lines from teleprompters. There's plainly faked office b-roll and the obvious use of green screens. There's even some enthusiastic (and awkward) sashaying to the New Orleans classic "Iko Iko" — one wonders if participating employees received compensation. Interestingly, "Iko Iko" is a song about a collision between two rival tribes, which opens with a threat to "set your flag on fire." But subtitles on the video translate the song's Creole patois word "Jockamo" into the corporate-positive phrase "we mean business." It's like the executives started their brainstorming session by watching 12 music videos, an iMovie editing tutorial and the entirety of "The Office" Season 1. Mixed in with the corporate b-roll of a copy machine spitting out paper and a too-loud video of a hand crushing a Dr. Pepper can, the company's executives sketch out the vibe of a return-to-office plan — though no specifics. The video ends with CEO Bob Brisco thanking the team, before gently adding "I want to leave you with this. We aren't asking or negotiating at this point. We're informing, of how we need to work together going forward.... "Thank you, in advance, for your help." The video has since started going viral on Reddit's "Work Reform" subreddit, with a headline calling it a "bizarre and cringe video mocking working from home and threatening employees who continue to avoid the office." (This take drew 1,300 upvotes, and 241 comments, like " 'By the way this is a threat' is a nice way to end it.") Footage of at least some of the executives was clearly just spliced in front of still photos showing what offices look like. But besides the wooden delivery, what really struck me is how generic all the words were: "Working together face-to-face helps us create ideas, faster, and better." "We're able to collaborate, and help each other to be better leaders." "We're better when we're together, and we need to be our best — to crush our competition." [Footage of the word "competition" being erased from a whiteboard. And then, of someone crushing a Dr. Pepper can...]

Read more of this story at Slashdot.

We already know that a precise range of wavelengths within daylight triggers a light-sensitive photoreceptor in the back of your eye, causing the body's internal clock to reset. Those receptors are called "intrinsically photosensitive retinal ganglion cells" (or ipRGCs), according to Science Alert — although the actual color is perceived by some nearby cones (which then send information back to those rceptors). But are our bodies really affected specifically by the perceived color? Chronobiologist Christine Blume investigated with a team from Switzerland's University of Basel and Germany's Max Planck Institute for Biological Cybernetics: Modern scientific wisdom advises us to avoid devices that emit a significant amount of blue radiance, such as our smartphones, computer monitors, and tablets, when we ought to be wrapping ourselves in darkness and resting. There's perfectly sound reasoning for this — the ipRGCs in our eyes react to short wavelengths of electromagnetic radiation, roughly 490 nanometers in size... Given blue light scatters from the sky during daylight hours, it makes sense our eyes would use this wavelength as a cue to mark the beginning and end of sleep time... Yet University of Basel chronobiologist Christine Blume had her suspicions that the way a light's mix of wavelengths influenced the color-reading cones could mean there's more to the phenomenon than meets the eye. "A study in mice in 2019 suggested that yellowish light has a stronger influence on the internal clock than bluish light," says Blume. To resolve whether the way cones perceive a range of wavelengths carries any weight in how the blue-triggered ipRGCs function, Blume and her team recruited eight healthy adult men and eight women in a 23-day-long experiment. After habituating to a specific bedtime for a week, the volunteers attended three visits to a lab where they were exposed to a constant controlled 'white' glow, a bright yellow, or dim blue light for one hour in the evening... None of the analyses revealed any indication that the perceived color of the light affected the duration or quality of the volunteers' sleep patterns. Instead, all three light conditions caused a sleep delay, suggesting light in general has a more complicated impact than previously thought. That's not to say ipRGCs aren't affected by 'blue' wavelengths of light. Rather, white light that is packed with blue waves but stimulates cone cells into seeing yellows, reds, or purples could still affect our sleep cycles. Similarly, light that looks blue but isn't intense enough to provoke the ipRGCs into functioning might have little influence over our body's daily rhythms. Phones of the future may one day allow us to switch into a night mode that we don't perceive in warmer tones. Thanks to long-time Slashdot reader schwit1 for sharing the article.

Read more of this story at Slashdot.

johnnyb (Slashdot reader #4,816) is a senior software R&D engineer who shares his proposed framework for "what AI legislation should cover, what policy goals it should aim to achieve, and what we should be wary of along the way." Some excerpts? Protect Content Consumers from AI The government should legislate technical and visual markers for AI-generated content, and the FTC should ensure that consumers always know whether or not there is a human taking responsibility for the content. This could be done by creating special content markings which communicate to users that content is AI-generated... This will enable Google to do things such as allow users to not include AI content when searching. It will enable users to detect which parts of their content are AI-generated and apply the appropriate level of skepticism. And future AI language models can also use these tags to know not to consume AI-generated content... Ensure Companies are Clear on Who's Taking Responsibility It's fine for a software product to produce a result that the software company views as advisory only, but it has to be clearly marked as such. Additionally, if one company includes the software built by another company, all companies need to be clear as to which outputs are derived from identifiable algorithms and which outputs are the result of AI. If the company supplying the component is not willing to stand behind the AI results that are produced, then that needs to be made clear. Clarify Copyright Rules on Content Used in Models Note that nothing here limits the technological development of Artificial Intelligence... The goal of these proposals is to give clarity to all involved what the expectations and responsibilities of each party are. OpenAI's Sam Altman has also been pondering this, but on a much larger scale. In a (pre-ouster) interview with Bill Gates, Altman pondered what happens at the next level. That is, what happens "If we are right, and this technology goes as far as we think it's going to go, it will impact society, geopolitical balance of power, so many things..." [F]or these, still hypothetical, but future extraordinarily powerful systems — not like GPT- 4, but something with 100,000 or a million times the compute power of that, we have been socialized in the idea of a global regulatory body that looks at those super-powerful systems, because they do have such global impact. One model we talk about is something like the IAEA. For nuclear energy, we decided the same thing. This needs a global agency of some sort, because of the potential for global impact. I think that could make sense... I think if it comes across as asking for a slowdown, that will be really hard. If it instead says, "Do what you want, but any compute cluster above a certain extremely high-power threshold" — and given the cost here, we're talking maybe five in the world, something like that — any cluster like that has to submit to the equivalent of international weapons inspectors. The model there has to be made available for safety audit, pass some tests during training, and before deployment. That feels possible to me. I wasn't that sure before, but I did a big trip around the world this year, and talked to heads of state in many of the countries that would need to participate in this, and there was almost universal support for it.

Read more of this story at Slashdot.

An 1979 Atari 8-bit system re-released in a tiny form factor? Yep. Retro Games Ltd. is releasing a "half-sized" version of its very first home computer, the Atari 400, "emulating the whole 8-bit Atari range, including the 400/800, XL and XE series, and the 5200 home console. ("In 1979 Atari brought the computer age home," remembers a video announcement, saying the new device represents "The iconic computer now reimagined.") More info from ExtremeTech: For those of you unfamiliar with it, the Atari 400 and 800 were launched in 1979 as the company's first attempt at a home computer that just happened to double as an incredible game system. That's because, in addition to a faster variant of the excellent 8-bit MOS 6502 CPU found in the Apple II and Commodore PET, they also included Atari's dedicated ANTIC, GTIA, and POKEY coprocessors for graphics and sound, making the Atari 400 and 800 the first true gaming PCs... If it's as good as the other Retro Games systems, the [new] 400Mini will count as another feather in the cap for Atari Interactive's resurgence following its excellent Atari50 compilation, reissued Atari 2600+ console, and acquisitions of key properties including Digital Eclipse, MobyGames, and AtariAge. The 2024 version — launching in the U.K. March 28th — will boast high-definition HDMI output at 720p 50 or 60Hz, along with five USB ports. More details from Retro Games Ltd. Also included is THECXSTICK — a superb recreation of the classic Atari CX-40 joystick, with an additional seven seamlessly integrated function buttons. Play one of the included 25 classic Atari games, selected from a simple to use carousel, including all-time greats such as Berzerk, Missile Command, Lee, Millipede, Miner 2049er, M.U.L.E. and Star Raiders II, or play the games you own from USB stick. Plus save and resume your game at any time, or rewind by up to 30 seconds to help you finish those punishingly difficult classics! Thanks to long-time Slashdot reader elfstones for sharing the article.

Read more of this story at Slashdot.

"Chinese company Betavolt has announced an atomic energy battery for consumers with a touted 50-year lifespan," reports Tom's Hardware: The Betavolt BV100 will be the first product to launch using the firm's new atomic battery technology, constructed using a nickel -63 isotope and diamond semiconductor material. Betavolt says that its nuclear battery will target aerospace, AI devices, medical, MEMS systems, intelligent sensors, small drones, and robots — and may eventually mean manufacturers can sell smartphones that never need charging... [T]he BV100, which is in the pilot stage ahead of mass production, doesn't offer a lot of power. This 15 x 15 x 5mm battery delivers 100 microwatts at 3 volts. It is mentioned that multiple BV100 batteries can be used together in series or parallel depending on device requirements. Betavolt also asserts that it has plans to launch a 1-watt version of its atomic battery in 2025. The new BV100 is claimed to be a disruptive product on two counts. Firstly, a safe miniature atomic battery with 50 years of maintenance-free stamina is a breakthrough. Secondly, Betavolt claims it is the only company in the world with the technology to dope large-size diamond semiconductor materials, as used by the BV100. It is using its 4th Gen diamond semiconductor material here... [T]he Betavolt BV100 is claimed to be safe for consumers and won't leak radiation even if subjected to gunshots or puncture... Betavolt's battery uses a nickel -63 isotope as the energy source, which decays to a stable isotope of copper. This, plus the diamond semiconductor material, helps the BV100 operate stably in environments ranging from -60 to 120 degrees Celsius, according to the firm... Betavolt will be well aware of devices with a greater thirst for power and teases that it is investigating isotopes such as strontium- 90, promethium- 147, and deuterium to develop atomic energy batteries with higher power levels and even longer service lives — up to 230 years. Thanks to long-time Slashdot reader hackingbear for sharing the news.

Read more of this story at Slashdot.

"Sal Kahn, the founder and CEO of Khan Academy predicted last year that AI tutoring bots would soon revolutionize education," writes long-time Slashdot reader theodp: theodp writes: His vision of tutoring bots tapped into a decades-old Silicon Valley dream: automated teaching platforms that instantly customize lessons for each student. Proponents argue that developing such systems would help close achievement gaps in schools by delivering relevant, individualized instruction to children faster and more efficiently than human teachers ever could. But some education researchers say schools should be wary of the hype around AI-assisted instruction, warning that generative AI tools may turn out to have harmful or "degenerative" effects on student learning. A ChatGPT-powered tutoring bot was tested last spring at the Khan Academy — and Bill Gates is enthusiastic about that bot and AI education in general (as well as the Khan Academy and AI-related school curriculums). From the original submission: Explaining his AI vision in November, Bill Gates wrote, "If a tutoring agent knows that a kid likes [Microsoft] Minecraft and Taylor Swift, it will use Minecraft to teach them about calculating the volume and area of shapes, and Taylor's lyrics to teach them about storytelling and rhyme schemes. The experience will be far richer—with graphics and sound, for example—and more personalized than today's text-based tutors." The New York Times article notes that similar enthusiasm greeted automated teaching tools in the 1960s, but predictions that that the mechanical and electronic "teaching machines' — which were programmed to ask students questions on topics like spelling or math — would revolutionize education didn't pan out. So, is this time different?

Read more of this story at Slashdot.

The fuel-leaking Peregrine lunar lander is now "on a parth towards Earth," according to Update #16 from Astrobotic, which predicts their spacecraft "will likely burn up in the Earth's atmosphere." "Our analysis efforts have been challenging due to the propellant leak... The team is currently assessing options and we will update as soon as we are able. The propellant leak has slowed considerably to a point where it is no longer the teams' top priority... We have now been operating in space for 5 days and 8 hours and are about 242,000 miles from Earth. "A soft landing on the Moon is not possible," the announcement emphasizes. NDTV explains: Shortly after it separated from the rocket, the spaceship experienced an onboard explosion and it soon became clear it would not make a soft lunar touchdown because of the amount of the propellant it was losing — though Astrobotic's team were able to power up science experiments they were carrying for NASA and other space agencies, and gather spaceflight data... Astrobotic itself will get another chance in November with its Griffin lander transporting NASA's VIPER rover to the lunar south pole.

Read more of this story at Slashdot.

jd (Slashdot reader #1,658) shared this story from BleepingComputer. The article notes that "Multiple implementations of the Kyber key encapsulation mechanism for quantum-safe encryption, are vulnerable to a set of flaws collectively referred to as KyberSlash, which could allow the recovery of secret keys." jd explains that Crystals-Kyber "was chosen to be the U.S. government's post-quantum cryptography system of choice last year, but a side-channel attack has been identified. But in the article, NIST says that this is an implementation-specific attack (the reference implementation) and not a vulnerability in Kyber itself." From the article: CRYSTALS-Kyber is the official implementation of the Kyber key encapsulation mechanism (KEM) for quantum-safe algorithm (QSA) and part of the CRYSTALS (Cryptographic Suite for Algebraic Lattices) suite of algorithms. It is designed for general encryption... The KyberSlash flaws are timing-based attacks arising from how Kyber performs certain division operations in the decapsulation process, allowing attackers to analyze the execution time and derive secrets that could compromise the encryption. If a service implementing Kyber allows multiple operation requests towards the same key pair, an attacker can measure timing differences and gradually compute the secret key... In a KyberSlash1 demo on a Raspberry Pi system, the researchers recovered Kyber's secret key from decryption timings in two out of three attempts... On December 30, KyberSlash2 was patched following its discovery and responsible reporting by Prasanna Ravi, a researcher at the Nanyang Technological University in Singapore, and Matthias Kannwischer, who works at the Quantum Safe Migration Center.

Read more of this story at Slashdot.

This week on his podcast Bill Gates asked Sam Altman how his team is doing after his (temporary) ouster, Altman replies "a lot of people have remarked on the fact that the team has never felt more productive or more optimistic or better. So, I guess that's like a silver lining of all of this. In some sense, this was like a real moment of growing up for us, we are very motivated to become better, and sort of to become a company ready for the challenges in front of us." The rest of their conversation was pre-ouster — but gave fascinating glimpses at the possible future of AI — including the prospect of very speedy improvements. Altman suggests it will be easier to understand how a creative work gets "encoded" in an AI than it would be in a human brain. "There has been some very good work on interpretability, and I think there will be more over time... The little bits we do understand have, as you'd expect, been very helpful in improving these things. We're all motivated to really understand them, scientific curiosity aside, but the scale of these is so vast...." BILL GATES: I'm pretty sure, within the next five years, we'll understand it. In terms of both training efficiency and accuracy, that understanding would let us do far better than we're able to do today. SAM ALTMAN: A hundred percent. You see this in a lot of the history of technology where someone makes an empirical discovery. They have no idea what's going on, but it clearly works. Then, as the scientific understanding deepens, they can make it so much better. BILL GATES: Yes, in physics, biology, it's sometimes just messing around, and it's like, whoa — how does this actually come together...? When you look at the next two years, what do you think some of the key milestones will be? SAM ALTMAN: Multimodality will definitely be important. BILL GATES: Which means speech in, speech out? SAM ALTMAN: Speech in, speech out. Images. Eventually video. Clearly, people really want that.... [B]ut maybe the most important areas of progress will be around reasoning ability. Right now, GPT-4 can reason in only extremely limited ways. Also reliability. If you ask GPT-4 most questions 10,000 times, one of those 10,000 is probably pretty good, but it doesn't always know which one, and you'd like to get the best response of 10,000 each time, and so that increase in reliability will be important. Customizability and personalization will also be very important. People want very different things out of GPT-4: different styles, different sets of assumptions. We'll make all that possible, and then also the ability to have it use your own data. The ability to know about you, your email, your calendar, how you like appointments booked, connected to other outside data sources, all of that. Those will be some of the most important areas of improvement. Areas where Altman sees potential are healthcare, education, and especially computer programming. "If you make a programmer three times more effective, it's not just that they can do three times more stuff, it's that they can — at that higher level of abstraction, using more of their brainpower — they can now think of totally different things. It's like, going from punch cards to higher level languages didn't just let us program a little faster — it let us do these qualitatively new things. And we're really seeing that... "I think it's worth always putting it in context of this technology that, at least for the next five or ten years, will be on a very steep improvement curve. These are the stupidest the models will ever be." He predicts the fastest technology revolution "by far," worrying about "the speed with which society is going to have to adapt, and that the labor market will change." But soon he adds that "We started investing a little bit in robotics companies. On the physical hardware side, there's finally, for the first time that I've ever seen, really exciting new platforms being built there." And at some point Altman tells Gates he's optimistic that AI could contribute to helping humans get along with each other.

Read more of this story at Slashdot.

porkchop_d_clown (Slashdot reader #39,923) writes: MOND — MOdified Newtonian Dynamics is a hypothesis that Newton's law of gravity is incorrect under some conditions. Now a paper claims that a study does indeed show that pairs of widely separated binary stars do show a deviation from Newton's Second Law, arguing that, at very low levels, gravity is stronger than the law predicts. Phys.org writes that the study "reinforces the evidence for modified gravity that was previously reported in 2023 from an analysis of the orbital motions of gravitationally bound, widely separated (or long-period) binary stars, known as wide binaries." But RockDoctor (Slashdot reader #15,477) calls the hypothesis "very much disputed." YouTubing-astrophysicist Dr Becky considered this report a couple of months ago (2023-Nov-09), under the title "HUGE blow for alternate theory of gravity MOND". At the very least, astrophysicists and cosmologists are deeply undecided whether this data supports or discourages MOND. (Shortened comment because verification problem.) Last week, I updated my annual count of MOND and other "alternative gravity" publications. While research on MOND (and others) continues, and any "suppression" the tin-foil-hat brigade want to scream about is ineffective, it remains an unpopular (not-equal-to "suppressed") field. Generally, astronomical publication counts are increasing, and MOND sticks with that trend. If anything is becoming more popular, it's the "MOG" type of "MOdified Gravity".

Read more of this story at Slashdot.

Long-time Slashdot reader Geoffrey.landis writes: Hundreds of British postal workers wrongly convicted of theft due to faulty accounting software could have their convictions reversed, according to a story from the BBC. Between 1999 and 2015, the Post Office prosecuted 700 sub-postmasters and sub-postmistresses — an average of one a week — based on information from a computer system called Horizon, after faulty software wrongly made it look like money was missing. Some 283 more cases were brought by other bodies including the Crown Prosecution Service. 2024 began with a four-part dramatization of the scandal airing on British television, and the BBC reporting today that its reporters originally investigating the story confronted "lobbying, misinformation and outright lies." Yet the Guardian notes that to this day in English and Welsh law, computers are still assumed to be "reliable" unless and until proven otherwise. But critics of this approach say this reverses the burden of proof normally applied in criminal cases. Stephen Mason, a barrister and expert on electronic evidence, said: "It says, for the person who's saying 'there's something wrong with this computer', that they have to prove it. Even if it's the person accusing them who has the information...." He and colleagues had been expressing alarm about the presumption as far back as 2009. "My view is that the Post Office would never have got anywhere near as far as it did if this presumption wasn't in place," Mason said... [W]hen post office operators were accused of having stolen money, the hallucinatory evidence of the Horizon system was deemed sufficient proof. Without any evidence to the contrary, the defendants could not force the system to be tested in court and their loss was all but guaranteed. The influence of English common law internationally means that the presumption of reliability is widespread. Mason cites cases from New Zealand, Singapore and the U.S. that upheld the standard and just one notable case where the opposite happened... The rise of AI systems made it even more pressing to reassess the law, said Noah Waisberg, the co-founder and CEO of the legal AI platform Zuva. Thanks to Slashdot reader Bruce66423 for sharing the article.

Read more of this story at Slashdot.

A new analysis suggests that wind turbines have little impact on bird populations, according to the Economist — and that oil-and-gas extraction may be worse: Erik Katovich [an economist at the University of Geneva] combined bird population and species maps with the locations and construction dates of all wind turbines in the United States, with the exceptions of Alaska and Hawaii, between 2000 and 2020. He found that building turbines had no discernible effect on bird populations. That reassuring finding held even when he looked specifically at large birds like hawks, vultures and eagles that many people believe are particularly vulnerable to being struck. But Dr. Katovich did not confine his analysis to wind power alone. He also examined oil-and-gas extraction. Like wind power, this has boomed in America over the past couple of decades, with the rise of shale gas produced by hydraulic fracturing, or fracking, of rocks. Production has risen from 37m cubic metres in 2007 to 740m cubic metres in 2020. Comparing bird populations to the locations of new gas wells revealed an average 15% drop in bird numbers when new wells were drilled, probably due to a combination of noise, air pollution and the disturbance of rivers and ponds that many birds rely upon. When drilling happens in places designated by the National Audubon Society as "important bird areas", bird numbers instead dropped by 25%. Such places are typically migration hubs, feeding grounds or breeding locations. Wind power, in other words, not only produces far less planet-heating carbon dioxide and methane than do fossil fuels. It appears to be significantly less damaging to wildlife, too. Thanks to long-time Slashdot reader SpzToid for sharing the article.

Read more of this story at Slashdot.