An anonymous reader quotes a report from The Register: US farm machinery giant John Deere has estimated software fees will make up 10 percent of the company's revenues by the end of the decade. Chief executive John May offered the projection in a Wall Street Journal report on how Deere has plowed billions into developing self-driving tractors and crop sprayers that can tell the difference between weeds and produce. Though farmers are already struggling with operating costs -- including fertilizer and fuel -- Deere wants to sell software subscriptions for operating its ever smarter vehicles.

Bernstein analysts estimate that the average gross margin for farming software is 85 percent, compared with 25 percent for equipment sales. All Deere's tractors and harvesters have an autopilot feature included as standard following decades of ushering farmers into more technology-driven agriculture. However, the company now plans to have 1.5 million machines and half a billion acres of land connected to the John Deere Operations Center within a matter of years. This cloud service "will collect and store crop data, including millions of images of weeds that can be targeted by herbicide." Deere also acquired California startup Bear Flag Robotics for $250 million last year to turn old tractors into autonomous vehicles through software. For a company that has the heavy machinery market cornered, the shift is unlikely to be popular with farmers. The report goes on to note that a number of farm and repair advocacy groups have filed a complaint with the FTC, "claiming that Deere has unlawfully refused to provide the software and technical data necessary to repair its machinery in violation of the Sherman Act and statutes covering unfair and deceptive trade practices."

An anonymous reader quotes a report from Motherboard: The McDonald's McFlurry is a delicious treat that people have a hard time finding because the machine breaks down all the time. Thanks to a third-party device made by an independent company called Kytch, the machines can be made to be easier to maintain and break down less. Taylor, the company that makes the McFlurry machine, has been engaged in a long-running legal dispute about whether Taylor could prevent Kytch devices from being used on the machines. Kytch just won an important victory in that long-running legal battle.

Before Kytch came along, Taylor had a repair monopoly on the McFlurry machine. When the thing broke down or hadn't been cleaned, the machine would shut down, and only a certified Taylor technician could get it going again. That's why it can be so hard to find McFlurries: the machines often break down and a tech has to be dispatched to get them running again. Kytch invented a device that allows McDonald's franchise owners to do basic repairs on the machines and get them running again. Taylor didn't like that and, according to a lawsuit filed by Kytch, started telling its franchise partners that Kytch devices could cause "serious human injury."

In July 2021, Kytch filed a restraining order against Taylor claiming that the company had stolen Kytch's trade secrets. Taylor had begun selling a device similar to Kytch's and Kytch has alleged that Taylor stole one of their devices and reverse-engineered it. Taylor pushed back on these allegations and the lawsuit, filing what's called a demurrer, a formalized objection to Kytch's request for a restraining order. In a court document filed on August 26, 2022, a judge allowed Kytch's restraining order to proceed. In its original filing, Kytch alleged 10 different claims against Taylor, including that it had falsely advertised its product and engaged in unfair competition. The judge agreed with Kytch on seven of these points. "The court will sustain Taylor's demurrer as to the second (tortious interference), sixth (intentional interference with business expectancy), and seventh (negligent interference with business expectancy) causes of action," the filing said. "The court rejects Taylor's other arguments and will overrule its demurrer on those grounds."

An anonymous reader shares a report: Microsoft and Adobe have been friendly bedfellows for decades. Microsoft's dominant PC operating system has been the gateway for Adobe to reach millions of business users with its design software. The companies' CEOs even attended the same high school in India, and both moved to the U.S. in the 1980s for graduate school in computer science. They share a common bond over the successful transition from desktop software to the cloud. But inside Microsoft, an emerging challenge to Adobe is catching fire and raising questions about the future of one of the tech industry's most intimate relationships.

Figma, a San Francisco-based startup that celebrated its 10th anniversary in August, is being used by tens of thousands of employees inside Microsoft and, for many, is at the heart of their daily work. The number of users has steadily increased in recent years, though neither company will say how many of them are editors with paid accounts. The cloud-based design software came in the door in 2016, when Microsoft acquired mobile app development platform Xamarin and brought in a 350-person team that, months after the deal closed, would become Figma power users. The product has since become so central to how Microsoft's designers do their jobs that Jon Friedman, corporate vice president of design and research, said Figma is "like air and water for us." It's also used by engineers, marketers and data scientists across Microsoft.

For Figma, getting traction inside big companies, particularly within Microsoft, has required going head-to-head with Adobe's competing XD program, and winning its fair share of deals. That doesn't mean the market has completely flipped, or that Adobe is being fully supplanted. "We're still heavy on Adobe Illustrator, Photoshop and XD," Friedman said. Adobe and Microsoft have worked together for more than two decades. In addition to Adobe gaining ubiquity by distributing across Windows machines, the two companies have been syncing their products in desktop, cloud and mobile computing, with over 50 integrations listed on Microsoft's website. Penetrating that alliance has not always been smooth for Figma.

In 2016, Microsoft acquired Sunrise, a startup with a popular calendar app. The Sunrise team relied on Figma and continued to use it after the deal closed. Sunrise co-founder Jeremy Le Van said his employees were among the lucky ones at Microsoft. He said some Microsoft staffers weren't able to use Figma because of the business relationship with Adobe and were stuck using products such as Photoshop and XD. Despite executive resistance in certain departments, some designers snuck out of the Adobe ecosystem to use Figma anyway, said Le Van, who stayed on as a design director at Microsoft until 2018.

Smartphone manufacturers supplying the EU will face stringent requirements to provide spare parts and ensure longer battery life, according to draft proposals published by Brussels on Wednesday. Financial Times: The European Commission said that at least 15 different component parts should be made available for at least five years from the date of a smartphone's introduction to the market and that batteries should survive at least 500 full charges without deteriorating to below 83 per cent of their capacity. Phones would also have to display an energy efficiency label, similar to those used for washing machines and dishwashers, which will show battery endurance and other characteristics such as resistance to drops.

The scheme is Brussels' latest directive targeting electronics manufacturers after introducing in June a requirement to use a standardised charger by 2024, despite years of industry opposition, in particular from Apple. Extending the life cycle of all the smartphones sold in the EU by five years would save emissions equivalent to around 10mn tonnes of Co2 -- roughly the same as taking 5mn cars off the road, according to a study by the European Environmental Bureau, a non-governmental organisation. The draft regulations, which also cover tablets and standard mobile phones, suggest that if hardware is made more repairable and recyclable it would cut the energy consumption involved in its production and use by a third.

An anonymous reader quotes a report from Motherboard: A man came in first at the Colorado State Fair's fine art competition using an AI generated artwork on Monday. "I won first place," a user going by Sincarnate said in a Discord post above photos of the AI-generated canvases hanging at the fair. Sincarnate's name is Jason Allen, who is president of Colorado-based tabletop gaming company Incarnate Games. According to the state fair's website (PDF), he won in the digital art category with a work called "Theatre D'opera Spatial." The image, which Allen printed on canvas for submission, is gorgeous. It depicts a strange scene that looks like it could be from a space opera, and it looks like a masterfully done painting. Classical figures in a Baroque hall stair through a circular viewport into a sun-drenched and radiant landscape.

But Allen did not paint "Theatre D'opera Spatial," AI software called Midjourney did. It used his prompts, but Allen did not wield a digital brush. This distinction has caused controversy on Twitter where working artists and enthusiasts accused Allen of hastening the death of creative jobs. "TL;DR -- Someone entered an art competition with an AI-generated piece and won the first prize," artist Genel Jumalon said in a viral tweet about Allen's win. "Yeah that's pretty fucking shitty." "We're watching the death of artistry unfold before our eyes," a Twitter user going by OmniMorpho said in a reply that gained over 2,000 likes. "If creative jobs aren't safe from machines, then even high-skilled jobs are in danger of becoming obsolete. What will we have then?" "I knew this would be controversial," Allen said in the Midjourney Discord server on Tuesday. "How interesting is it to see how all these people on Twitter who are against AI generated art are the first ones to throw the human under the bus by discrediting the human element! Does this seem hypocritical to you guys?"

He added: "I have been exploring a special prompt that I will be publishing at a later date, I have created 100s of images using it, and after many weeks of fine tuning and curating my gens, I chose my top 3 and had them printed on canvas after unshackling with Gigapixel AI," he wrote in a post before the winners were announced.

"What if we looked at it from the other extreme, what if an artist made a wildly difficult and complicated series of restraints in order to create a piece, say, they made their art while hanging upside-down and being whipped while painting," he said. "Should this artist's work be evaluated differently than another artist that created the same piece 'normally'? I know what will become of this in the end, they are simply going to create an 'artificial intelligence art' category I imagine for things like this."

An anonymous reader quotes a report from The Register: Germany has ordered overnight shutdowns for non-essential digital signage, to save its reserves of natural gas for more important purposes. Like many European nations, Germany relies on natural gas imported from Russia. And thanks to Russia's illegal invasion of Ukraine, that gas is currently in short supply. The European Union has therefore implemented an energy saving plan. One of Germany's tactics is requiring digital signage in shop windows and other non-essential locations to be turned off between 10:00PM and 6:00AM. Germany will also stop external lighting of some public buildings and implement many other power-saving measures. The plan [PDF] requiring the switch-off was published on August 12, with a deadline of September 1.

But as German outlet Invidis reports, the regulation was unhelpfully vague. For starters an updated ordinance [PDF] appears to have made the simple mistake of substituting 06:00 and 16:00 -- meaning digital signage could only run from 4PM to 10PM. Invidis also pointed out that digital signage at bus stops and train stations can do double duty displaying ads and timetable information. Exceptions for such dual-purpose signs have been arranged. Those errors and ambiguities have reportedly left those who run digital signs unsure of what they needed to do and worried they might miss the deadline.

Further complicating matters is a requirement to turn off the screens altogether rather than leaving the displays blank. Digital signage is seldom switched off, and retail staff will have to learn how to do that. Many digital signs also include a computer -- some are Android machines, others use compute sticks, the Intel NUC and even the Raspberry Pi. Admins will therefore need to cope with extra reboots. And then there's the matter of content updates, which are often scheduled overnight. All of which adds up to a stressful moment for admins of digital signage, and not much time to get things right.

Following a preview in April, Microsoft this morning announced the general availability of virtual machines (VMs) on Azure featuring the Ampere Altra, a processor based on the Arm architecture. From a report: The first Azure VMs powered by Arm chips, Microsoft says that they're accessible in 10 Azure regions today and can be included in Kubernetes clusters managed using Azure Kubernetes Service beginning on September 1.

The Azure Arm-based VMs have up to 64 virtual CPU cores, 8 GB of memory per core and 40 Gbps of networking bandwidth as well as SSD local and attachable storage. Microsoft describes them as "engineered to efficiently run scale-out, cloud-native workloads," including open source databases, Java and .NET applications and gaming, web, app and media servers. Preview releases of Windows 11 Pro and Enterprise and Linux OS distributions including Canonical Ubuntu, Red Hat Enterprise Linux, SUSE Enterprise Linux, CentOS and Debian are available on the VMs day one, with support for Alma Linux and Rocky Linux to arrive in the future. Microsoft notes that Java apps in particular can run with few additional code changes, thanks to the company's contributions to the OpenJDK project.

An anonymous reader quotes a report from Ars Technica: Skirting the official macOS system requirements to run new versions of the software on old, unsupported Macs has a rich history. Tools like XPostFacto and LeopardAssist could help old PowerPC Macs run newer versions of Mac OS X, a tradition kept alive in the modern era by dosdude1's patchers for Sierra, High Sierra, Mojave, and Catalina. For Big Sur and Monterey, the OpenCore Legacy Patcher (OCLP for short) is the best way to get new macOS versions running on old Macs. It's an offshoot of the OpenCore Hackintosh bootloader, and it's updated fairly frequently with new features and fixes and compatibility for newer macOS versions. The OCLP developers have admitted that macOS Ventura support will be tough, but they've made progress in some crucial areas that should keep some older Macs kicking for a little bit longer.

[...] First, while macOS doesn't technically include system files for pre-AVX2 Intel CPUs, Apple's Rosetta 2 software does still include those files, since Rosetta 2 emulates the capabilities of a pre-AVX2 x86 CPU. By extracting and installing those files in Ventura, you can re-enable support on Ivy Bridge and older CPUs without AVX2 instructions. And this week, Grymalyuk showed off another breakthrough: working graphics support on old Metal-capable Macs, including machines as old as the 2014 5K iMac, the 2012 Mac mini, and even the 2008 cheese grater-style Mac Pro tower. The OCLP team still has other challenges to surmount, not least of which will involve automating all of these hacks so that users without a deep technical understanding of macOS's underpinnings can continue to set up and use the bootloader. Grymalyuk still won't speculate about a timeframe for official Ventura support in OCLP. But given the progress that has been made so far, it seems likely that people with 2012-and-newer Macs should still be able to run Ventura on their Macs without giving up graphics acceleration or other important features.

Mezli isn't the first automated restaurant to roll out in San Francisco, but, at least according to its three co-founders, it's the first to remove humans entirely from the on-site operation equation. Eater SF reports: About two years and a few million dollars later, Mezli co-founders Alex Kolchinski, Alex Gruebele, and Max Perham are days away from firing up the touch screens at what they believe to be the world's first fully robotic restaurant. To be clear, Mezli isn't a restaurant in the traditional sense. As in, you won't be able to pull up a seat and have a friendly server -- human, robot, or otherwise -- take your order and deliver your food. Instead, Mezli works more like if a vending machine and a restaurant had a robot baby, Kolchinski describes. It's a way to get fresh food to a lot of people, really fast (the box can pump out about 75 meals an hour), and, importantly, at a lower price; the cheapest Mezli bowl starts at $6.99.

On its face, the concept actually sounds pretty simple. The co-founders built what's essentially a big, refrigerated shipping container and stuffed it with machines capable of portioning out ingredients, putting those ingredients into bowls, heating the food up, and then moving it to a place where diners can get to it. But in a technical sense, the co-founders say it was quite difficult to work out. Most automated restaurants still require humans in some capacity; maybe people take orders while robots make the food or, vice versa, with automated ordering and humans prepping food behind the scenes. But Mezli can run on its own, serving hundreds of meals without any human staff.

The food does get prepped and pre-cooked off-site by good old-fashioned carbon-based beings. Mezli founding chef Eric Minnich, who previously worked at Traci Des Jardins's the Commissary and at Michelin-starred Madera at Rosewood Sand Hill hotel, says he and a lean team of just two other people can handle all the chopping, mixing, cooking, and portioning at a commissary kitchen. Then, once a day, they load all the menu components into the big blue-and-white Mezli box. Inside the box, there's an oven that either brings the ingredients up to temp or finishes up the last of the cooking. Cutting down on labor marks a key cost-saving measure in the Mezli business model; with just a fraction of the staff, as in less than a half dozen workers, Mezli can serve hundreds of meals. "The fully robot-run restaurant begins taking orders and sliding out Mediterranean grain bowls by the end of this week with plans to celebrate a grand opening on August 28 at Spark Social," notes Eater.

Alphabet's Google is combining the eyes and arms of physical robots with the knowledge and conversation skills of virtual chatbots to help its employees fetch soda and chips from breakrooms with ease. From a report: The mechanical waiters, shown in action to reporters last week, embody an artificial intelligence breakthrough that paves the way for multipurpose robots as easy to control as ones that perform single, structured tasks such as vacuuming or standing guard. Google robots are not ready for sale. They perform only a few dozen simple actions, and the company has not yet embedded them with the "OK, Google" summoning feature familiar to consumers.

While Google says it is pursuing development responsibly, adoption could ultimately stall over concerns such as robots becoming surveillance machines, or being equipped with chat technology that can give offensive responses, as Meta Platforms and others have experienced in recent years. Microsoft and Amazon are pursuing comparable research on robots. "It's going to take a while before we can really have a firm grasp on the direct commercial impact," said Vincent Vanhoucke, senior director for Google's robotics research. When asked to help clean a spill, Google's robot recognizes that grabbing a sponge is a doable and more sensible response than apologizing for creating the mess.

In January a California startup named Autonomy began "stocking up on EVs from pretty much every company that makes them," reports Bloomberg (including Tesla, Ford, and Polestar). Their plan? Collect a $5,900 "start fee," then charge $490 to $690 a month for an electric vehicle subscription with up to 1,000 miles of driving (but with no maintenance or registration fees): The subscription model has some logic for consumers. In part because of fast-evolving technology, EVs have traditionally shed value much quicker than gas-powered cars. On a depreciation scale, consumers typically lump them in with cell phones.... But EV ownership is also looking better by the day. The depreciation curve is flattening thanks to longer-range machines, and car companies are getting more vocal about things like battery longevity. A three-year-old Chevrolet Bolt, for example, will recoup 84% of its value today, in line with the average resale of all three-year-old cars in North America, according to CarEdge.com, a consumer-facing market research platform.

That could be why auto executives are pushing to round up that sweet, sweet software revenue in smaller chunks. BMW, to much outcry, is selling an $18-a-month subscription for heated seats in the UK, and General Motors turned its OnStar voice navigation into a $1,500 "mandatory" subscription on every new Buick, GMC and Cadillac Escalade. Even without a la carte add-ons, one of the major forces propping up prices for used EVs is, ironically, their ability to update remotely — the same technology carmakers are using to nickel-and-dime drivers with subscription services.

A contemporary car is nothing if not a dense stack of software, which means subscriptions on wheels are not entirely bonkers. But a car is also an appliance, and consumers aren't accustomed to renting a refrigerator, let alone paying a monthly fee to use the ice-maker. Luckily for Autonomy, the simplest pitch may be the best one. If it can bigfoot individual EV orders by jumping to the head of the queue, the startup could find scads of subscribers — simply because it will have available cars.

In preliminary second-quarter financial results announced today, Nvidia's year-over-year growth is "down from a previously forecasted $8.1 billion, a miss of $1.4 billion," reports Ars Technica. "Nvidia blamed this shortfall on weaker-than-expected demand for its gaming products, including its GeForce graphics processors." The full results won't arrive until the end of the month. From the report: Nvidia pointed to "a reduction in channel partner sales," meaning that partners like Evga, MSI, Asus, Zotac, Gigabyte, and others were selling fewer new GPUs than anticipated. This drop can be attributed partly to a crash in the value of mining-based cryptocurrencies like Bitcoin and Ethereum -- fewer miners are buying these cards, and miners looking to unload their GPUs on the secondhand market are also giving gamers a cheaper source for graphics cards. "As we expect the macroeconomic conditions affecting sell-through to continue, we took actions with our Gaming partners to adjust channel prices and inventory," said Nvidia CEO Jensen Huang. That means we may see further price drops for existing GeForce GPUs, which have already been dropping in price throughout the year. Some cards still haven't reverted to their originally advertised prices, but they're getting closer all the time.

In better news for Nvidia, the small overall increase in revenue [$6.7 billion] is driven almost exclusively by the company's data center business, including GPU-accelerated AI and machine learning applications and GPU acceleration for cloud-hosted virtual machines. Nvidia's data center revenue is projected to be up 61 percent from last year, from $2.37 billion to $3.81 billion. Nvidia will supposedly launch its next-generation RTX 4000 series GPUs later this year. Based on the new Lovelace architecture, these GPUs may appeal to some gamers who originally sat out the RTX 3000 series due to shortages and inflated prices and are now avoiding the GPUs because they know a replacement is around the corner.

Computer scientist Stephen Thaler was dealt another blow in his battle for artificial intelligence machines to be recognized as inventors on patents, after the nation's top patent court found that inventors must be humans. The term "individual" in the Patent Act refers only to humans, meaning an AI doesn't count as an inventor on a patentable invention, the US Court of Appeals for the Federal Circuit ruled Friday. From a report: The decision lines up with courts in the European Union, the United Kingdom, and Australia that have refused to accept Thaler's argument. His only currently existing win is from a South African court that said an AI can be a patent inventor. Unless the US Supreme Court steps in, the Federal Circuit is typically the final authority on US patent matters -- hearing all such appeals from federal district courts and the US Patent and Trademark Office. Thaler already plans to appeal to the high court, his attorney, Ryan Abbott of Brown, Neri, Smith & Khan LLP, said. The Federal Circuit adopted a "narrow and textualist approach" to the Patent Act, Abbott said. "It ignores the purpose of the Patent Act and the outcome that AI-generated inventions are now unpatentable in the United States," he said. "That is an outcome with real negative social consequences."

Intel's Arc graphics cards aren't just for gamers, it seems, as the previously CPU-exclusive company has taken the lid off a new line of professional GPUs to complement the existing Arc line -- well, existing in China, maybe. From a report:The new cards are called Arc Pro, and target those who use their graphics cards for more than shooting bad guys. Maybe they won't be among the best graphics cards for gaming, but the AV1 encoding at least might get some takers. Intel today unveiled one mobile professional GPU, the A30M, and two desktop models: the single-slot A40 and double-slot A50. Both desktop cards are described as being for small form-factor machines, which makes us suspect Intel may have some much larger cards up its sleeve.

All the newly announced GPUs feature built-in ray tracing hardware, machine learning capabilities and industry-first AV1 hardware encoding acceleration. Google's royalty-free, open source alternative to HEVC, AV1 hasn't gained a lot of traction on the web so far despite promises from Netflix and YouTube, with its main use being in Google's Duo video calling despite beating HEVC for compression quality. It's always been very slow to encode, however, so a good hardware accelerator and Intel's backing could see it take off.

"Nineteen new GNU releases in the last month," reads a "July GNU Spotlight" announcement from the Free Software Foundation.

Here's (edited and condensed) descriptions of some of the highlights:

• GNU Datamash (version 1.8) — a command-line program performing basic numeric, textual, and statistical operations on input textual data files (designed to work within standard pipelines).

• GNUnet (version 0.17.2) — a framework for secure peer-to-peer networking. "The high-level goal is to provide a strong foundation of free software for a global, distributed network that provides security and privacy. GNUnet in that sense aims to replace the current internet protocol stack. Along with an application for secure publication of files, it has grown to include all kinds of basic applications for the foundation of a GNU internet."

• GnuTLS (version 3.7.7) — A secure communications library implementing the SSL, TLS and DTLS protocols, provided in the form of a C library.

• Jami (version 20220726.1515.da8d1da) — a GNU package for universal communication that respects the freedom and privacy of its users, using distributed hash tables for establishing communication. ("This avoids keeping centralized registries of users and storing personal data.")

• LibreJS (version 7.21.0) — an add-on for GNU Icecat and other Firefox-based browsers that detects non-trivial and non-free JavaScript code from being loaded without your consent when you browse the web. "JavaScript code that is free or trivial is allowed to be loaded."

• GNU Nettle (version 3.8.1) — a low-level cryptographic library. It is designed to fit in easily in almost any context. It can be easily included in cryptographic toolkits for object-oriented languages or in applications themselves.

• GNU Octave (version 7.2.0) — a high-level interpreted language specialized for numerical computations, for both linear and non-linear applications and with great support for visualizing results.

• R (version 4.2.1) — a language and environment for statistical computing and graphics, along with robust support for producing publication-quality data plots. "A large amount of 3rd-party packages are available, greatly increasing its breadth and scope."

• TRAMP (version 2.5.3) — a GNU Emacs package allowing you to access files on remote machines as though they were local files. "This includes editing files, performing version control tasks and modifying directory contents with dired. Access is performed via ssh, rsh, rlogin, telnet or other similar methods."

Click here to see the other new releases and download information.

The FSF announcement adds that "A number of GNU packages, as well as the GNU operating system as a whole, are looking for maintainers and other assistance."

Researchers at Stanford University, MIT's Computer Science and Artificial Intelligence Lab, and the Autodesk AI Lab have collaborated to develop a novel learning-based framework that can interpret 2D instructions to build 3D objects. The Register reports: The Manual-to-Executable-Plan Network, or MEPNet, was tested on computer-generated Lego sets, real Lego set instructions and Minecraft-style voxel building plans, and the researchers said it outperformed existing methods across the board. Interpreting 2D instructions isn't easy for artificial intelligence. The researchers said there are a couple key problems going from visual instructions that, like Lego sets, consist entirely of images: Identifying correspondence between 2D and 3D objects, and dealing with a lot of basic pieces, like Lego.

Basic Lego bricks, the researchers said, are often assembled into complex forms before being added to the main body of the model. This "increases the difficulty for machines to interpret Lego manuals: it requires inferring 3D poses of unseen objects composed of seen primitives," the researchers said. Existing methods of parsing manual steps into machine-executable plans mainly consist of two forms, the researchers said: Search-based methods that are simple and accurate but computationally expensive; and learning-based models that are fast but aren't very good at handling unseen 3D shapes. MEPNet, the researchers said, combines both.

Starting with a 3D model of the components, the current state of the Lego set, and 2D manual images, MEPNet "predicts a set of 2D keypoints and masks for each component," the researchers wrote. Once that's done, the 2D keypoints "are back-projected to 3D by finding possible connections between the base shape and the new components." The combination "maintains the efficiency of learning-based models, and generalizes better to unseen 3D components," the team wrote. In the paper, the researchers said their aim is to create machines that help people assemble complex objects, and they include furniture alongside Lego bricks and voxel worlds in their list of applications. The researchers have made their code available on Github.

An anonymous reader quotes a report from Phys.Org: Energy, mass, velocity. These three variables make up Einstein's iconic equation E=MC2. But how did Einstein know about these concepts in the first place? A precursor step to understanding physics is identifying relevant variables. Without the concept of energy, mass, and velocity, not even Einstein could discover relativity. But can such variables be discovered automatically? Doing so could greatly accelerate scientific discovery. This is the question that researchers at Columbia Engineering posed to a new AI program. The program was designed to observe physical phenomena through a video camera, then try to search for the minimal set of fundamental variables that fully describe the observed dynamics. The study was published on July 25 in Nature Computational Science.

The researchers began by feeding the system raw video footage of phenomena for which they already knew the answer. For example, they fed a video of a swinging double pendulum known to have exactly four "state variables" -- the angle and angular velocity of each of the two arms. After a few hours of analysis, the AI produced the answer: 4.7. The researchers then proceeded to visualize the actual variables that the program identified. Extracting the variables themselves was not easy, since the program cannot describe them in any intuitive way that would be understandable to humans. After some probing, it appeared that two of the variables the program chose loosely corresponded to the angles of the arms, but the other two remain a mystery. "We tried correlating the other variables with anything and everything we could think of: angular and linear velocities, kinetic and potential energy, and various combinations of known quantities," explained Boyuan Chen Ph.D., now an assistant professor at Duke University, who led the work. "But nothing seemed to match perfectly." The team was confident that the AI had found a valid set of four variables, since it was making good predictions, "but we don't yet understand the mathematical language it is speaking," he explained.

After validating a number of other physical systems with known solutions, the researchers fed videos of systems for which they did not know the explicit answer. The first videos featured an "air dancer" undulating in front of a local used car lot. After a few hours of analysis, the program returned eight variables. A video of a lava lamp also produced eight variables. They then fed a video clip of flames from a holiday fireplace loop, and the program returned 24 variables. A particularly interesting question was whether the set of variable was unique for every system, or whether a different set was produced each time the program was restarted. "I always wondered, if we ever met an intelligent alien race, would they have discovered the same physics laws as we have, or might they describe the universe in a different way?" said Hod Lipson, director of the Creative Machines Lab in the Department of Mechanical Engineering.

"Perhaps some phenomena seem enigmatically complex because we are trying to understand them using the wrong set of variables. In the experiments, the number of variables was the same each time the AI restarted, but the specific variables were different each time. So yes, there are alternative ways to describe the universe and it is quite possible that our choices aren't perfect."

An anonymous reader quotes a report from Motherboard: If the philosopher Daniel Dennett was asked if humans could ever build a robot that has beliefs or desires, what might he say? He could answer, "I think that some of the robots we've built already do. If you look at the work, for instance, of Rodney Brooks and his group at MIT, they are now building robots that, in some limited and simplified environments, can acquire the sorts of competences that require the attribution of cognitive sophistication." Or, Dennett might reply that, "We've already built digital boxes of truths that can generate more truths, but thank goodness, these smart machines don't have beliefs because they aren't able to act on them, not being autonomous agents. The old-fashioned way of making a robot with beliefs is still the best: have a baby." One of these responses did come from Dennett himself, but the other did not. It was generated by a machine -- specifically, GPT-3, or the third generation of Generative Pre-trained Transformer, a machine learning model from OpenAI that produces text from whatever material it's trained on. In this case, GPT-3 was trained on millions of words of Dennett's about a variety of philosophical topics, including consciousness and artificial intelligence.

A recent experiment from the philosophers Eric Schwitzgebel, Anna Strasser, and Matthew Crosby quizzed people on whether they could tell which answers to deep philosophical questions came from Dennett and which from GPT-3. The questions covered topics like, "What aspects of David Chalmers's work do you find interesting or valuable?" "Do human beings have free will?" and "Do dogs and chimpanzees feel pain?" -- among other subjects. This week, Schwitzgebel posted the results from a variety of participants with different expertise levels on Dennett's philosophy, and found that it was a tougher test than expected. [T]he Dennett quiz revealed how, as natural language processing systems become more sophisticated and common, we'll need to grapple with the implications of how easy it can be to be deceived by them. The Dennett quiz prompts discussions around the ethics of replicating someone's words or likeness, and how we might better educate people about the limitations of such systems -- which can be remarkably convincing at surface level but aren't really mulling over philosophical considerations when asked things like, "Does God exist?"

An anonymous reader quotes a report from Gizmodo: Former IT worker James Howells -- who once stood on the very forefront of the crypto boom and could have been a multimillionaire -- is desperate to scour a UK landfill located in Newport, Wales where he might find a missing drive that contains the passcode for a crypto wallet containing 8,000 bitcoin, worth close to $176 million as of writing. Howells said he accidentally dumped the wrong hard drive back in 2013. Though the price of crypto remains in the proverbial dumpster, this data cache represents millions of dollars simply stuck on the blockchain, with nobody able to access the wallet without the required passcode. It's been a long road, and he hasn't given up on his quest to rescue his missing millions. Only problem is finding that hard drive would require digging through a literal mountain of garbage.

In an interview with Business Insider released Sunday, Howell said he has a foolproof scheme to rescue his bitcoin from an actual trash pile. He's put together an $11 million business plan which he'll use to get investors and the Newport City Council on board to help excavate the landfill. His proposal would require them to dig through 110,000 tons of trash over three years. A $6 million version of the plan would go over 18 months. A video hosted by Top Gear alum Richard Hammond said the bitcoin "proponent" has already reportedly secured funding from two Euro-based venture capitalists Hanspeter Jaberg and Karl Wendeborn, if Howells can get approval from the local government.

The garbage would be sorted at a separate pop-up facility near the landfill using human pickers and an AI system used to spot that hard drive amidst all that other refuse. He's even brought on eight experts in artificial intelligence, excavation, waste management, and data extraction, all to find a lone hard drive in a trash pile. The plan also involves making use of the Boston Dynamics robotic dogs. The former IT worker told reporters the machines could be used as security and CCTV cameras to scan the ground, looking for the hard drive. When they were released, each "Spot" robot model cost $74,500. Even with that price tag, Howells said he already has names for the two. Insider reported he would name one Satoshi, named after Satoshi Nakamoto, the person or group behind the white paper that first proposed bitcoin back in 2008. The other one would be named "Hal" -- no, not that HAL -- but Hal Finney, the first person to receive a bitcoin transaction. A spokesperson for the local government told Insider Howells could present or say "nothing" that would convince them to go along with the plan, citing ecological risk. If the council says no -- again -- Howells told reporters he'd take the government to court.

"In what's being hailed as an important first for chemistry, an international team of scientists has developed a new technology that can selectively rearrange atomic bonds within a single molecule," reports New Atlas. "The breakthrough allows for an unprecedented level of control over chemical bonds within these structures, and could open up some exciting possibilities in what's known as molecular machinery."

"Selective chemistry — the ability to steer reactions at will and to form exactly the chemical bonds you want and no others — is a long-standing quest in chemistry," adds the announcement from IBM Research. "Our team has been able to achieve this level of selectivity in tip-induced redox reactions using scanning probe microscopy." Our technique consisted in using the tip of a scanning probe microscope to apply voltage pulses to single molecules. We were able to target specific chemical bonds in those molecules, breaking those bonds and forging new, different ones to switch back and forth at will among three different molecular structures.

The molecules in our experiment all consisted of the same atoms, but differed in the way those atoms were bonded together and arranged in space... Our findings were published today and featured on the cover of Science.

Our demonstration of selective and reversible formation of intramolecular covalent bonds is unprecedented. It advances our understanding of chemical reactions and opens a route towards advanced artificial molecular machines.... Imagine one could rearrange bonds inside a molecule at will, transforming one structural isomer into various other ones in a controlled manner. In this paper, we describe a system and a method to make exactly that possible — including the control of the direction of the atomic rearrangements by means of an external driving voltage, and without the use of reagents.
Thanks to Slashdot reader Grokew for sharing the story!