2-year-old MacBooks with Apple's T2 security chip are being turned into parts because recyclers have no way to login and factory reset the machines, reports Motherboard. "It's a boon for security and privacy and a plague on the second hard market." From the report: "How many of you out there would like a 2-year-old M1 MacBook? Well, too bad, because your local recycler just took out all the Activation Locked logic boards and ground them into carcinogenic dust," John Bumstead, a MacBook refurbisher and owner of the RDKL INC repair store, said in a recent tweet. First introduced in 2018, the laptop makes it impossible for anyone who isn't the original owner to log into the machine. "Like it has been for years with recyclers and millions of iPhones and iPads, it's pretty much game over with MacBooks now -- there's just nothing to do about it if a device is locked," Bumstead told Motherboard. "Even the jailbreakers/bypassers don't have a solution, and they probably won't because Apple proprietary chips are so relatively formidable." When Apple released its own silicon with the M1, it integrated the features of the T2 into those computers.

"The functionality of T2 is built into Apple silicon, so it's the same situation. But whereas T2 with activation lock is basically impossible to overcome, bypass developers are finding the m1/m2 chips with activation lock even more difficult," Bumstead said. "Many bypassers have claimed solutions to T2 macs (I have not tried or confirmed they work... I am skeptical) but they admit they have had no success with M1. Regardless, a bypassed Mac is a hacked machine, which reverts to the lock if wiped and reset, so it is not ethical to sell bypassed macs in the retail environment."

Responsible recyclers and refurbishers wipe the data from used devices before selling them on. In these cases, the data is wiped, but cannot be assigned to a new user, making them effectively worthless. Instead of finding these machines a second home, Bumstead and others are dismantling them and selling the parts. These computers often end up at recycling centers after corporations go out of business or buy all new machines. [...] Motherboard first reported on this problem in 2020, but Bumstead said it's gotten worse recently. "Now we're seeing quantity come through because companies with internal 3-year product cycles are starting to dump their 2018/2019s, and inevitably a lot of those are locked," he said. "When we come upon a locked machine that was legally acquired, we should be able to log into our Apple account, enter the serial and any given information, then click a button and submit the machine to Apple for unlocking," Bumstead said. "Then Apple could explore its records, query the original owner if it wants, but then at the end of the day if there are no red flags and the original owner does not protest within 30 days, the device should be auto-unlocked."

An anonymous reader quotes a report from VentureBeat: With all the hype and excitement surrounding generative AI technologies, there has been a corresponding growth in the interest that businesses have in using artificial intelligence (AI). While there is lots of interest in figuring out how to use AI to help a business, finding the right people to help an organization use AI effectively isn't necessarily an easy task. It's an area that has led to a surge of interest on freelance marketplace Fiverr, with a 1,400% increase in searches for AI-related services over the last six months. Organizations are looking for individuals that are able to help them take advantage of all manner of AI technologies, including generative AI capabilities for image and text generation that can help to improve marketing, sales and business operations. Fiverr has a history of helping organizations fill talent needs, growing strongly during the pandemic as demand for freelance remote skills accelerated.

Fiverr is now turning its attention to AI, today introducing a series of new categories to its freelance marketplace to help businesses find the talent they need to benefit from the power that AI can bring. "We've seen a trend of increasing searches for AI-related services," Yoav Hornung, head of verticals and innovation at Fiverr, told VentureBeat. "We've also started seeing more freelancers creating offerings that are related to the world of generative AI, for the most recent tools like ChatGPT, GPT-3, Midjourney, Dall-E and Stable Diffusion." Hornung explained that Fiverr creates new categories on its service as a way to help both organizations and freelancers connect. It's an approach that isn't just about providing a specific category, but also about providing the right structure to help a company make a request to bring in the right skills to achieve a business outcome.

For generative AI, there has been growing demand for AI artists that are skilled in the use of the various tools that exist. Hornung said that Fiverr has also seen a surge in companies trying to use or create products that use generative AI engines. To that end, he said companies have been looking for skilled freelancers that can help them build AI-powered applications as well. [...] The explosion in the use of generative AI tools for text generation has also led to a new demand for freelancers to help organizations with proofreading as well as fact checking. "Up until now, proofreading an article was one thing and today, proofreading or editing an article that was generated by AI is different," Hornung said.

Tom's Hardware: Finding the best TIM can be a tricky endeavor, but some people are more adventurous than others. Case in point: An enthusiast recently broadened his GPU thermal paste search to include several interesting substances ranging from regular thermal paste to thermal pads, cheese, ketchup, toothpaste, diaper rash ointment, and even potatoes. The user originally set out to test different types of thermal pads but decided to expand into other substances, making for an interesting and entertaining study in GPU cooling with some substances that are definitely not safe for long-term use. The test system used a Radeon R7 240 with a 30W TDP, with temperature readings from a five-minute run of Furmark. As such, these tests aren't a great indicator of the long-term feasibility of using a potato to cool your chip, so here's a statement of the obvious: Don't try this at home. The user shared a spreadsheet showing the findings, including 22 different tested thermal "paste" materials. The list includes several standard thermal pads of different sizes, including Arctic TP2 0.5mm, 1mm, 1.5mm, Arctic TP3 1mm, 1.5mm, EC360 Blue 0.5mm, EC360 Gold 1mm, 0.5mm EKWB, and Thermal Grizzly Minus 8 thermal pads.

Several items caused the GPU to engage its thermal throttling mechanism due to overheating as the GPU hit its maximum temperature of 105C, including the sliced cheese and potato slices. Some thermal pads also didn't fare well, with throttling occurring with the EC360 Blue 0.5mm thermal pad, 0.5mm EKWB pad, Arctic TP2 1mm pad, Arctic TP2 1.5mm pad, Thermal Grizzly Minus 8 1.5mm pad copper tape. The double-sided aluminum adhesive pad was the worst offender of them all -- it caused the system to shut down. The Pentaten Creme (for diaper rashes) and copper paste were also problematic. However, the rest of the thermal applications were functional and did not cause the GPU to thermal throttle. This includes the 0.5mm Arctic TP2 thermal pad, 1mm Alphacool Apex thermal pad, Arctic TP3 1mm thermal pad, 1mm EC360 Gold thermal pad, and 1.5mm Arctic TP3 thermal pad. All of these thermal pads kept the GPU anywhere between 61C and 79C. The various different kinds of toothpaste did decently well, too, with the Amasan T12 coming out on top at 63C, Silber Wl.paste at 65C, and the plain no-named toothpaste being the worst, hitting 90C. Surprisingly, the Ketchup did exceptionally well, keeping the GPU at 71C.

Despite aggregate productivity for the US economy having doubled over the past 50 years, the country's construction sector has diverged considerably, trending downward throughout that period. And this is no slight decrease. Raw BEA data suggest that the value added per worker in the construction sector was about 40 percent lower in 2020 than in 1970. From a report: How can a sector like construction, with average value-added of 4.3 percent of GDP between 1950 and 2020, experience such a precipitous decline in productivity relative to the rest of the economy? To answer this question, researchers have focused on issues relating to data measurement, hypothesizing that measurement errors largely explain this phenomenon. This new research updates some of those efforts and, importantly, extends them to investigate other hypotheses to find the following:

1. Using measures of physical productivity in housing construction (i.e., number of houses or total square footage built per employee), the authors confirm that productivity is indeed falling or, at best, stagnant over multiple decades. Importantly, these facts are not explained by the incidence of price measurement problems.
2. Instead of data error, the authors investigate two other possible explanations. First, they find that the construction sector's ability to transform intermediate goods into finished products has deteriorated.
3. And second, the authors describe the curious fact that producers located in more-productive areas do not grow at expected rates. Indeed, rather than construction inputs flowing to areas where they are more productive, the activity share of these areas either stagnates or even falls. The authors suggest that this problem with allocative efficiency may accentuate the aggregate productivity problem for the industry.

Following controversial changes to Dungeons & Dragons' decades-old Open Gaming License (OGL), "many prominent third-party RPG publishers now say they're abandoning the OGL, regardless of what changes [publisher Wizards of the Coast (WotC)] officially releases in a coming new version," reports Ars Technica. "What's more, many in the community have now lost faith in WotC's stewardship of the licensed rules system that has underpinned so much of the industry's last two decades." From the report: Pathfinder publisher Paizo Inc. is behind perhaps the biggest effort to move the industry away from WotC's OGL. The company announced last Thursday that it is creating a new Open RPG Creative License (ORC) designed to be "open, perpetual, and irrevocable." [...] Regardless of the legal fate of the OGL, Paizo says it wants to "irrevocably and unquestionably keep alive the spirit of the Open Game License" with its new ORC. The system-agnostic license, designed with the help of IP law firm Azora Law, will eventually be controlled and protected by a nonprofit akin to the Linux Foundation, the company says. Until that new license is ready, upcoming Paizo products will be printed without any explicit license, the company says.

Paizo's ORC effort has already drawn some significant support from the community. Call of Cthulhu and Runequest publisher Chaosium, which never used the WotC OGL for its products in the first place, nonetheless writes that it's "very happy to be working with the rest of the industry to come up with a system-wide OGL that anyone can use." Popular D&D module publisher Kobold Press has also lent its support to Paizo's ORC product but stopped just short of committing to use it for its just-announced Core Fantasy ruleset, codenamed Project Black Flag. Instead, Kobold says it is "wait[ing] to see exactly what shape the Open Gaming License might take in this new era" and "will review the terms and consider whether they fit the needs of our audience and our business goals" when the updated OGL is eventually released. Mutants & Masterminds publisher Green Ronin is also on board with the ORC, with founder and President Chris Pramas publicly comparing the current OGL fiasco to WotC's disastrous attempt to push a new Game System License for the 4th edition of Dungeons & Dragons back in 2008.

Apart from the companies backing Paizo's ORC -- including Legendary Games and Rogue Genius -- some tabletop publishers are creating their own licenses or finding other ways to extricate themselves from the WotC OGL. Blade Runner RPG and Mutant: Year Zero publisher Free League, for instance, says it's overhauling its unique Year Zero Engine to remove any WotC OGL content. At the same time, it's creating a new "irrevocable, worldwide, and royalty-free" license for anyone who wants to use that engine in their own games. [...] Old-School Essentials publisher Necrotic Gnome has similarly announced that it's "moving away from the OGL" for its future products. The company is leaving a bit of wiggle room, saying it will be "keeping an eye on developments" and that its next move "will depend on how the OGL topic develops over the coming months." But Necrotic Gnome adds that "the direction is clear," and that direction is toward "an alternative open license," which could end up being Paizo's ORC. Arcadia publisher MCDM and publisher Basic Fantasy also have plans to abandon the D&D 5th edition ruleset. "Troll Lord Games, meanwhile, publicly abandoned the OGL weeks ago and liquidated its existing stock of 5th-edition D&D products, 'never to be revisited again, in any edition,'" adds Ars.

The Federal Election Commission has dismissed a complaint from Republicans that Google's Gmail app aided Democratic candidates by sending GOP fundraising emails to spam at a far higher rate than Democratic solicitations. From a report: The Republican National Committee and others contended that the alleged benefit amounted to unreported campaign contributions to Democrats. But in a letter to Google last week, the FEC said it "found no reason to believe" that Google made prohibited in-kind corporate contributions, and that any skewed results from its spam filter algorithms were inadvertent. "Google has credibly supported its claim that its spam filter is in place for commercial reasons and thus did not constitute a contribution" within the meaning of federal campaign laws, according to an FEC analysis reviewed by The Wall Street Journal.

The Republican National Committee, the National Republican Senatorial Committee and the National Republican Congressional Committee complained to the FEC last year, citing an academic study that showed that nearly 70% of emails from Republican candidates were sent to spam compared with fewer than 1 in 10 from Democrat candidates from 2019 to 2020. The RNC and other campaign committees argued that Google's "overwhelmingly disproportionate suppression of Republican emails" constituted an illegal corporate contribution to Democratic candidates. But the FEC disagreed, finding that Google established that it maintains its spam filter settings to aid its business in keeping out malware, phishing attacks and scams, and not for the purpose of benefiting any political candidates.

Google is launching its final Stadia game today and is promising to release a tool next week to enable Bluetooth connections on its Stadia Controller. The Verge reports: The last Stadia game to launch on the service is Worm Game, a test game that was technically available on Stadia before Stadia launched publicly in November 2019. Developers at Google have decided to release the game just before the streaming service disappears next week. [...] Alongside the new game, Google is also committing to enabling Bluetooth on Stadia controllers. Google Stadia owners will be pleased to hear there's a self-serve tool coming next week that will enable Bluetooth on the Stadia Controller. "We'll share details next week on how to enable this feature," says a Google Stadia community manager in a forum post.

Google originally launched the Stadia Controller as a device that connects directly to Stadia services and had the Bluetooth chip disabled. After news broke of the Stadia shutdown, fans have been finding ways to save the controller from an e-waste fate by using workarounds to connect it wirelessly to other devices. Workarounds like connecting to an Android device will no longer be required thanks to this new tool. It means that most Stadia players that purchased a Founders or Premiere edition will have been effectively gifted a free Bluetooth controller thanks to Google's refunds.

Big global banks are eying some of the world's most fragile countries for a new experiment in financial engineering: debt relief in exchange for environmental protections. From a report: Called "debt-for-nature swaps," they present a tempting solution for the rising number of nations in distress, particularly those with ecosystems to protect. A country gets to avoid default and lower its debt burden, as long as it's willing to earmark some of the savings to salvage a coral reef, preserve a forest or build a wind farm, for example. Global investors get better returns and enhanced green credentials. Wall Street takes a cut.

As much as $2 trillion of developing country debt may be eligible for this kind of restructuring, according to a rough estimate by the Nature Conservancy, a US nonprofit that's taking a lead role in these deals. Belize inked a $364 million nature swap in 2021; Gabon signaled plans for a $700 million restructuring in October; Ecuador is said to be working on a $800 million transaction, and Sri Lanka is considering a $1 billion deal. Buoyed by the finance industry's newfound enthusiasm for biodiversity, backers of this latest flavor of swap are finding eager partners in investment banks and institutional investors.

These are "turbocharged swaps," said Daniel Munevar, economic affairs officer at the United Nations Conference on Trade and Development and former adviser to finance ministries in Greece and Colombia. "The limit in these operations isn't the money to fund the swaps, it's how much debt can be swapped." Behind the feel-good headlines, it's unclear whether these kinds of swaps will deliver the promised benefits. The terms can be murky. Transaction costs are high. Experts question whether the complex and costly deals will achieve long-term financial stability. In December, as negotiators gathered at the United Nations' COP15 biodiversity conference in Montreal, Greenpeace and dozens of other non-profits called for debt-nature swaps to be rejected.

An anonymous reader quotes a report from TechCrunch: A government watchdog has published a scathing rebuke of the Department of the Interior's cybersecurity posture, finding it was able to crack thousands of employee user accounts because the department's security policies allow easily guessable passwords like 'Password1234'. The report by the Office of the Inspector General for the Department of the Interior, tasked with oversight of the U.S. executive agency that manages the country's federal land, national parks and a budget of billions of dollars, said that the department's reliance on passwords as the sole way of protecting some of its most important systems and employees' user accounts has bucked nearly two decades of the government's own cybersecurity guidance of mandating stronger two-factor authentication. It concludes that poor password policies puts the department at risk of a breach that could lead to a "high probability" of massive disruption to its operations.

The inspector general's office said it launched its investigation after a previous test of the agency's cybersecurity defenses found lax password policies and requirements across the Department of the Interior's dozen-plus agencies and bureaus. The aim this time around was to determine if the department's security defenses were enough to block the use of stolen and recovered passwords. [...] To make their point, the watchdog spent less than $15,000 on building a password-cracking rig -- a setup of a high-performance computer or several chained together -- with the computing power designed to take on complex mathematical tasks, like recovering hashed passwords. Within the first 90 minutes, the watchdog was able to recover nearly 14,000 employee passwords, or about 16% of all department accounts, including passwords like 'Polar_bear65' and 'Nationalparks2014!'. The watchdog also recovered hundreds of accounts belonging to senior government employees and other accounts with elevated security privileges for accessing sensitive data and systems. Another 4,200 hashed passwords were cracked over an additional eight weeks of testing. [...]

The watchdog said it curated its own custom wordlist for cracking the department's passwords from dictionaries in multiple languages, as well as U.S. government terminology, pop culture references, and other publicly available lists of hashed passwords collected from past data breaches. By doing so, the watchdog demonstrated that a well-resourced cybercriminal could have cracked the department's passwords at a similar rate, the report said. The watchdog found that close to 5% of all active user account passwords were based on some variation of the word "password" and that the department did not "timely" wind down inactive or unused user accounts, leaving at least 6,000 user accounts vulnerable to compromise. The report also criticized the Department of the Interior for "not consistently" implementing or enforcing two-factor authentication, where users are required to enter a code from a device that they physically own to prevent attackers from logging in using just a stolen password.

An anonymous reader quotes a report from Motherboard: A team of security researchers managed to gain "super administrative access" into Reviver, the company behind California's new digital license plates which launched last year. That access allowed them to track the physical GPS location of all Reviver customers and change a section of text at the bottom of the license plate designed for personalized messages to whatever they wished, according to a blog post from the researchers. "An actual attacker could remotely update, track, or delete anyone's REVIVER plate," Sam Curry, a bug bounty hunter, wrote in the blog post. Curry wrote that he and a group of friends started finding vulnerabilities across the automotive industry. That included Reviver.

California launched the option to buy digital license plates in October. Reviver is the sole provider of these plates, and says that the plates are legal to drive nationwide, and "legal to purchase in a growing number of states." [...] In the blog post, Curry writes the researchers were interested in Reviver because the license plate's features meant it could be used to track vehicles. After digging around the app and then a Reviver website, the researchers found Reviver assigned different roles to user accounts. Those included "CONSUMER" and "CORPORATE." Eventually, the researchers identified a role called "REVIVER," managed to change their account to it, which in turn granted them access to all sorts of data and capabilities, which included tracking the location of vehicles. "We could take any of the normal API calls (viewing vehicle location, updating vehicle plates, adding new users to accounts) and perform the action using our super administrator account with full authorization," Curry writes. "We could additionally access any dealer (e.g. Mercedes-Benz dealerships will often package REVIVER plates) and update the default image used by the dealer when the newly purchased vehicle still had DEALER tags." Reviver told Motherboard in a statement that it patched the issues identified by the researchers. "We are proud of our team's quick response, which patched our application in under 24 hours and took further measures to prevent this from occurring in the future. Our investigation confirmed that this potential vulnerability has not been misused. Customer information has not been affected, and there is no evidence of ongoing risk related to this report. As part of our commitment to data security and privacy, we also used this opportunity to identify and implement additional safeguards to supplement our existing, significant protections," the statement read.

"Cybersecurity is central to our mission to modernize the driving experience and we will continue to work with industry-leading professionals, tools, and systems to build and monitor our secure platforms for connected vehicles," it added.

"Hey Disney" is the answer to a riddle that nobody asked: What do you get when you cross Amazon's Alexa voice assistant with the voices of Disney characters?

Long-time Slashdot reader destinyland writes: In a few months (and for a few bucks) you'll be able to purchase what Amazon calls a "first-of-its-kind voice assistant" for your Echo devices. Yes, your favorite Disney, Pixar, and Star Wars characters will be available to tell you jokes or play trivia games — whether it's Mickey Mouse, Dory the fish from Finding Nemo, or Olaf the snowman from Frozen.

The PS5 looks to have a design fault that can take months to appear and only seems to happen if you use the console while it's in a vertical orientation. From a report: As Wololo reports, hardware repair specialists working on PS5 consoles that fail to boot are finding the problem is caused by the liquid metal thermal interface Sony used on the custom AMD Zen 2 CPU. When the PS5 is oriented in a vertical position, over time the liquid metal is moving and spilling out on to the components surrounding the CPU. This also means the liquid metal is no longer evenly spread across the chip it's meant to help cool.

An anonymous reader quotes a report from The Verge: The PlayStation 5 shortage is finally over — at least according to Sony. "Everyone who wants a PS5 should have a much easier time finding one at retailers globally, starting from this point forward," said Jim Ryan, the company's gaming boss, during its CES presentation. The company also announced that it's sold around 30 million consoles at this point. That's around 5 million more than the last time it released sales numbers in November.

It's been a long time getting to this point; the PlayStation 5 launched in November 2020, and it wasn't until August 2022 that we declared it "almost easy to buy." For most of the console's life, getting one required you to be either very lucky, relatively on-the-spot with drops, or willing to pay scalpers or Best Buy significantly more than the console's MSRP. Even recently, the main way to get it was via a $550 bundle, which included the PS5 with disk drive, and a game -- that's fine if you were already going to buy Horizon Forbidden West or God of War Ragnarok, but a pretty steep markup if you were hoping for the $399.99 digital console. During the presentation, Ryan thanked PlayStation fans for their patience as the company "managed unprecedented demand amid global challenges over the past two years."

An anonymous reader quotes a report from Ars Technica: Immune therapies have rewritten the game when it comes to cancer treatment, earning the "fifth pillar" label next to more tried and true treatments like radiation therapy, surgery, and chemotherapy. And no immunotherapy has garnered quite the same excitement as CAR T-cell therapy, first approved in 2017 by the Food and Drug Administration to treat a form of acute lymphoblastic leukemia. At the time, then-FDA Commissioner Scott Gottlieb called the approval "a new frontier in medical innovation," and it seemed like the possibilities for CAR T were near-endless. Flash-forward almost six years, and six therapies have been approved for blood cancers, including lymphomas, leukemia, and multiple myeloma. There's no question that when CAR T works, it works incredibly well. But why it doesn't work for the majority of patients or cancer types has befuddled researchers.

CAR T therapies also haven't yet been expanded to treat solid tumors, which make up the majority of cancers. The immune therapy hasn't been able to crack physical barriers, idiosyncratic tumor cells, and a suppressive microenvironment that characterize these cancers. But a new generation of CAR T therapies are emerging, equipped with highly effective small molecules that scientists hope will solve their low success rate for both blood cancers and solid tumors. Known as "armored CAR T," these infusions have been boosted with additional layers of protection and cancer-fighting proteins. Early research shows that the armored flavor of CAR T might have what it takes for immunotherapy to go the extra mile.

Briefly, immunotherapy can boost or restore the body's immune system by lowering cancer cells' defenses, priming the immune system's T cells to destroy tumors, or -- in the case of CAR T-cell therapy -- genetically editing a patient's T cells. Scientists do this by isolating a patient's T cells from their blood and inserting a gene for a chimeric antigen receptor -- a type of synthetic protein that has been specially made to bind to another protein present on the surface of that patient's cancer cells. Then, upon infusing these modified T cells back into a patient, the immune fighters will recognize and destroy the tumor cells when the patient's normal T cells have failed. That, at least, is the idea. But when CAR T doesn't work, a few factors could be at play. One, Lim said, is the tumor microenvironment, the set of chemicals and structures present in solid cancers that naturally suppress pushback from the body's immune system. Tumor heterogeneity is also a factor -- depending on the type and stage of cancer, tumor cells may not express the protein that the CAR T cells' receptors have been designed to recognize, blocking the ability of the CAR T cells to attack the cancer Finally, there are physical barriers on the outside of a solid tumor that can even prevent the T cells from entering inside to destroy the cancer. Armored CAR T is meant to overcome these difficulties. With this form of therapy, not only are T cells engineered to express a tumor cell's surface protein, they are also given potent cargo often in the form of small proteins called cytokines. If deployed on their own, such molecules can be toxic -- but pairing them with T cells designed to release them at the tumor site and nowhere else represents a promising new strategy. Jakub Svodoba, an oncologist at the University of Pennsylvania, "is helping to lead a clinical trial using armored CAR T-cell therapy to treat patients with non-Hodgkin lymphoma for whom previous CAR T therapy has failed," reports the Daily Beast. "Last month, he presented findings that the first seven patients treated with this therapy all responded to it and were alive eight months after receiving it. Svoboda said an important additional finding was that toxicities experienced by the patients -- a concern with cytokines -- were comparable to those from traditional CAR T therapy."

Wendell Lim, a cellular and molecular pharmacology researcher at the University of California, San Francisco, and his team are also working on an armored CAR-T therapy, with the potential for it to be used to treat solid tumors. "In a paper published in Science on Dec. 16, he and his colleagues designed T cells to release a cytokine directly to the tumors of mice with pancreatic cancer and melanoma," reports the Daily Beast. "In the study, they wrote that these cancers are 'nearly completely resistant' to treatment with traditional CAR T, but releasing a cytokine allowed the engineered T cells to get past the tumor microenvironment -- effectively solving one of the issues that has set the therapy back."

"The first open source equivalent of OpenAI's ChatGPT has arrived," writes TechCrunch, "but good luck running it on your laptop — or at all." This week, Philip Wang, the developer responsible for reverse-engineering closed-sourced AI systems including Meta's Make-A-Video, released PaLM + RLHF, a text-generating model that behaves similarly to ChatGPT [listed as a work in progress]. The system combines PaLM, a large language model from Google, and a technique called Reinforcement Learning with Human Feedback — RLHF, for short — to create a system that can accomplish pretty much any task that ChatGPT can, including drafting emails and suggesting computer code.

But PaLM + RLHF isn't pre-trained. That is to say, the system hasn't been trained on the example data from the web necessary for it to actually work. Downloading PaLM + RLHF won't magically install a ChatGPT-like experience — that would require compiling gigabytes of text from which the model can learn and finding hardware beefy enough to handle the training workload.... PaLM + RLHF isn't going to replace ChatGPT today — unless a well-funded venture (or person) goes to the trouble of training and making it available publicly.

In better news, several other efforts to replicate ChatGPT are progressing at a fast clip, including one led by a research group called CarperAI. In partnership with the open AI research organization EleutherAI and startups Scale AI and Hugging Face, CarperAI plans to release the first ready-to-run, ChatGPT-like AI model trained with human feedback. LAION, the nonprofit that supplied the initial dataset used to train Stable Diffusion, is also spearheading a project to replicate ChatGPT using the newest machine learning techniques.

The time has come: Dark Sky, the (mostly) beloved weather app for iOS is going to stop working on January 1st, according to in-app warnings. From a report: The sunsetting has been in the forecast for a while -- Apple announced it was planning on shutting down the service last year after acquiring it in 2020, and it removed Dark Sky from the App Store a few months ago, according to 9to5Mac. But if you've been putting off finding a new weather app, now's the time to finally get around to it. As for what alternatives iPhone users have available (the Android app was axed in 2020), perhaps the most obvious is Apple's own built-in Weather app. The company even has a support document titled "How Dark Sky users can use the Apple Weather app," which talks about how features from the former have been added to the later. Further reading: The World's Best Terrible Weather App.

An anonymous reader quotes a report from Android Headlines: In a recent voice assistant test conducted by popular YouTuber MKBHD, Google Assistant emerged as the best voice assistant, outperforming Apple's Siri, Samsung's Bixby, and Amazon's Alexa. There are several reasons why Google Assistant stands out as the top voice assistant. Firstly, it is backed by Google's powerful artificial intelligence, which helps it to understand and interpret user requests accurately. Secondly, Google Assistant has access to a vast amount of data from its users, which allows it to provide a more personalized experience. The company also collects data from various services such as search, maps, and email to improve the functionality and performance of Google Assistant. However, one of the biggest reasons behind Google Assistant's win is its strong conversation skills. Google's AI uses natural language processing (NLP) algorithms to understand the meaning and context of words and phrases, which helps to keep the conversation going.

Apple's Siri took second place in the competition. It performed well when asked to complete tasks like setting a timer and searching the internet, but struggled when asked to answer more complex or conversational questions. Additionally, Siri was unable to perform tasks that required interacting with apps. In contrast, Samsung's Bixby excelled in device control thanks to its integration with Samsung devices. This integration enables Bixby to control system settings and integrate more deeply with apps than any other voice assistant. Bixby can send text messages, check sports scores, turn down screen brightness, check your calendar, launch apps, and more.

Of all the digital assistants, Amazon's Alexa performed the worst in the voice assistant test. This is due to several factors. Firstly, Alexa is not integrated into smartphones, which means it lacks the personalized touch of other voice assistants. This can make it feel less intuitive and less convenient to use. Secondly, Alexa's inaccuracy in finding facts, inability to interact with other apps and poor conversational models all combine to create a subpar experience when used on a phone. These issues make it difficult for Alexa to provide useful and reliable information, which is a key expectation of voice assistants. In addition, the inclusion of Amazon advertisements between tasks can be annoying and disrupt the user experience.

CNN shares the historic close-up snowflake photos of Wilson Bentley, the first person to capture the details of the individual "snow crystal" ice that makes up snowflakes.

It was 1885, just 69 years after the invention of the camera, and after years of trial and error, "He went on to photograph more than 5,000 of these "ice flowers" during his lifetime — never finding any duplicates — and the images still mesmerize to this day." Every snow crystal shares a common six-sided or six-pointed structure — it's how frozen water molecules arrange themselves — but they will always vary from one another because each falls from the sky in its own unique way and experiences slightly different atmospheric conditions on its travel down to earth. Some of their arms may look long and skinny. Others may appear short and flat or somewhere in between. The possibilities are endless and fascinating....

"He had the mind of a scientist and the soul of a poet, and you can see that in his writings," said Sue Richardson, Bentley's great-grandniece who is vice president of the board for the Jericho Historical Society. "He wrote many, many articles over the years for scientific publications and for other magazines like Harper's Bazaar and National Geographic. "He also kept very detailed weather records and very detailed journals of every photograph that he took of a snow crystal — the temperature, the humidity, what part of the storm it came from. He kept very detailed information, and then these weather records that he kept and the theories that he developed about how snow crystals formed in the atmosphere, those were proven true...."

It wasn't easy, however, to get those snow crystals on camera. It took almost three years, Richardson said, for Bentley to figure out how to successfully photograph one — which he did just a month shy of his 20th birthday. The first obstacle was figuring out how to attach the microscope to the camera. And then there was the challenge of getting each crystal photographed before it could melt away. "He worked in an unheated woodshed at the back of the house. He had to," Richardson said. "And the microscope slides, everything, had to be an ambient temperature or they'd melt" the crystal....

A children's book about him won the Caldecott Medal in 1999.
Bentley never had formal education, according to his grandniece (who grew up hearing stories about this famous ancestor). One says that when Wilson Bentley was given an old microscope at age 15, "The first time he looked at a snow crystal under it, he was hooked. Just the beauty, the intricate detail. He was totally hooked."

An anonymous reader quotes a report from MIT News: What if the U.S. placed a tax on robots? The concept has been publicly discussed by policy analysts, scholars, and Bill Gates (who favors the notion). Because robots can replace jobs, the idea goes, a stiff tax on them would give firms incentive to help retain workers, while also compensating for a dropoff in payroll taxes when robots are used. Thus far, South Korea has reduced incentives for firms to deploy robots; European Union policymakers, on the other hand, considered a robot tax but did not enact it. Now a study by MIT economists scrutinizes the existing evidence and suggests the optimal policy in this situation would indeed include a tax on robots, but only a modest one. The same applies to taxes on foreign trade that would also reduce U.S. jobs, the research finds.

"Our finding suggests that taxes on either robots or imported goods should be pretty small," says Arnaud Costinot, an MIT economist, and co-author of a published paper detailing the findings. "Although robots have an effect on income inequality ... they still lead to optimal taxes that are modest." Specifically, the study finds that a tax on robots should range from 1 percent to 3.7 percent of their value, while trade taxes would be from 0.03 percent to 0.11 percent, given current U.S. income taxes. "We came in to this not knowing what would happen," says Ivan Werning, an MIT economist and the other co-author of the study. "We had all the potential ingredients for this to be a big tax, so that by stopping technology or trade you would have less inequality, but ... for now, we find a tax in the one-digit range, and for trade, even smaller taxes."

[...] Apart from its bottom-line tax numbers, the study contains some additional conclusions about technology and income trends. Perhaps counterintuitively, the research concludes that after many more robots are added to the economy, the impact that each additional robot has on wages may actually decline. At a future point, robot taxes could then be reduced even further. "You could have a situation where we deeply care about redistribution, we have more robots, we have more trade, but taxes are actually going down," Costinot says. If the economy is relatively saturated with robots, he adds, "That marginal robot you are getting in the economy matters less and less for inequality." The paper, "Robots, Trade, and Luddism: A Sufficient Statistic Approach to Optimal Technology Regulation," appears in advance online form in The Review of Economic Studies.

Business Insider reports: An official Amazon study run as part of the company's "Earth's Best Employer" initiative showed mounting employee frustration and other challenges around the time the crucial project was getting off the ground.... The 11-page document, created in October 2021 for Amazon's most senior leaders, gives a brutally honest assessment of employee sentiment.... "We learned we are not seen as distinctively innovative and that our innovation culture is not fun. Innovation at Amazon is associated with stress, burnout, churn, and a cut-throat atmosphere," the document said....

A competitive labor market was one of the main drivers for change, according to the report. Amazon anticipated a "tech talent shortage" of 6 million to 8 million workers in the US by 2030. The Amazon Web Services cloud business alone was projected to need 210,000 to 336,000 tech employees by 2031, or at least 123,000 additional workers. Yet, Amazon is not perceived as a particularly attractive workplace even by its own employees, the report found. Corporate employees said they were subject to long hours and exhausting workloads. Sometimes, even finding answers to trivial HR policies led to confusion. For frontline workers, Amazon wants to provide more behavioral health support and reduce injuries like musculoskeletal disorders, which are more common at its warehouses than other similar facilities

. "Neither corporate nor front-line employees feel Amazon is a place they can clearly grow their careers and achieve personal success," the report stated.
The report identified six areas where Amazon sees room for improvement, including work-life balance and compensation and benefits.