Category: Technology

Qualcomm has announced its latest flagship mobile chipset, the Snapdragon 8 Gen 2. Along with making it more powerful and efficient than Gen 1 chips, Qualcomm says it has packed more AI smarts into the new platform.

The Snapdragon 8 will tap into the latest Qualcomm AI Engine and upgraded Hexagon processor to offer “faster natural language processing with multi-language translation and advanced AI camera features,” the company claims. The processor has architectural upgrades that will enable up to 4.35 times the AI performance of Gen 1 chips, according to Qualcomm. There will be support for an AI precision format called Int4, which the company suggests will lead to a 60 percent performance/watt improvement over the previous-gen chipset for sustained AI inferencing. Meanwhile, the Sensing Hub will have dual AI processors, which can support features such as custom wake words.

Qualcomm notes that the upgrades will enable new camera-focused features as will. It says Snapdragon 8 Gen 2 will enhance photos and videos automatically and in real-time using semantic segmentation. This harnesses an AI neural network to make the camera aware of and individually optimize elements such as faces, hair, clothes and skies. Qualcomm has tuned the chipset to support new sensors, such as Samsung’s 200-megapixel Isocell HP3. Meanwhile, this is the first Snapdragon model with an AV1 codec that supports 8K HDR playback at up to 60 frames per second.

There’s an intriguing feature for mobile gamers: Qualcomm says there will be real-time hardware-accelerated ray tracing, which should improve the visuals on supported games. The company claims the latest Adreno GPU and Kyro CPU deliver over 25 percent faster performance and 40 percent more power efficiency than the Gen 1 chip. Qualcomm adds that it will offer the first mobile-optimized support for the Unreal Engine 5 Metahumans Framework, claiming that this will let players “experience photorealistic human characters in their games.”

On the connectivity front, the Snapdragon 8 Gen 2 will use AI to manage 5G download speeds, latency, coverage and power efficiency. There will be support for dual 5G SIMs as well as WiFi 7 (with speeds of up to 5.8 Gbps). In terms of audio, phones that use the chipset will be able to offer spatial audio with dynamic head tracking, 48kHz lossless music streaming and latency as low as 48ms.

Qualcomm notes that the first phones with the Gen 2 chipset should be available by the end of the year. Partners including ASUS Republic of Gamers, Honor, Motorola, OnePlus, Oppo, Sharp, Sony, Xiaomi and ZTE plan to use the chip in their devices. There’s likely to be more news from Qualcomm in the coming days, as the company is hosting its Snapdragon Summit this week.

 

In YouTube’s latest move to woo TikTok creators onto YouTube Shorts, the streaming giant announced today that short-term video creators would soon be allowed up to a minute of copyrighted music in their Shorts. The change is a significant boost from the previous 15-second limit for any licensed song.

YouTube says its short-form creators will be able to use between 30 and 60 seconds of licensed music “for most tracks.” However, some songs will remain on the previous 15-second limit, with licensing agreements determining which tracks fall in which window. Creators can quickly see how much time each song allows in the YouTube app’s audio picker. The new song-length options begin rolling out today and will continue “over the next few weeks” on iOS and Android.

As user-created videos have exploded during the past decade, aggressive DMCA takedown notices have become a headache for streamers on all platforms. Record labels often automate their copyright enforcement, leading to overzealous claims. For example, creators have seen their videos flagged for accidentally including a few seconds of copyrighted audio from a passing car’s stereo. Some police officers have even exploited the DMCA to their advantage, blasting Taylor Swift songs to prevent bystanders from sharing their legally recorded videos.

The song-limit boost is YouTube’s latest attempt to woo TikTok creators (and therefore viewers and ad dollars) onto Shorts. In September, the company announced an ad-revenue sharing program to give qualified creators a 45 percent cut of ad revenue, regardless of whether they use music. TikTok launched a similar sharing program earlier this year following widespread complaints about its previous “static pool of money” approach.

YouTube’s aggressive approach appears to be paying off, with Shorts tallying views from over 1.5 billion logged-in users per month. As of September 2021, TikTok had racked up 1 billion monthly users. However, given YouTube’s overall dominance in the video space, those numbers likely include people who were already on YouTube watching other content. TikTok’s numbers, on the other hand, more clearly include people looking for its distinctive short-form videos — an audience that YouTube and other media giants like Meta are continuing to chase.

 

NASA and University of Michigan researchers have found a security flaw that could pose a threat to spacecraft in the right (or rather, wrong) conditions. The team has discovered that time-triggered Ethernet (TTE), a feature that lets critical systems sit alongside minor ones on the same networking hardware, is vulnerable to a spoofing attack. An intruder can send fake sync messages by conducting electromagnetic interference through copper Ethernet cables into network switches, creating a “gap” in a switch’s activity that lets bogus data slide through. Over time, the TTE device will lose sync and behave erratically.

The attack requires placing a small device on the network, so remote breaches are unlikely. However, the consequences could be serious. The scientists tested the vulnerability using real NASA machinery to recreate a planned asteroid redirection test. In a simulation, the TTE exploit produced a knock-on effect severe enough that the crew capsule strayed from its course and missed a crucial docking procedure.

There are simple safeguards, if not necessarily problem-free. Vehicle designers could replace copper Ethernet with fiber optic cables or place optical isolators between switches and devices, so long as they’re willing to accept sacrifices in cost and performance. Engineers could also change the network layout to prevent fake sync messages from taking the same path as legitimate ones, although that will clearly take time.

NASA and the university stress that there’s no “current” known threat. However, the method could also be used to compromise aircraft, power plants and industrial control systems that rely on TTE. Theoretically, a saboteur with physical access could disrupt critical infrastructure in a way that might not be immediately obvious.

 

We’ve finally reached the point where a robot can vacuum for us — but smart vacs aren’t cheap. Eufy, from parent company Anker, makes solid and affordable robot vacuums that often go on sale. In fact, we recommend waiting until a good deal like this comes around before diving in so you can join the robot revolution while saving some cash. Right now Amazon is hosting a sale on Eufy vacuums with discounts of up to 44 percent on the smart vacs. The sale includes the Eufy RoboVac 11S, which is down to $140 from its list price of $230, just be sure to click the box to apply the $60 coupon.   

Buy Eufy RoboVac 11S at Amazon – $140Shop the Eufy robot vacuum sale at Amazon

We tested the 11S in our budget robot vacuum guide and named it the best bang for your buck. We particularly liked the way the slimmer profile was able to get under low-slung furniture and thought its collision avoidance was remarkable. It’s relatively quiet, and did a good job of cleaning the carpets, only missing the occasional crumb. It doesn’t have WiFi, but the included remote takes on most of the scheduling features you’d find in an app. Overall, it impressed us with its smarts, especially for the price, which is now an even better deal. 

Another robot mentioned in our guide is the Eufy RoboVac G30, it’s just $180 right now, which is a steep 44 percent lower than its $319 list price. The G30 is a newer model than the 11S, and features Dynamic Navigation 2.0, improved navigation software that relies less on random patterns to choose its cleaning path. Anker gives the Pascal (Pa) units to rate the suction of Eufy vacuums, and the G30 is rated to 2000pa, which is on the higher end of the spectrum (the 11S above rates at 1300Pa). With WiFi and a dedicated app, you can control it from your phone, or even with Alexa if you have an Echo smart speaker or display. 

Eufy RoboVac G30 at Amazon – $180

The Eufy RoboVac X8 Hybrid got a well-deserved honorable mention in our best robot vacuum guide. It’s a robot vacuum and mop in one, which you can get for $320 right now, instead of its usual $650. The sale price takes 15 percent off, and the clickable coupon discounts another $230. We liked the amount of customization you get with the X8, including a “tap and go” feature that gets the vac to clean an exact spot. You can also straight up control the robot manually, which we thought felt like controlling a slow and slightly clumsy RC car. 

The “hybrid” in the name refers to the mopping features. When you want to wipe down your hard floors, there’s an on-board water tank to fill and included mopping cloths to attach. You’ll also want to mark off the no-go zones wherever there’s carpet. 

The X8 uses LiDAR laser scanning to navigate and unlike some smart vacs, the X8 can learn where not to go through virtual no-go zones you set within the app. We were impressed by how well the X8 navigated, but you should note that if you prefer to set your boundaries using boundary strips, the X8 doesn’t come with any, you’ll have to buy those separately. 

Buy Eufy RoboVac X8 Hybrid at Amazon – $320

Get the latest Black Friday and Cyber Monday offers by following @EngadgetDeals on Twitter and subscribing to the Engadget Deals newsletter.

 

Starting today, Netflix users will be able to see all the devices linked to their account and have the option to remotely log out of each one individually. For each device, you’ll see the type of hardware (say, a smart TV or Android phone), the profile that was most recently used to watch something on that device, the last time Netflix streamed something to it and the approximate location based on the IP address. The new Manage Access and Devices page will be available on the web, iOS and Android.

Netflix suggests this could be helpful for when you’re traveling over the holiday period. Perhaps you might use your account on a hotel’s smart TV, but you forget to log out before leaving. From the Manage Access and Devices page on your account settings, you’ll be able to sign out from that specific device remotely. Until now, Netflix has only offered the option to log out of all devices.

The company has been making moves in recent weeks that will make it easier for you to kick someone out of your account ahead of a crackdown on password sharing. It started rolling out the option to transfer a profile with all data and preferences to a new account. If you’re generous enough, you can give your ex, old roommate or kids a chance to keep their watch history, My List, game save data and so on before you give them the boot.

A key reason why this feature is important is the fact that Netflix will start charging extra fees for users who are accessing the service outside of the account holder’s immediate household. So, you can more easily get rid of moochers who are using your account before Netflix starts charging you more for them.

Moreover, the Manage Access and Devices feature could help you figure out if someone is accessing your account using a leaked password. In any case, if you remotely log a certain device out, it might be a good idea to change your password to ensure only the people you want to access your Netflix account can do so.

 

Hybrid cars aren’t just valuable for their fuel efficiency, apparently. Consumer Reports has published annual reliability survey data indicating that hybrids are generally more reliable than their gas-only equivalents. Hybrid cars were the most reliable among vehicle types, with their SUV siblings ranking third. Certain models were stand-outs, including the Ford Maverick pickup, Lexus NX luxury SUV and Toyota Corolla sedan — they all had above-average reliability on top of major fuel savings.

That trustworthiness doesn’t always extend to other electrified cars. The publication found that plug-in hybrids aren’t as reliable. Toyota’s Prius Prime and RAV4 Prime are less reliable than their conventional hybrid versions, and the Chrysler Pacifica hybrid was one of the most unreliable vehicles in the survey. EVs continue to struggle, too. While there are some exceptions, such as the “outstanding” reliability of the Kia EV6, the category is still plagued with glitches — and not just Tesla’s build quality issues. Ford’s Mustang-Mach-E dipped to below average due to its electronics flaws. Only four out of 11 models with enough survey data had average or better reliability.

A straightforward hybrid isn’t always the best choice, either. Consumer Reports warns that BMW, Mercedes, Ram and others offer “mild” hybrids that don’t offer much in the way of fuel savings, and are sometimes focused more on adding power. These vehicles weren’t included in the hybrid reliability rankings.

The greater reliability of hybrids isn’t a total surprise. While they offer improved fuel economy, they’re ultimately based on familiar model lines using well-established combustion engine technology. EVs are more likely to be brand new models based on young electric motor systems and don’t have years of refinement.

Automakers will have to improve their safety tech if they want to stay in Consumer Reports‘ good graces, whatever powertrain they’re using. As of November, the outlet will penalize models that don’t include pedestrian-aware automatic emergency braking as a standard feature. CR will also stop handing out bonus points to vehicles that only have blind spot warnings (they’ll need rear cross traffic warnings as well) and forward collision alerts. This will theoretically push car creators to strengthen their default safety packages and potentially save lives.

 

Now might be a good moment to splurge on music-making tools. Arturia is running a Black Friday sale that offers 50 percent off its “flagship” virtual instruments, effects and software. The 33-instrument V Collection 9 set has dropped from $599 to $299 for newcomers, while FX Collection 3’s 26 effects plugins are $199 instead of the usual $399. And if you’re looking for an accessible synthesizer, Pigments is down from $199 to $99. The promo runs until December 6th, and you’ll get personalized offers if you already use the company’s products.

We’ve considered these tools to be valuable even at their regular prices. V Collection 9 offers a range of pianos, classic synths and other instruments that offer substantial flexibility, including new items like a Korg MS-20 emulator and the cinematic Augmented Strings. FX Collection 3 brings vintage distortions and lo-fi effects. Pigments, meanwhile, encourages you to create synth sounds from scratch with the help of extras like a distortion module.

You can use the V Collection and Pigments tools by themselves, but all of the software on sale can work with other digital audio workstations through support for formats like AAX, Audio Unit, NKS and VST. With these discounts, Arturia’s audio packages could make more sense if you’re either new to music production or determined to expand your toolset for the lowest price possible.

 

There are more Apple Watches than ever before. Between the Series 8, Watch Ultra and second-gen SE, shopping for a smartwatch is a lot trickier than in years past. Are the Watch Ultra’s outdoors-friendly features worth the $800 and bulkier size? What differentiates the Series 8 from the new SE? Check out the video below detailing how they compare against each other to help you make your decision if you’re in the market for a new Apple Watch.

 

With global temperatures expected to rise as much as 2.8 degrees Celsius by the end of the century, the island nation of Tuvalu says it has no choice but to build a digital version of itself. On Tuesday, Simon Kofe, the country’s foreign minister, told the COP27 climate summit Tuvalu would look to the metaverse to preserve its culture and history amid rising sea levels (via Reuters).

“As our land disappears, we have no choice but to become the world’s first digital nation. Our land, our ocean, our culture are the most precious assets of our people. And to keep them safe from harm, no matter what happens in the physical world, we’ll move them to the cloud,” Kofe said in a video that sees the camera slowly zooming out to reveal that he’s in front of a greenscreen recreation of his home.

At last year’s COP26 summit, Kofe famously addressed the conference standing knee-deep in seawater to highlight the existential threat climate change poses to island nations like Tuvalu. In his latest address, the metaverse is framed as a potential home for all countries if there’s not a global effort to address the problem.

“Only concerted global effort can ensure that Tuvalu does not move permanently online and disappear forever from the physical plane,” he said. “Without a global conscience and a global commitment to our shared well-being, we may soon find the rest of the world joining us online as their lands disappear.

Tuvalu is an archipelago consisting of nine islands located between Australia and Hawaii. It’s home to approximately 12,000 people. Climate scientists anticipate the entire country will be underwater by the end of the 21st century.

To achieve the 1.5C target put forward by the Paris Agreement and avoid significantly worse climate outcomes, the world has eight years to reduce annual global emissions by a further 45 percent, compared with projections based on current policies. To limit the rise in temperatures to under 2C, an extra 30 percent reduction in emissions is needed.

 

Last year, MIT developed an AI/ML algorithm capable of learning and adapting to new information while on the job, not just during its initial training phase. These “liquid” neural networks (in the Bruce Lee sense) literally play 4D chess — their models requiring time-series data to operate — which makes them ideal for use in time-sensitive tasks like pacemaker monitoring, weather forecasting, investment forecasting, or autonomous vehicle navigation. But, the problem is that data throughput has become a bottleneck, and scaling these systems has become prohibitively expensive, computationally speaking.

On Tuesday, MIT researchers announced that they have devised a solution to that restriction, not by widening the data pipeline but by solving a differential equation that has stumped mathematicians since 1907. Specifically, the team solved, “the differential equation behind the interaction of two neurons through synapses… to unlock a new type of fast and efficient artificial intelligence algorithms.”

“The new machine learning models we call ‘CfC’s’ [closed-form Continuous-time] replace the differential equation defining the computation of the neuron with a closed form approximation, preserving the beautiful properties of liquid networks without the need for numerical integration,” MIT professor and CSAIL Director Daniela Rus said in a Tuesday press statement. “CfC models are causal, compact, explainable, and efficient to train and predict. They open the way to trustworthy machine learning for safety-critical applications.”

So, for those of us without a doctorate in Really Hard Math, differential equations are formulas that can describe the state of a system at various discrete points or steps throughout the process. For example, if you have a robot arm moving from point A to B, you can use a differential equation to know where it is in between the two points in space at any given step within the process. However, solving these equations for every step quickly gets computationally expensive as well. MIT’s “closed form” solution end-arounds that issue by functionally modeling the entire description of a system in a single computational step. AS the MIT team explains:

Imagine if you have an end-to-end neural network that receives driving input from a camera mounted on a car. The network is trained to generate outputs, like the car’s steering angle. In 2020, the team solved this by using liquid neural networks with 19 nodes, so 19 neurons plus a small perception module could drive a car. A differential equation describes each node of that system. With the closed-form solution, if you replace it inside this network, it would give you the exact behavior, as it’s a good approximation of the actual dynamics of the system. They can thus solve the problem with an even lower number of neurons, which means it would be faster and less computationally expensive.

By solving this equation at the neuron-level, the team is hopeful that they’ll be able to construct models of the human brain that measure in the millions of neural connections, something not possible today. The team also notes that this CfC model might be able to take the visual training it learned in one environment and apply it to a wholly new situation without additional work, what’s known as out-of-distribution generalization. That’s not something current-gen models can really do and would prove to be a significant step towards the generalized AI systems of tomorrow.