Rivian is working on an e-bike
Would you buy a Rivian e-bike? | Photo by Nilay Patel / The Verge
Electric automaker Rivian is developing an electric bike, according to sources speaking to Bloomberg. CEO RJ Scaringe apparently told staff about the e-bike effort during a company meeting on Friday, noting that a small team is currently working on it.
Bloomberg says it’s unclear if Scaringe was talking about a battery-assisted bicycle (aka an e-bike) or an electric motorcycle. However, Scaringe has teased an entry into micromobility in the past and the company has patents for e-bike designs and components.
Last week, Rivian laid off about six percent of its workforce for the second time in less than a year as part of broad cost-cutting efforts. Rivian has struggled to hit manufacturing targets amidst supply chain problems and a looming EV price war.
Despite the revenue opportunity presented by the new category, car makers have yet to find any notable success with their e-bikes despite brands like Porsche, Mercedes, Jeep, GM, Hummer, and others all giving it a go. Even e-bike darlings like Rad Power and VanMoof have struggled recently after finding success during the covid lockdowns. Would you buy a Rivian e-bike?
‘Get something that’s fun to play, then think about the story’: how Nintendo keeps levelling up
With 100m sales of its Switch console, a new Mario movie and even a California theme park, the Japanese tech giant is riding high again. What’s its secret?
Every Nintendo fan remembers the game that converted them. Perhaps it was running and jumping around as Mario in an abstract, toylike playspace, thrilling at the lightness and precision of his movement. It could have been becoming hypnotised by falling Tetris blocks on the Game Boy’s tiny monochrome screen, or choosing a first Pokémon, marvelling at how the little collection of fat pixels representing your chosen critter instantly assumed an imagined personality. Millions of people had their first Nintendo moment during 2020’s lockdowns, moving to a virtual deserted island full of quirky neighbours in Animal Crossing.
For more than 40 years, this Japanese giant of entertainment has been making video games that have shaped the tastes of the people who played them as children; there is surely no game developer working today who is untouched by its influence. Its latest console, meanwhile – the Nintendo Switch, released in 2017 – recently became the fastest ever to reach 100m sales, and stands a good chance of becoming the bestselling console ever. Barring an extremely unlikely sales slump over Christmas, the Switch will leapfrog to No 3 on the bestselling list when Nintendo announces its quarterly results this week. Only the DS and Sony’s PlayStation 2 are ahead of it.
Hi-Fi Rush review – a brawler set to the beat of a drum
TK; Xbox This music-centred adventure, an unexpected release, pits a teenage boy against robotic assassins in a colourful title for gamers of all talents
The video game industry’s hype cycles are typically measured in months and years, not minutes and seconds. So the simultaneous announcement and release in late January of Hi-Fi Rush – the kind of “go and buy it right now” revelation Apple is known for – feels breezily countercultural. So too does its bright, cartoonish styling: this game is as brazenly colourful as a Jet Set Radio fever dream, and even as plastic Guitar Hero instruments clog up the nation’s cupboards, it’s refreshing to play a game that is so unashamedly music-centred.
This is a brawler set to the beat of a drum. You play as Chai, an ebullient teenage boy who enrols in a biological augmentation program with a shady pharmaceutical company. The operation goes wrong when an old iPod-style music player is fitted to Chai’s chest. Marked for extermination, you must guide Chai to safety, then help him to take down his oppressor. As you move through these Shibuya-inspired streets, the world and everything in it pulses to the rhythm of the game’s soundtrack (a blend of originals and licensed tracks from bands such as the Black Keys and Nine Inch Nails).
Do You Work in the Auto Industry? We Want to Hear From You. Sales of electric vehicles are growing fast, and automakers are investing billions of dollars in new technology and factories. We want to know how jobs are changing.
Monday’s top tech news: an ultra year for tech?
Last year’s iPhone 14 Pro and Pro Max. | Photo by Amelia Holowaty Krales / The Verge
Apple is reportedly thinking about introducing a higher tier iPhone this year
Just a few short days after Samsung announced not one but two “Ultra”-branded devices comes a report that Apple might be about to slap that same label on an upcoming iPhone. Over the weekend, Bloomberg’s Mark Gurman reported that Apple’s 2022 flagship iPhone lineup could include its first Ultra model. Apparently there’s a chance it could offer a higher performance camera system, faster chip, and / or a larger display than the Pro Max.
Second, if you have even a passing interest in Mario Kart, The Last of Us, or Pedro Pascal, then I implore you to check out Pascal’s appearance on Saturday Night Live. I won’t spoil it, but it’s-a good stuff.
Finally, I want to highlight a couple of interesting reviews that went up over the weekend. Announced last week alongside the Samsung Galaxy S23 lineup, the SmartThings Station is not only a hub to control your smart home, but it’s also a controller button and a wireless charging pad — both functions that provide a great excuse to have it easily accessible at all times. There’s also this fantastically overkill gadget from Nix which aims to help athletes keep hydrated while exercising.
At This School, Computer Science Class Now Includes Critiquing Chatbots Move over, coding. Some schools are asking student programmers to think critically about rapid advances in artificial intelligence.
Blueprints for a dream: the new age of virtual architecture
Aspiring architects are using using social media, AI and digital technology to showcase their fantastical creations to the full
“Something big is happening,” says Hamza Shaikh. “Architecture is entering a new age.” The ways in which buildings are imagined and communicated are, he argues, being transformed by a combination of social media and the ever-evolving techniques of digital drawing, to which artificial intelligence is adding new capabilities. And indeed, if it is not yet clear how blocks of flats or schools or shopping centres near you might be changed by this revolution, the energy and invention behind it are undeniable.
There is also, as Shaikh justifiably claims, a social transformation. If, in the past, aspiring architects had to claw their way up a profession that favoured those with connections and money, now anyone from anywhere can make a name for themselves, if they have the talent, determination and access to technology. They do this not by realising completed buildings, but through compelling images of imaginary architecture. They don’t all use the most advanced techniques all the time – some work by hand, some (Shaikh included) with hybrids of manual and digital – but all use the internet to spread their work and exchange ideas.
That pain has now been compounded by the recent discovery that Archie was bullied online in the months before the catastrophic brain injury on 7 April last year, she told the Guardian.
‘They filmed me without my consent’: the ugly side of #kindness videos
Some social media users are building a following through ‘feelgood’ videos, in which, for instance, they give flowers to a stranger. The stranger then becomes their clickbait. Is there anything we can do to stop this?
Maree only wanted to buy some shoes. A pair that she liked the look of had gone on sale, so she made a trip into the city to try them on. It was late in the day in June, mid-winter in Melbourne, and the shopping centre was quiet. After making her purchase, Maree stopped for a coffee. “And that’s when it happened,” she says.
A young man approached her holding a posy of flowers. He asked Maree to hold them for him as he put on his jacket. “I wish I’d trusted my instincts and said no,” she says. “It was all so quick.” Maree took the flowers – then the man walked away, wishing her “a lovely day”. She held them out after him, bemused.
TikTok: how the west has turned on gen Z’s favourite app
US and European fears about China exploiting TikTok’s data harvest and promoting Beijing’s worldview look set to inspire an urgent overhaul in data privacy laws
The FBI has called it a national security threat. The US government has passed a law forcing officials to delete it from their phones. Texas senator Ted Cruz has denounced it as “a Trojan horse the Chinese Communist party can use to influence what Americans see, hear, and ultimately think”. And in March its CEO will defend its existence before the US Congress. For those unaware of the debate broiling on the other side of the Atlantic, the target of this strong rhetoric might prove surprising: an app best known for viral dances, launching generation Z media stars, and sucking teens down an hours-long content abyss.
But the rancorous debate over TikTok that began under the Trump administration has rolled on under President Biden. In addition to a ban of the app on all federal government devices, at least 27 states have blocked TikTok on devices they’ve issued, affecting a number of state schools and universities, too. A bipartisan bill, introduced in Congress last December, stipulates banning the app’s use by everyone in the United States.
Nothing, Forever is an endless, AI-generated version of the show that has been streaming on Twitch since mid-December. It tells the “story” – if you can call it that – of four characters, Larry, Fred, Yvonne and Kakler, who look like what would happen if Jerry, George, Elaine and Kramer were sucked into a 1990s computer game. They spend their days discussing their lives and other trivial matters. And it never, ever stops: log on at any hour and there they are, talking about coffee quality or a difficult Monopoly game.
On Friday, the Treasury Department updated the way it classifies vehicles that qualify for its $7,500 EV tax credit as part of the Inflation Reduction Act (IRA). The change should allow more vehicles — including the Model Y — to qualify for the credit, as it no longer puts certain SUV crossovers in the same category as sedans.
But now that the government uses the Environmental Protection Agency’s (EPA) Fuel Economy Labeling standard instead of the EPA’s corporate average fuel economy (CAFE) standard to classify vehicles, the Model Y and Cadillac Lyriq now fall under the SUV category. This gives Tesla more wiggle room when it comes to pricing, as vehicles in this category can be priced at up to $80,000 to qualify for the tax credit.
It’s still unclear how the Treasury Department’s list of qualified vehicles will change come March, though. That’s when the agency’s expected to release its guidance on how to apply the IRA’s strict rules surrounding the sourcing and manufacturing of the minerals and battery components used in EVs.
‘I’m not Snow White. I have to think like a criminal’: how I became a burglar for hire
Jenny Radcliffe is a professional ‘people hacker’ – someone who claims she can get past anyone and get in anywhere. No building is secure. How does she do it? Plus, an extract from her memoir
‘Do I look like someone to mess with?” says Jenny Radcliffe, folding her arms in a really-don’t-mess-with-me kind of way. Her tattoos seem to be making the point, too. On her left forearm is a Latin phrase – facta non verba, actions not words – with a pair of devil’s horns; on her right, a feather, from the wings of an angel. Which is she, I wonder. Her boots – DM-like, many eyelets – suggest no angel; but the T-shirt is emblazoned with “Trust Me”.
Radcliffe has an unusual job: she’s a social engineer. “Also known as a professional burglar, physical penetration tester … though it’s difficult to say that one to old ladies on trains,” she says. Yes, I can see that.
Artists must be protected from piracy in the new world of AI | Letter
We must recognise the critical importance of strong copyright law and fair remuneration, writes Christian Zimmermann
Artists, illustrators and photographers have often led the way in embracing new technology. The concerns that creators such as Harry Woodgate have about AI programs (‘It’s the opposite of art’: why illustrators are furious about AI, 23 January) that “rely entirely on the pirated intellectual property of countless working artists, photographers, illustrators and other rights holders” must be heeded.
Evidence published recently by the House of Lords, gathered from the first-hand experience of visual artists, galleries and experts, demonstrates that the government’s proposed copyright exception will have far-reaching, detrimental consequences.
Are You Switching to a Heat Pump? We Want to Hear From You. Some people in extremely cold areas of the U.S. are switching to heat pumps from oil or gas furnaces. But others are not convinced that these devices are right for them.
Advisory Firm Sues Elon Musk’s Twitter, Saying It Hasn’t Been Paid Innisfree M&A says Twitter, which Elon Musk bought last year, has not paid it $1.9 million for services it rendered for the deal.
Contractors who work on YouTube Music are striking
Illustration by Alex Castro / The Verge
Over 40 contractors for YouTube Music are going on strike — a first at Google, according to the Alphabet Workers Union (or AWU). The action is in response to an order to return to in-person work next week, something that many of the workers say they can’t do. They’re demanding a return-to-work policy that’s “fair, flexible, and does not threaten the safety and livelihoods of workers,” according to an AWU press release.
The workers are part of the YouTube Music Content Operations team via Cognizant, a subcontractor for Alphabet, Google and YouTube’s parent company. Their jobs are to “ensure music content is available and approved” for the platform, according to a prior press release from the AWU.
The objections to the return to office plan stem from pay and availability. According to the AWU, the contractors are paid as little as $19 an hour, making it difficult to afford the relocation, travel, or childcare costs that they didn’t have to pay when working remotely, instead of at an office in Austin, TX.
An unnamed spokesperson for Cognizant told Engadget that the return to office policy had been “communicated to [the workers] repeatedly since December 2021,” and that they had taken the positions “with the understanding that they were accepting in-office positions, and that the team would work together at a physical location based in Austin.” Google would not provide an on-the-record comment for this story, but the company has told the National Labor Relations board that it does not see the workers as its employees, according to Bloomberg.
The contractors are currently attempting to unionize with the AWU, which filed a petition with the National Labor Relations Board to represent the contractors in October. Last week, the AWU filed an unfair labor practice charge against Alphabet and Cognizant, claiming that the return to office was being used to “interfere with the fair voting conditions mandated by federal law,” as Sam Regan, one of the workers put it in a press release.
There have been previous organized labor actions at Google. In 2018, tens of thousands of workers walked out to protest how Google handled sexual harassment, spurred on by reports that it had paid Android co-founder Andy Rubin $90 million in severance after he was accused of sexual assault. And in 2022, a group of Cognizant contractors working on Google Maps were able to get their return-to-office pushed back after threatening to strike.
Electrify America is increasing prices at its DC fast charging stations
Photo by Umar Shakir / The Verge
Remember how rising gas prices weren’t a concern to electric car owners? Electrify America is giving users of its DC fast charging network a reminder about one of the biggest frustrations of owning internal combustion engine cars again — by increasing electricity rates at its stations (via Ars Technica).
Starting on March 6th, Electrify America will be charging US customers a standard rate of $0.48 per kilowatt hour. The company, a subsidiary of Volkswagen Auto Group, was charging $0.43 per kWh, but now, it is emailing customers to share the price increase details.
Tara Geiger of Electrify America’s corporate communications emailed the company’s statement on the price changes to The Verge:
Electrify America has been able to maintain consistent and uniform pricing since September 2020; however, rising operational and energy costs have now made adjusting our pricing necessary. Our focus remains on meeting the needs of electric vehicle drivers of today and tomorrow by investing in our network expansion and enhancement to customer experience.
Customers paying for Electrify America’s $4 per month Pass Plus subscription will also be affected. The company had advertised that Pass Plus saves “about 25 percent on charging” with the monthly fee plus a reduced $0.31 per kilowatt hour rate. Now that rate is increasing to $0.36 per kilowatt hour. Geiger adds that Pass Plus subscribers “can still benefit from savings of about 25% on charging,” confirming that the announced price increases apply to Electrify America only and that they are not temporary.
Electrify America also is increasing per-minute charging rates in applicable states like Massachusetts and Texas, where companies aren’t allowed to directly bill customers by the kWh. Users in those states will now be paying $0.19 per minute for sessions with peak charging speeds up to 90kW, up from $0.16. For quicker charging cars, up to 350kW, the rate is increasing from $0.32 per minute to $0.36.
And Pass Plus subscribers in those states will see rates increase from $0.12 per minute to $0.15 for sessions peaking up to 90kW and $0.24 per minute to $0.29 for quicker charging cars.
Comparatively, Tesla owners using the Supercharger network pay variable rates that can range from around $0.25 per kWh to as much as $0.50 depending on the state and time of use. Tesla has been increasing Supercharger prices for years but has also recently decreased prices in some areas.
Personally, I’ve experienced an average rate of $0.38 per kWh at Superchargers while driving in the Mid-Atlantic region this past December. In one session, I’ve added 46kWh of electricity to my battery, which cost me about $17. If the rates matched that of Electrify America, it would have cost me about $20 at the old standard rate and $22 at the new one.
Electrify America, after Tesla, is the largest coast-to-coast fast charging network in the US. In addition to its 800-plus stations in North America, the company is working with Ikea and TravelCenters of America to increase its reach to EV drivers. There are even cool Electrify America Sky Club Lounge-like locations in the works, which might just make the whole price increase worth it — whenever it gets built.
Google is holding an event about search and AI on February 8th
Illustration: The Verge
Google is about to share more about its work in artificial intelligence. Next week, Google will be holding an event about how it’s “using the power of AI to reimagine how people search for, explore and interact with information, making it more natural and intuitive than ever before to find what you need,” according to an invite sent to The Verge. The 40-minute event will be streamed on YouTube on February 8th at 8:30AM ET.
The timing of the event is interesting given that Google CEO Sundar Pichai just announced that the company is planning on letting people “interact directly” with its “newest, most powerful language models as a companion to search” soon. Google, long the de facto way to find information on the internet, is likely facing some pressure from Microsoft, which is reportedly planning on integrating ChatGPT into Bing. That could allow Bing to offer more intelligent summaries and results, which is something that Google has struggled with.
Image: Google
An image from Google’s invite.
Of course, it’s very possible the presentation will be more about tools we’re already familiar with; the invitation contains references to Google Lens, Translate, Shopping, and Maps. It’d be hard to complain if the show is just about small improvements to those tools, as those can make a big difference for some of the millions of people who use them a day.
TikTok’s transparency theater
One of the displays inside TikTok’s Transparency and Accountability Center in Los Angeles. | Photo by Allison Zaucha for The Verge
What I observed during a recent visit to TikTok’s first transparency center.
TikTok is staring down the barrel of an outright ban in the US. It has already been prohibited on federal employee devices, blocked by dozens of universities across the country, and lawmakers are calling for its removal from US app stores.
It’s with that context that I and a handful of other journalists were invited to the company’s Los Angeles headquarters earlier this week for the first media tour of its “Transparency and Accountability Center.” It’s a space that, like the political discussion about TikTok these days, seems more about virtue signaling than anything else. Company officials say the center is designed for regulators, academics, and auditors to learn more about how the app works and its security practices. We were told that a politician-who-would-not-be-named had toured it the day before. TikTok eventually plans to open more centers in Washington, DC, Dublin, and Singapore.
Our tour was part of a multi-week press blitz by TikTok to push Project Texas, a novel proposal to the US government that would partition off American user data in lieu of a complete ban. The CEO of TikTok, Shou Zi Chew, was in DC last week giving a similar pitch to policymakers and think tanks. In March, he is expected to testify before Congress for the first time.
Photo by Allison Zaucha for The Verge
What you see when you first enter TikTok’s transparency center.
TikTok isn’t the first embattled tech company to lean on the spectacle of a physical space during a PR crisis. In 2018, Facebook invited journalists to tour its election “War Room,” which was really just a glorified conference room packed with employees staring at social media feeds and dashboards. Photos were taken, stories were written, and then the War Room was closed about a month later.
In a similar way, TikTok’s transparency center is a lot of smoke and mirrors designed to give the impression that it reallycares. Large touchscreens explain how TikTok works at a high level, along with a broad overview of the kind of trust and safety efforts that have become table stakes for any large platform.
A key difference, however, is a room my tour group wasn’t allowed to enter. Behind a wall with Death Star-like mood lighting, TikTok officials said a server room houses the app’s source code for outside auditors to review. Anyone who enters is required to sign a non-disclosure agreement, go through metal detectors, and lock away their phone in a storage locker. (It wasn’t clear who exactly would be permitted to enter the room.)
Photo by Allison Zaucha for The Verge
A room where you can interact with a mock version of the moderation software TikTok uses.
The interactive part of the center I was allowed to experience included a room with iMacs running a mock version of the software TikTok says its moderators use to review content. There was another room with iMacs running “code simulators.” While that sounded intriguing, it was really just a basic explanation of TikTok’s algorithm that seemed designed for a typical member of Congress to grasp. Close-up photos of the computer screens weren’t allowed. And despite it being called a transparency center, TikTok’s PR department made everyone agree to not quote or directly attribute comments made by employees leading the tour.
At the moderator workstation, I was shown some potentially violating videos to review, along with basic information like the accounts that posted them and each video’s number of likes and reshares. When I pulled up one of a man talking into the camera with the caption of “the world bringing up 9/11 to justify Muslims as t3rrori$ts,” the moderator system asked me to select whether it violated one of three policies, including one on “threats and incitement to violence.”
At the code simulator iMac in the other room, I was hoping to learn more about how TikTok’s recommendations system actually works. This was, after all, a physical place you had to travel to. Surely there would be some kind of information I couldn’t find anywhere else?
What I got was this: TikTok starts by using a “coarse machine learning model” to select “a subset of a few thousand videos” from the billions hosted by the app. Then, a “medium machine learning model further narrows the recall pool to a smaller pool of videos” it thinks you’ll be interested in. Lastly, a “fine machine learning model” makes the final pass before serving up videos it thinks you will like in your For You page.
The information displayed was frustratingly vague. One slide read that TikTok “recommends content by ranking videos based on a combination of factors, including the interests that new users convey to TikTok the first time they interact with the app, as well as changing preferences over time.” That’s exactly how you would expect it to work.
Photo by Allison Zaucha for The Verge
Eric Han, head of USDS Trust and Safety at TikTok.
TikTok first tried to open this transparency center in 2020, when then-President Donald Trump was trying to ban the app and Kevin Mayer was its CEO for all of three months. But then the pandemic happened, delaying the center’s opening until now.
In the past three years, TikTok’s trust deficit in DC has only deepened, fueled by a growing anti-China sentiment that started on the right and has since become more bipartisan. The worst revelation was in late December, when the company confirmed that employees improperly accessed the location data of several US journalists as part of an internal leak investigation. That same month, FBI director Chris Wray warned that China could use TikTok to “manipulate content, and if they want to, to use it for influence operations.”
TikTok’s answer to these concerns is Project Texas, a highly technical, unprecedented plan that would wall off most of TikTok’s US operations from its Chinese parent company, ByteDance. To make Project Texas a reality, TikTok is relying on Oracle, whose billionaire founder Larry Ellison leveraged his connections as an influential Republican donor to personally secure Trump’s blessing in the early phase of negotiations. (No one from Oracle was present at the briefing I attended, and my request to speak with someone there for this story wasn’t answered.)
Photo by Allison Zaucha for The Verge
I was given a brief overview of Project Texas before the tour, though I was asked to not quote the employees who presented directly. One graphic I was shown featured a Supreme Court-like building with five pillars showing the issues Project Texas is meant to address: org design, data protection and access control, tech assurance, content assurance, and compliance and monitoring.
TikTok says it has already taken thousands of people and over $1.5 billion to create Project Texas. The effort involves TikTok creating a separate legal entity dubbed USDS with an independent board from ByteDance that reports directly to the US government. More than seven outside auditors, including Oracle, will review all data that flows in and out of the US version of TikTok. Only American user data will be available to train the algorithm in the US, and TikTok says there will be strict compliance requirements for any internal access to US data. If the proposal is approved by the government, it will cost TikTok an estimated $700 million to $1 billion per year to maintain.
Whether Project Texas satisfies the government or not, it certainly seems like it will make working at TikTok more difficult. The US version of TikTok will have to be fully deconstructed, rebuilt, and published by Oracle to US app stores. Oracle will also have to review every app update. Duplicate roles will be created for TikTok in the US, even if the same roles already exist for TikTok elsewhere. And app performance could suffer when Americans are interacting with users and content in other countries since American user data has to be managed inside the country.
Photo by Allison Zaucha for The Verge
One name that wasn’t uttered during the entire briefing: ByteDance. I got the impression that TikTok employees felt uncomfortable talking about their relationship with their parent company.
While ByteDance was directly unacknowledged, its ties to TikTok weren’t hidden, either. The Wi-Fi for the building I was in was named ByteDance and conference room screens in the transparency center displayed Lark, the in-house communications tool ByteDance developed for its employees around the world. At one point during the tour, I tried asking what would hypothetically happen if, once Project Texas is greenlit, a Bytedance employee in China makes an uncomfortable request to an employee in TikTok’s US entity. I was quickly told by a member of TikTok’s PR team that the question wasn’t appropriate for the tour.
Ultimately, I was left with the feeling that, like its powerful algorithm, TikTok built its transparency center to show people what it thinks they want to see. The company seems to have realized that it won’t save itself from a US ban on the technical merits of its Project Texas proposal. The debate is now purely a matter of politics and optics. Unlike the tour I went on, that’s something TikTok can’t control.
It sounds like Google’s getting ready to compete with ChatGPT
Illustration: The Verge
Google may be gearing up to compete with OpenAI’s ChatGPT by letting people “interact directly” with its “newest, most powerful language models as a companion to search,” according to CEO Sundar Pichai. It would be a big move for the company — as systems like ChatGPT and DALL-E have gone viral, Google — a company that’s been flexing its AI muscles for years and producing tons of research in the area — hasn’t had a public answer to those sorts of tools, some of which could threaten its core businesses.
During an earnings call today, Pichai talked about how the company plans to “unlock the incredible opportunities AI enables,” saying the tech is “reach[ing] an inflection point.” He also says that it was Google’s earlier AI research that helped spawn “the generative AI applications you’re starting to see today.”
As for its own applications, Pichai says the company was planning on doing a lot within the next few months, something it’s been preparing for since “early last year.” Google says the first model people will be able to engage directly with is LaMDA, the company’s conversational AI model, though it’s not exactly clear what form that will take.
With some people talking about how you could use ChatGPT to replace Google search (side note: that does not currently work very well for many reasons), I’m sure the company has been feeling the pressure to do something with all the AI tech it has. In fact, Google reportedly has employees testing its own AI-powered chatbot rivals to ChatGPT right now, and it wouldn’t surprise me if it was doing something similar with its currently internal tools for turning text into images or even video.
It seems like Google might still be taking it slow and won’t be giving people carte blanche access to do whatever they want. (No publically traded company wants to make another Tay.) Pichai says it’s perusing AI with a “deep sense of responsibility.” Later in the call, he said that Google would be the company would be “careful” and would be launching AI “more as labs features in certain cases, beta features in certain cases, and just slowly scaling up from there.”
Still, he says he believes the market is ready for these kinds of products. Whether society may be is an open question, but we’ve clearly seen that people want to play around with AI — and it seems like Google is ready to make its own version available.
Tech’s Biggest Companies Discover Austerity, to the Relief of Investors After years of expansion and billions in profits, Big Tech is pulling back from its famously lavish spending as a long boom finally ends.
Apple’s iPhone 14 Pro supply problems sank its holiday revenues
Photo by Amelia Holowaty Krales / The Verge
Apple reported its Q1 2023 earnings this afternoon. During the holiday quarter, the company took in $117.2 billion of revenue, down 5 percent year over year, and earnings per share of $1.88. It was the first such YOY for Apple since before the covid pandemic. Most of the curiosity around this quarter’s numbers was tied to iPhone sales; in early November, Apple warned of “longer wait times” for its flagship iPhone 14 Pro and Pro Max. Both phones were hard to come by during the height of the holiday shopping season, though stock has since leveled out.
But the supply issues, combined with consumers being extra mindful of spending amid an uncertain economic outlook, led to an 8 percent drop in iPhone revenue. “As we all continue to navigate a challenging environment, we are proud to have our best lineup of products and services ever, and as always, we remain focused on the long term and are leading with our values in everything we do,” CEO Tim Cook said in Apple’s earnings press release.
The iPhone 14 Pro / Pro Max delays were related to strict restrictions affecting Foxconn, the China-based factory Apple uses to assemble iPhones. Workers clashed with police during protests in the weeks that followed, with employees angry over delayed payments and deteriorating working conditions that stemmed from covid containment measures.
Apple also predicted that Mac sales would “decline substantially” compared to the year-ago quarter. And they sure did, dropping by 29 percent. The company introduced completely redesigned MacBook Pros in October 2021 that drew rave reviews and created buzz throughout the laptop industry. This spurred excitement and sales well beyond the Mac’s typical performance. By comparison, the holiday 2022 lineup was already well established — and the M2 MacBook Air and Mac Studio had already been on the market for a few months.
Apple’s Q1 fiscal quarter did see the introduction of new iPad models, however. The revamped entry-level iPad now has a design that takes after the iPad Pro and iPad Air. The Pro tablets also got a spec bump up to the M2 chip and added a new “hover” feature for the Apple Pencil. iPad revenue was up by 30 percent, signaling that the new models were well received. Wearables (a segment that includes the Apple Watch and AirPods) fell by 8 percent, however.
“During the December quarter, we achieved a major milestone and are excited to report that we now have more than 2 billion active devices as part of our growing installed base,” Cook said. Revenue for Apple’s all-important services division also rose 6 percent. On the software side, Apple rolled out its Emergency SOS via satellite feature for the iPhone 14 lineup in November. And macOS Ventura was released in late October with some useful features like Continuity Camera — though the overhauled settings interface hasn’t gone over well with all users.
Apple is widely expected to introduce a virtual reality headset sometime this spring. The device, rumored to be called Reality Pro, will mark Apple’s debut in a new product category that has been dominated by Meta to this point. There’s a lot riding on the headset after years of development and frequent rumors. Looking further out, the iPhone 15 Pro line is expected to switch over to USB-C, and Apple could finally bow to regulatory pressure and permit third-party app stores on iOS beginning with iOS 17 this fall.
Google’s fastest-growing business is insuring companies against their workers’ health
Granular Insurance, part of Verily, part of Alphabet, which is Google. | Image: Granular
I’ve heard people joke that Google only has a couple of successful businesses, primarily advertising. But it may have found another hit: insuring other companies against their workers’ potentially pricey medical care.
The Information is reporting that its healthcare company, Verily, more than doubled its revenue to become the biggest Alphabet subsidiary after Google proper — and that its health insurance business, Granular, is the biggest contributor to that growth. Granular’s revenue “rose nearly sixfold through the first nine months of last year to $151 million, from $27 million a year earlier,” writes The Information.
Exclusive: Google parent company Alphabet has never broken out financials of individual “other bets.” So here are financials of @Verily.
Turns out Alphabet’s biggest non-Google business is…selling insurance.
But Granular doesn’t sell health insurance to employees. It sells “stop-loss” insurance to employers who are worried that their own workers’ medical claims might hurt them.
See, not every company helps their employees pay for traditional insurance premiums where your doctors bill, say, UnitedHealthcare or Anthem or Aetna for your care (though those companies may be middlemen anyhow). Some think it’d be more cost-effective to “self-fund” and pay the medical claims of employees themselves.
Anyhow, Google / Alphabet / Verily’s Granular Insurance is one of many stop-loss insurance companies that promise to pay claims over a certain dollar threshold in exchange for its own regular premiums. Yes, that means companies that sign up are paying for insurance instead of paying for insurance — they’re betting that most employees won’t have enough claims to justify traditional insurance premiums but also betting that some workers might have huge ones.
What makes Granular different from other stop-loss providers? That’s less clear. The company advertises that “Granular uses an intelligent framework to better protect self-funded employers from the cost volatility of a workforce with diverse health-related needs,” but I think that just means it’s cheaper. I dug up some local government meeting materials from the San Joaquin Valley Insurance Authority in Fresno County, California, and they mostly seemed to be considering Granular to replace their existing stop-loss provider because the offer was competitive.
But perhaps it’s more competitive because Google thinks its data makes for more accurate bets. The San Joaquin Valley Insurance Authority notes that Granular would be providing its service alongside “Point6,” which appears to be this company that says it “brings actionable and integrated solutions focused on the 0.6% of the employee population that is driving 35% of employer healthcare expenditures.”
Either way, it’s not exactly the image that I typically associate with Google’s health efforts. Originally, Verily was most closely associated with the dream of a smart contact lens that’s long since been shelved.
Tears, blunders and chaos: inside Elon Musk’s Twitter
In the three months since Musk bought Twitter for £44bn, thousands have been sacked and the company has nosedived. Here, staff tell of a firm in disarray and an owner whose reputation is also plummeting
In April 2022, Elon Musk acquired a 9.2% stake in Twitter, making him the company’s largest shareholder, and was offered a seat on the board. Luke Simon, a senior engineering director at Twitter, was ecstatic. “Elon Musk is a brilliant engineer and scientist, and he has a track record of having a Midas touch when it comes to growing the companies he’s helped lead,” he wrote on the workplace messaging platform Slack.
Twitter had been defined by the leadership of Jack Dorsey – a co-founder who was known for going on long meditation retreats, fasting 22 hours a day, and walking five miles to the office – who was seen by some as an absentee landlord, leaving Twitter’s strategy and daily operations to a handful of trusted deputies. To Simon and those like him, it was hard to see Twitter as anything other than wasted potential.
Amazon’s delivery drones served fewer than 10 houses in their first month
Amazon’s hexagonal MK27-2 delivery drone. | Image: Amazon
It’s been nearly a decade since Amazon’s Jeff Bezos promised us delivery drones, but they aren’t off to a particularly impressive start. Roughly a month after Amazon Prime Air made its first deliveries in California and Texas, it’d served fewer than 10 households — and it’s already laid off more than half the employees at those locations.
That’s according to a pair of new reports at The Information and Business Insider, and Amazon isn’t denying it. Amazon spokesperson Maria Boschetti didn’t contest those numbers in an email to The Verge when we asked. But she also said that Prime Air is actually working to expand drone deliveries in both California and Texas, with the FAA’s approval.
And there may be a very good reason why Amazon doesn’t have a lot of customers for drone deliveries quite yet, as The Information points out: Amazon’s drone isn’t allowed to fly over roads by itself.
To cross the road while still abiding by FAA rules, Amazon employees had to act as spotters to make sure no vehicles were coming when the drone needed to fly across the street, a plan the FAA approved.
Seems ridiculous, right? Amazon’s drone is effectively a five-year-old who needs to hold hands to cross the street. A drone designed to replace humans needs humans to go places.
But before you decide Amazon’s drone isn’t capable, before you conclude that the FAA is halting progress in its tracks — either or both of which might be true! — I suggest you actually read the FAA’s decision that made things this way. I’ve embedded it at the bottom of this story.
If you’ve never dug into an FAA drone filing before, you might not be aware: the FAA doesn’t exactly hand out licenses to operate autonomous drones and drone delivery services. It creates specific exemptions to the United States’ strict airspace regulations, each with a long list of conditions that companies must follow.
Until last November, Amazon couldn’t even fly its drones outside of “sparsely populated areas,” couldn’t fly over buildings or within 100 feet of a building, and had to stick to flying over property under Amazon’s total control. The FAA required Amazon’s drone pilots to have the kind of private pilot license that’d let you fly a plane, not just a drone. If I’m reading correctly, every flight needed as many as six human beings, including observers and ground station operators.
That seems to have been wise. There were five crashes in four months at Amazon’s testing facilities in Oregon, and one crash ignited a 25-acre brush fire. The drone weighs nearly 90 pounds.
But those rules were during the experimental stage, and Amazon successfully argued last November that its experience and its new, safer and more autonomous MK27-2 drone didn’t need as many humans or safeguards. Among other things, the FAA cited its “enhanced perception system that allows for detection of people or obstacles below the UA during delivery or landing,” its auto-abort feature, remote alerts, and the fact that it can fly even if one of its six motors fails as reasons to nix those specific restrictions and more.
But not all of them, not by a long shot. Here are just some of the ground rules that still exist:
i. Operations over people are prohibited, unless otherwise approved by the Administrator;
ii. Overflight of power plants is prohibited;
iii. Overflight of schools during times of operation (e.g., elementary, middle, high, preschool and daycare facilities) is prohibited;
iv. Operations over or within 250 ft. laterally of moving vehicles are prohibited, unless otherwise approved by the Administrator
v. Overflight of any area deemed high risk by the Operator during the flight route design process are prohibited;
vi. Sustained flight within 250 ft. laterally of roadways is prohibited, and transitions over roadways is prohibited, unless otherwise approved by the Administrator;
vii. The UA must remain at least 100 ft. laterally from any person during all phases of flight, unless otherwise approved by the Administrator.
You’ll notice that Amazon’s drone still can’t cross a road by itself, and can’t come near or fly over people. That means Amazon’s customers can’t stand in their own backyards when the package gets dropped — unless their backyards are bigger than my house — and the FAA states that Amazon has to explicitly warn customers about that when they sign up.
In addition, both its California and Texas operations were limited to flights within just 3-4 miles of their drone launch sites. That’ll limit the number of potential customers.
Now, however, Amazon no longer needs nearly as many people for each flight, and the FAA is no longer requiring those employees to have the same flight or medical training — so it’s not surprising that Amazon would lay off many workers now it’s made them expendable.
“The recent staffing reductions do not impact our plans to deliver in these locations,” Amazon spokesperson Maria Boschetti tells The Verge. “We remain committed to our delivery operations in Lockeford and College Station and will continue to offer a safe and exceptional drone delivery service to our customers in both locations. We’ll gradually expand deliveries to more customers in those areas over time.”
She says the FAA approved flights to more customers in Lockeford, California and College Station, Texas just last week, and that Amazon is continuing to work on its next-generation drone, the MK30.
You can read the FAA’s revised requirements for Amazon’s MK27-2 delivery drone (and the reasons for its decisions) in the full document below.
Mark Zuckerberg says Meta is making this the ‘year of efficiency’
Meta CEO Mark Zuckerberg. | Laura Normand / The Verge
During Meta’s fourth-quarter earnings call with investors today, CEO Mark Zuckerberg explained why he wants to make this the “year of efficiency.”
“I just think we’ve entered somewhat of a phase change for the company,” he said, noting that headcount steadily climbed for nearly two decades, making it “very hard to really crank on efficiency while you’re growing that quickly.” Now, after laying off roughly 11,000 employees and putting a pause on most hiring, he is focused on “increasing the efficiency of how we make decisions.”
Practically, Zuckerberg said this means “flattening our org structure and removing some layers of middle management to make decisions faster.” As I reported in last week’s edition of my newsletter Command Line, he put it more plainly to employees during a recent all-hands meeting: “I don’t think you want a management structure that’s just managers managing managers, managing managers, managing managers, managing the people who are doing the work.” Indeed.
It turns out that Wall Street loves austerity these days. Meta’s stock price shot up nearly 20 percent after its earnings report was released with Zuckerberg’s commentary. The company took a one-time $4.2 billion charge related to the layoffs, the canceling of some building leases, and the backing out of costly data center projects. It also hinted that more layoffs could be on the horizon in its press release: “We may incur additional restructuring charges as we progress further in our efficiency efforts.”
Meta’s core business of serving ads remains challenged, with overall revenue declining by one percent in 2022 compared to 2021. But Zuckerberg struck an optimistic tone on the earnings call, saying that commentary about the company is lagging behind the progress he is seeing internally on key initiatives like the performance of Reels.
And for all the doom and gloom about the slow decay of Facebook itself, the numbers tell a different story; daily users hit a staggering two billion in the fourth quarter for the first time.
Image: Meta
Facebook’s daily active users count topped 2 billion for the first time.
“We’re going to be more proactive about cutting projects that aren’t performing or may no longer be as crucial,” Zuckerberg said. For now, that doesn’t apply to his metaverse efforts, which remain as costly as ever. Reality Labs, the division of Meta building its Quest headsets and coming AR glasses, reported an operating loss of $13.72 billion for 2022. That number is, amazingly, expected to increase this year.
Since Facebook rebranded to Meta in the fall of 2021, investors have grown increasingly worried that Zuckerberg is spending on his metaverse dreams with reckless abandon and little to show for it. That could still be true. But with this new “efficiency” push, he may be able to get away with it.
The best webcam to buy right now
Photo by Amelia Holowaty Krales / The Verge
You can get a great 1080p webcam for under $75. If you want to spend more, you can get 4K recording, AI head tracking, and even a real gimbal.
If you don’t already have a laptop with a built-in webcam — or the one on your laptop isn’t very good — you might need a standalone webcam. Maybe it’s to be able to show your face during work or school. Maybe it’s for chatting with friends or for side projects, like livestreaming or recording yourself.
But assuming you simply want to purchase a webcam, our goal is to make it easier for you to find the one that matches your needs and your budget. We tested popular models from brands like Logitech, Microsoft, Dell, Opal, Elgato, as well as some models from brands that you might not have heard of.
Each camera’s video quality was judged by using it with Zoom (where footage is prone to compression, as it is with other web-based video calling apps), as well as with OBS Studio or VLC to see how it compares with a locally stored recording.
The Logitech C920S Pro HD is a decent, inexpensive 1080p webcam with a built-in privacy shutter.
Logitech C920S Pro HD
Usually priced around $60, Logitech’s C920S Pro HD provides better video and mic quality than others that I tested in this price range. It can record in 1080p resolution at up to 30 frames per second, and while you won’t have an issue finding other similarly priced webcams with those specs, the C920S Pro’s out-of-box color balance, exposure, and relatively fast autofocus made it stand out from others I tested.
The picture quality has sufficient detail, and even in my relatively dark apartment, the C920S Pro had no trouble making my facial features look sharp. But it wasn’t a flawless presentation overall, as is to be expected for the price. In less-than-ideal lighting scenarios, the C920S Pro made my skin look oversaturated, with red spots where the webcam couldn’t compensate for the lack of lighting. Though in a work or play environment flush with natural light, this was noticeably less of an issue.
Picture quality aside, the C920S Pro has several welcome features, like a generous 78-degree field of view, status lights that activate when the webcam is being used, and an included privacy shutter. Some other nice features at this price include its strong articulating stand, which can sit atop your monitor or just as easily screw into a tripod. It’s just a great value for the price.
The best 4K webcam
Photo by Cameron Faulkner / The Verge
The Insta360 Link can record in 4K/60fps, and its gimbal gives it excellent tracking capabilities.
If you have $300 to spend on a webcam, the Insta360 Link is our new go-to recommendation. That’s around the same price as Opal’s C1 and Elgato’s Facecam Pro mentioned below, but some key factors give the Link an edge. It’s a 4K-ready webcam with a gimbal and loads of features.
That gimbal might be the star of the show for you. Instead of sitting stationary like most other webcams, the 0.5-inch Sony sensor is mounted to a motorized, three-axis arm that lets it move around. The gimbal allows the Link to track your head, and with an AI feature switched on, it can zoom in on your head, the top half of your body, or your whole body, if you’re standing far enough away. Insta360’s expertise in the action cam sector is showing in this consumer webcam.
The gimbal allows for a few alternate modes that might cut down on the tech you need to show off your work or to put your hobby on display on Twitch. One is deskview mode, which aims the Link’s slightly downward to show off your desk. There’s an overhead mode that aims the sensor directly at the ground (you’ll need to get clever with a tripod mount to utilize this or mount it oddly to your monitor). Then there’s a whiteboard mode, which tells the camera to look for four guiding stickers (included in the box) that prompt it to zoom in once it discovers them. This one seems gimmicky, but it works as intended and it could be great for people giving presentations.
Aside from those interesting features that make its $300 price more digestible, the Link delivers great video quality that’s better than any other webcam that I’ve tried so far. In my review, I primarily compared it to the Opal C1, which matches the Link’s 4K/30fps spec. Out of the box, the Opal C1 delivers a more contrast-rich image by default, while the Link looks more true to life and sharp but slightly duller by comparison (this can be tweaked to your taste in Insta360’s companion app). As for mic quality, neither option delivered stellar audio nor canceled out voices completely coming from across the room, although the Link was a little better at reducing some computer fan noise.
As I said in my review, I’d be happy to have either one of these two webcams on my desk for a virtual meeting. But when it comes to features, the Link is the more impressive option if you want to pay $300 for a webcam.
Other sub-$100 webcams we tested that you might like
The Microsoft Modern Webcam is okay, but its picture quality is not as good as the C920s Pro HD. Weirdly, its microphone doesn’t work at all in macOS and requires a separate utility to enable in Windows. For the price, just get the Logitech C920s.
Logitech’s C270 HD is the cheapest webcam-like object you can get, at around $20. Its 720p/30fps picture is fuzzy at best. Most people should spend more.
For this category, we also tested the Logitech Brio 500, but at $129.99, its 1080p/30fps video quality wasn’t that much better than the C920S Pro HD, although it added AI face tracking, which is a nice feature to have at this price. We hope that its clever privacy shutter and flexible mounting system make it into future (and more affordable) iterations, but this one isn’t totally worth its price unless you’re in love with the design.
Lastly, Dell’s barrel-shaped QHD webcam is tempting at $99.99, but its video quality looked awful. The capture feed doesn’t look natural, with darks having strange and overblown highlights. Plus, its longer shape is awkward if you plan to mount it to your laptop lid.
Other good high-end webcams we tested
The Opal C1 is an unconventional 4K webcam with DSLR-like quality and interesting machine-learning image effects. I was impressed with the image quality and features when I first reviewed it in late 2021, but it costs $300, and its software is still in beta and only works on Mac. The Insta360 Link is also $300 but doesn’t have the latter two drawbacks.
The Elgato Facecam Pro records in 4K/60fps but doesn’t have a built-in microphone. Most dedicated streamers won’t use the webcam’s built-in mic anyway, but others should take note.
The $140ish Logitech StreamCam is a great 1080p/60fps webcam with dual mics and smooth video capture, but features-wise, it’s stuck in the middle between the sub-$100 Logitech C920s and the Insta360 Link.
For this category, I also tested the OSBOT Tiny 4K, which like the Insta360 Link can rotate and move around to follow you in the frame. It also supports optional hand gestures to control its AI face tracking and zoom level. It even advertises the same 4K resolution at 30 frames per second recording. However, I was less impressed with its recording quality, and its companion software — while comprehensive — isn’t as polished as Insta360’s. Also, it’s physically larger, so it doesn’t sit as easily on top of a laptop display lid (and the included monitor mount is cheaply constructed). I think if you find it for $200 or less, I’d consider getting that model, but not when it’s priced within $100 or less of the better option that is Insta360’s Link.
Updated February 1st, 6:00PM ET:For 2023, we’ve tested new models, like the OSBOT Tiny 4K, Dell’s QHD webcam, and more. We also added a new pick for the best premium webcam, the Insta360 Link.
Meta earnings dropped by less than analysts expected
Parent company of Facebook and Instagram reports $32bn in revenue for fourth quarter, prompting rise in stock price
Meta investors got some good news in the social media company’s latest earnings report on Wednesday, which showed a smaller drop in revenues than analysts had expected.
Meta, which also owns Instagram and Whatsapp, reported $32bn in revenue for the fourth quarter, which drove a rise in its stock price in extended trading on Wednesday.
EA reportedly cancels new Titanfall single-player game
The wait for a new single-player Titanfall game continues. | Image: EA
EA has canceled an unannounced single-player game set in the same universe as Titanfall and Apex Legends, according to Bloomberg. The news comes on the heels of EA’s most recent earnings report on Tuesday, where it announced that it would be shutting down Apex Legends Mobile, halting development on a planned Battlefield mobile game, and delaying the release of Star Wars Jedi: Survivorby six weeks.
Apex Legends, a battle royale live service shooter from Respawn Entertainment that was born out of a plan to make a new Titanfall, has become one of EA’s marquee titles. Fans of Titanfall have been hoping for a new single-player game in that series following the well-received single-player campaign in Titanfall 2, but given Bloomberg’s report, it seems like that may not become a reality anytime soon.
The canceled project was codenamed Titanfall Legends,Bloomberg says. EA will try to find other jobs at the company for the game’s team of “about 50,” though those who don’t get a new job will be laid off with severance. EA didn’t immediately reply to a request for comment.
Respawn will still have a lot on its plate. In addition to working on Apex Legends, it’s developing two more Star Wars games in addition to Jedi: Survivor. And Respawn CEO Vince Zampella was put in charge of the Battlefield franchise shortly after the disappointing launch of Battlefield 2042.
EA has also shut down its Industrial Toys game studio, which was developing the Battlefield mobile game, according to VentureBeat.
ChatGPT maker OpenAI releases ‘not fully reliable’ tool to detect AI generated content
OpenAI is calling on educators to give their feedback on how the tool is used, amid rising concerns around AI-assisted cheating at universities
OpenAI, the research laboratory behind AI program ChatGPT, has released a tool designed to detect whether text has been written by artificial intelligence, but warns it’s not completely reliable – yet.
In a blog post on Tuesday, OpenAI linked to a new classifier tool that has been trained to distinguish between text written by a human and that written by a variety of AI, not just ChatGPT.
Dissecting Elon Musk’s Tweets: Memes, Rants, Private Parts and Echo Chamber As he fights for free speech online and promotes his companies, the new owner of Twitter spends a lot of time replying to his fans.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24405706/unnamed__12_.png)
Image: Google
/cdn.vox-cdn.com/uploads/chorus_asset/file/24401452/236514_Tiktok_transparency_center_AZaucha_0001.jpg)
Photo by Allison Zaucha for The Verge
/cdn.vox-cdn.com/uploads/chorus_asset/file/24401465/236514_Tiktok_transparency_center_AZaucha_0014.jpg)
Photo by Allison Zaucha for The Verge
/cdn.vox-cdn.com/uploads/chorus_asset/file/24401472/236514_Tiktok_transparency_center_AZaucha_0021.jpg)
Photo by Allison Zaucha for The Verge
/cdn.vox-cdn.com/uploads/chorus_asset/file/24401464/236514_Tiktok_transparency_center_AZaucha_0013.jpg)
Photo by Allison Zaucha for The Verge/cdn.vox-cdn.com/uploads/chorus_asset/file/24401453/236514_Tiktok_transparency_center_AZaucha_0002.jpg)
Photo by Allison Zaucha for The Verge
/cdn.vox-cdn.com/uploads/chorus_asset/file/24402787/meta_q4_22_slides_facebook_dau.jpg)
Image: Meta
/cdn.vox-cdn.com/uploads/chorus_asset/file/22820369/akrales_210831_4734_0071.jpg)
/cdn.vox-cdn.com/uploads/chorus_asset/file/23909869/cfaulkner_220727_5351_0003.jpg)
Photo by Cameron Faulkner / The Verge
