January 2024 – Page 44 – Tech News Feed

Intel’s Humbling

Comments

Google spent billions of dollars to lay people off

Illustration: The Verge

Google spent $2.1 billion on severance and other expenses as it laid off more than 12,000 employees over the course of 2023. And the layoff charges keep coming: in just the one month of 2024 so far, the company has already spent $700 million on employee severance charges as part of layoffs targeting another 1,000-plus roles. Alphabet, Google’s parent company, revealed the figure alongside its fourth-quarter earnings release on Tuesday.
Despite the cuts, Google ended 2023 with growth across most of its core business lines. Google posted $86 billion in revenue for the fourth quarter of 2023, a 13 percent increase year over year. The search giant’s core digital ads and cloud computing businesses also showed steady growth, which CEO Sundar Pichai credited to Google’s investments in generative AI.
Closing down offices cost Google, too
Google’s primary revenue generator, its search engine business, generated $48 billion, a nearly 13 percent jump year over year. The company’s subscription services and devices segment generated $10.7 billion, thanks primarily to subscriptions to YouTube Premium and Music, YouTube TV, and Google One, according to Pichai. YouTube’s advertising revenue was $9.2 billion, a 15 percent spike from last year.
On a call with investors, Pichai referred to 2024 as Alphabet’s “Gemini era,” referring to the company’s AI language model that it expects to work across all of Google’s core products. “Gemini is the first realization of the vision we had when we formed Google DeepMind, bringing together our two world-class research teams,” Pichai said. Coming around the bend is Gemini Ultra, an update of the original model which the company is still working on. “Gemini Ultra is coming soon. The team is already working on the next version and bringing it to our products, starting with Search.”
Google is currently the third-largest cloud provider in the world, trailing behind Microsoft’s Azure and AWS. But the cloud division picked up steam last year. The company reported $9.19 billion in revenue for Google Cloud, a considerable boost of 25.6 percent year over year.
Although layoffs cost Google a pretty penny, so did the real estate cutbacks due to said layoffs (including in locations like the pricey Bay Area). Shutting down physical office space cost Google a total of $1.8 billion for the entirety of 2023.
And Google’s continued investments in AI, or its “Gemini era,” means that even more layoffs are to follow in 2024 as the company makes cuts in other divisions to compensate.

Illustration: The Verge

Google spent $2.1 billion on severance and other expenses as it laid off more than 12,000 employees over the course of 2023. And the layoff charges keep coming: in just the one month of 2024 so far, the company has already spent $700 million on employee severance charges as part of layoffs targeting another 1,000-plus roles. Alphabet, Google’s parent company, revealed the figure alongside its fourth-quarter earnings release on Tuesday.

Despite the cuts, Google ended 2023 with growth across most of its core business lines. Google posted $86 billion in revenue for the fourth quarter of 2023, a 13 percent increase year over year. The search giant’s core digital ads and cloud computing businesses also showed steady growth, which CEO Sundar Pichai credited to Google’s investments in generative AI.

Closing down offices cost Google, too

Google’s primary revenue generator, its search engine business, generated $48 billion, a nearly 13 percent jump year over year. The company’s subscription services and devices segment generated $10.7 billion, thanks primarily to subscriptions to YouTube Premium and Music, YouTube TV, and Google One, according to Pichai. YouTube’s advertising revenue was $9.2 billion, a 15 percent spike from last year.

On a call with investors, Pichai referred to 2024 as Alphabet’s “Gemini era,” referring to the company’s AI language model that it expects to work across all of Google’s core products. “Gemini is the first realization of the vision we had when we formed Google DeepMind, bringing together our two world-class research teams,” Pichai said. Coming around the bend is Gemini Ultra, an update of the original model which the company is still working on. “Gemini Ultra is coming soon. The team is already working on the next version and bringing it to our products, starting with Search.”

Google is currently the third-largest cloud provider in the world, trailing behind Microsoft’s Azure and AWS. But the cloud division picked up steam last year. The company reported $9.19 billion in revenue for Google Cloud, a considerable boost of 25.6 percent year over year.

Although layoffs cost Google a pretty penny, so did the real estate cutbacks due to said layoffs (including in locations like the pricey Bay Area). Shutting down physical office space cost Google a total of $1.8 billion for the entirety of 2023.

And Google’s continued investments in AI, or its “Gemini era,” means that even more layoffs are to follow in 2024 as the company makes cuts in other divisions to compensate.

USCIS announces strengthened integrity measures for H-1B program

Comments

Replika’s new AI therapy app tries to bring you to a zen island

Tomo, the AI avatar | Luka, Inc.

AI companion company Replika partnered with the team behind the AI dating simulator Blush to release Tomo, a wellness and meditation app with an AI-generated avatar guiding users. Its the kind of concept that seemed destined as soon as generative AI took off, but in the time I’ve spent with Tomo, I found myself wondering if I can ever open up to an AI like I can with a real therapist.
Tomo, now generally available on the Apple iOS store, brings users to a virtual island retreat, greeted by an AI-generated avatar guide named… Tomo. Tomo offers programs to help people explore personal growth, mental well-being, and fulfillment. It also provides guided meditation, yoga, affirmation classes, and, most of all, talk therapy.
Users can try out Tomo for free for three days, after which they have to choose between paying $7.99 a week or $49.99 per year.
I got to try Tomo before the launch. The first time I launched the app, I was greeted by spa music, signaling that the app wants users to feel like they’re in a retreat rather than a therapist’s office. Then the avatar Tomo, drawn as a young woman standing in front of a traditional Japanese house on an island, asks me if I’m ready to begin. Tomo started asking me some questions to figure out what I hoped to work on.
“We worked with coaches and psychologists to come up with the programs for Tomo. We focused on the most common problems but also thought about what would work best with conversational AI,” Eugenia Kuyda, founder and CEO of Replika, tells The Verge in an email. “We had a lot of experience building coaching programs for Replika with clinical psychologists from UC Berkeley; for Tomo, we expanded that to mindfulness teachers to combine Eastern and Western practices.”
The approach felt like texting a therapist on text-based therapy services like BetterHelp. I already go to in-person therapy, so the experience of sharing more about myself was not new. Yet I have never been a text therapy fan; I prefer stream-of-consciousness conversation to typing out my anxieties. But for the sake of a hands-on, I kept texting with Tomo. It began building a profile based on my answers. My profile shows I like to focus on work, have money anxieties, and need help coping with stress; in other words, a journalist existing in late-stage capitalism. Tomo summarized our conversation, but it did mistakenly assume I had a “determination to pursue pottery” when really I’m just curious about it as a possible stress reliever.

Screenshot: Luka, Inc.
Screenshot of a Tomo therapy session.

After the initial conversation, users can explore other activities or “areas” of the island. There was no virtual pottery for me to apparently pursue; instead, the programs Tomo offers range from mastering the art of work-life balance, driving motivation, and improving sleep, which comes in the form of modules that users can finish in anywhere from two weeks to a month. Eventually, the developers said, people can unlock 3D objects around the island “that facilitate a deeper exploration of their inner sanctuary.”
Tomo, the avatar guide, is supposed to be powered by generative AI to have better conversations with users. But honestly, talking to Tomo didn’t feel much different from speaking with a regular chatbot. I couldn’t get it to participate in a little art therapy with me (it wouldn’t draw) or retail therapy (it couldn’t help with shopping or travel tasks I tried to assign it) — which meant that instead of feeling like a fully formed digital being on which I could unload my troubles, it really just felt like someone put background music on ChatGPT.
And while I found the guided meditations helpful, it also felt like other guided meditation apps I’ve tried before. Eventually I found myself losing interest, mainly because by the time I remembered to open Tomo, I already had assignments from my therapist to work on.
Using AI for mental health therapy remains controversial, especially as privacy protections still fall short for many technology-based mental health solutions. It’s a tricky business, one for which Replika, which is behind Tomo’s digital avatar, has already gotten into hot water. Italy banned Replika last year for failing to meet security standards in its Replika chat app. But Replika’s CEO says it is taking more precautions with Tomo, though the company did not give me its full privacy policy.
“We don’t share any information with any third parties and rely on a subscription business model. What users tell Tomo stays private between them and their coach,” Kuyda said.
Tomo is only available on iPhones; an Android version will be released later this year. Replika also plans to launch an app on Apple’s Vision Pro, paving the way for an even more immersive Tomo-guided meditation.

Tomo, the AI avatar | Luka, Inc.

AI companion company Replika partnered with the team behind the AI dating simulator Blush to release Tomo, a wellness and meditation app with an AI-generated avatar guiding users. Its the kind of concept that seemed destined as soon as generative AI took off, but in the time I’ve spent with Tomo, I found myself wondering if I can ever open up to an AI like I can with a real therapist.

Tomo, now generally available on the Apple iOS store, brings users to a virtual island retreat, greeted by an AI-generated avatar guide named… Tomo. Tomo offers programs to help people explore personal growth, mental well-being, and fulfillment. It also provides guided meditation, yoga, affirmation classes, and, most of all, talk therapy.

Users can try out Tomo for free for three days, after which they have to choose between paying $7.99 a week or $49.99 per year.

I got to try Tomo before the launch. The first time I launched the app, I was greeted by spa music, signaling that the app wants users to feel like they’re in a retreat rather than a therapist’s office. Then the avatar Tomo, drawn as a young woman standing in front of a traditional Japanese house on an island, asks me if I’m ready to begin. Tomo started asking me some questions to figure out what I hoped to work on.

“We worked with coaches and psychologists to come up with the programs for Tomo. We focused on the most common problems but also thought about what would work best with conversational AI,” Eugenia Kuyda, founder and CEO of Replika, tells The Verge in an email. “We had a lot of experience building coaching programs for Replika with clinical psychologists from UC Berkeley; for Tomo, we expanded that to mindfulness teachers to combine Eastern and Western practices.”

The approach felt like texting a therapist on text-based therapy services like BetterHelp. I already go to in-person therapy, so the experience of sharing more about myself was not new. Yet I have never been a text therapy fan; I prefer stream-of-consciousness conversation to typing out my anxieties. But for the sake of a hands-on, I kept texting with Tomo. It began building a profile based on my answers. My profile shows I like to focus on work, have money anxieties, and need help coping with stress; in other words, a journalist existing in late-stage capitalism. Tomo summarized our conversation, but it did mistakenly assume I had a “determination to pursue pottery” when really I’m just curious about it as a possible stress reliever.

Screenshot: Luka, Inc.
Screenshot of a Tomo therapy session.

After the initial conversation, users can explore other activities or “areas” of the island. There was no virtual pottery for me to apparently pursue; instead, the programs Tomo offers range from mastering the art of work-life balance, driving motivation, and improving sleep, which comes in the form of modules that users can finish in anywhere from two weeks to a month. Eventually, the developers said, people can unlock 3D objects around the island “that facilitate a deeper exploration of their inner sanctuary.”

Tomo, the avatar guide, is supposed to be powered by generative AI to have better conversations with users. But honestly, talking to Tomo didn’t feel much different from speaking with a regular chatbot. I couldn’t get it to participate in a little art therapy with me (it wouldn’t draw) or retail therapy (it couldn’t help with shopping or travel tasks I tried to assign it) — which meant that instead of feeling like a fully formed digital being on which I could unload my troubles, it really just felt like someone put background music on ChatGPT.

And while I found the guided meditations helpful, it also felt like other guided meditation apps I’ve tried before. Eventually I found myself losing interest, mainly because by the time I remembered to open Tomo, I already had assignments from my therapist to work on.

Using AI for mental health therapy remains controversial, especially as privacy protections still fall short for many technology-based mental health solutions. It’s a tricky business, one for which Replika, which is behind Tomo’s digital avatar, has already gotten into hot water. Italy banned Replika last year for failing to meet security standards in its Replika chat app. But Replika’s CEO says it is taking more precautions with Tomo, though the company did not give me its full privacy policy.

“We don’t share any information with any third parties and rely on a subscription business model. What users tell Tomo stays private between them and their coach,” Kuyda said.

Tomo is only available on iPhones; an Android version will be released later this year. Replika also plans to launch an app on Apple’s Vision Pro, paving the way for an even more immersive Tomo-guided meditation.

UPS To Cut 12,000 Jobs, Invest In AI For Efficiency

sdinfoserv writes: UPS is cutting 12,000 jobs, or about 2.5% of its global workforce. The cuts mostly effect managers and contractors. Meanwhile, as the company wrestles with struggling profits and sales, workers are to return to the office five days a week and the company is “investing in artificial intelligence (AI) as it pushes to become more efficient,” according to the BBC. [The job cuts are expected to reduce costs by $1 billion this year.] The BBC reports: The company said that reflected economic weakness in Europe and parts of Asia, as well as disruption in the US, where a strike threatened by staff over the summer led some customers to shift their business to rivals. UPS said it had since won back about 60% of that business and expected modest growth to start to return this year, with average daily volumes flat or up 2% in the US and flat or up 3% internationally. But its forecast was weaker than analysts had expected, sending shares down more than 7%.

It also warned that costs associated with its new contract with the Teamsters union would continue to weigh on the company over the next six months. As part of that deal, the average full-time driver won a pay and benefits package worth about $170,000 a year by the end of the five-year contract in 2028.

Gamers rejoice: AMD may be cooking up a GPU with RTX 4080 performance at half the price

AMD could be working on only mid-range graphics cards in the next generation, according to a new report.

According to a new report, AMD may be working on a graphics card that offers the same performance as the Nvidia RTX 4080 but at only half the price. This isn’t the first time these rumors cropped up either, with a report back in 2023 about the manufacturer focusing on the mid-range market.

The latest report asserts that the mid-range options of the AMD Radeon 8000 series will allegedly cost between $400-600 (around £315-£472 / AU$606-AU$910), according to PCGamesN. This is a huge price cut on its own but becomes a downright steal when compared to the MSRP of the RTX 4080. And in a graphics card generation that’s already suffering from a severe lack of affordable options, the possibility of AMD pushing out a card so powerful and budget-friendly is an excellent way to appease a significant market while standing out from its competition.

There are also rumors that this level of performance for the AMD Radeon 8000 will max out here, leaving the high-end graphics cards to Nvidia in the next generation. While it sounds like a poor decision, it’s a strategy that makes sense for the manufacturer, since this would leave AMD to focus development on its mid-range and lower-end options instead.

It’s also important to note that AMD has been struggling against Nvidia in terms of ray-tracing options in its graphics cards. Team Green also has the advantage of its DLSS 3 versus Team Red’s FSR 3, which is upscaling tech that takes a lower resolution frame and upscales the resolution to a higher one (so 1080p to 4K) so they are quicker to render, boosting frame rates. It should be noted though that this advantage is narrowing.

According to YouTube leaker Moore’s Law is Dead, AMD had a high-end Radeon RX 8000 graphics card, codenamed Navi 4C. A diagram shown in the video asserts that the card features three “Shader Engine Dies” placed on top of each “Active Interposer die,” and a “Multimedia and I/O Die” – which would have resulted in a card with nine separate Shader Engine Dies.

But it seems that this high-end card wasn’t fated to be, as the card could have demonstrated top-tier performance but at an extremely expensive price point. We’re talking possibly $2,500. And considering that AMD still struggles with ray tracing, that asking price would be a tall order for a market that already is balking at expensive components.