Category: Startups

Sources: Twitter only has one staff left on a team dedicated to removing CSAM after layoffs, despite Musk saying removing child exploitation is top priority (Morgan Meaker/Wired)

Morgan Meaker / Wired:
Sources: Twitter only has one staff left on a team dedicated to removing CSAM after layoffs, despite Musk saying removing child exploitation is top priority  —  Just one person remains to enforce the company’s ban on child sexual abuse content across Japan and the Asia Pacific region.

Analysis: high-profile Republican members of Congress saw an uptick in Twitter followers in the first few weeks of Musk’s reign, while Democrats lost followers (Washington Post)

Washington Post:
Analysis: high-profile Republican members of Congress saw an uptick in Twitter followers in the first few weeks of Musk’s reign, while Democrats lost followers  —  Shift in follower counts for Elizabeth Warren, Ted Cruz show how the platform is beginning to change under Musk

Nvidia and Microsoft team up to build massive AI cloud computer

Enlarge / Nvidia and Microsoft are teaming up on an AI cloud supercomputer. (credit: Nvidia)

On Wednesday, Nvidia announced a collaboration with Microsoft to build a “massive” cloud computer focused on AI. It will reportedly use tens of thousands of high-end Nvidia GPUs for applications like deep learning and large language models. The companies aim to make it one of the most powerful AI supercomputers in the world.

In turn, the new supercomputer will feature thousands of units of what is arguably the most powerful GPU in the world, the Hopper H100, which Nvidia launched in October. Nvidia will also provide its second most powerful GPU, the A100, and utilize its Quantum-2 InfiniBand networking platform, which can transfer data at 400 gigabits per second between servers, linking them together into a powerful cluster.

Meanwhile, Microsoft will contribute its Azure cloud infrastructure and ND- and NC-series virtual machines. Nvidia’s AI Enterprise platform will tie the whole thing together. The companies will also collaborate on DeepSpeed, Microsoft’s deep learning optimization software.

Read 4 remaining paragraphs | Comments

Amazon begins layoffs of up to 10,000 jobs, blames “uncertain” economy

Enlarge / An Amazon Echo Dot 4th Generation smart speaker. (credit: Getty Images | Gado )

Amazon has started a round of layoffs that reportedly could cut about 10,000 jobs. Layoffs began yesterday in the Amazon hardware division that makes products including Echo, Alexa, Fire, and Kindle devices.

“We notified impacted employees yesterday, and will continue to work closely with each individual to provide support, including assisting in finding new roles,” Amazon Senior VP of Devices & Services Dave Limp wrote to his department’s staff today in a memo posted publicly by Amazon. Limp wrote that Amazon “continue[s] to face an unusual and uncertain macroeconomic environment… After a deep set of reviews, we recently decided to consolidate some teams and programs.”

“It pains me to have to deliver this news as we know we will lose talented Amazonians from the Devices & Services org as a result… While I know this news is tough to digest, I do want to emphasize that the Devices & Services organization remains an important area of investment for Amazon,” the memo said.

Read 6 remaining paragraphs | Comments

ISP deploys fiber service with a wrinkle—the users themselves own each network

Enlarge / Horizontal boring equipment installing fiber in Los Altos Hills, California. (credit: Los Altos Hills Community Fiber)

Our recent article about Silicon Valley residents who formed a co-op Internet service provider might have people wondering what it would take to get the same thing in their hometowns. The most obvious obstacle is price—in Los Altos Hills, California, residents have had to pay anywhere from $5,000 to $12,000 upfront for a fiber-to-the-home Internet connection.

But the company that built the Los Altos Hills network says its model isn’t just for wealthy people. “This is not the 1 percent solution, as people derisively call it to my face,” Next Level Networks CEO David Barron told Ars in a phone interview a few weeks ago. “Los Altos Hills was unique.”

Los Altos Hills residents were the first to contract with Next Level Networks, and Barron said the company has “a fairly aggressive expansion plan to go into a number of markets throughout the United States in the next five years.”

Read 53 remaining paragraphs | Comments

3D for everyone? Nvidia’s Magic3D can generate 3D models from text

Enlarge / A poison dart frog rendered as a 3D model by Magic3D. (credit: Nvidia)

On Friday, researchers from Nvidia announced Magic3D, an AI model that can generate 3D models from text descriptions. After entering a prompt such as, “A blue poison-dart frog sitting on a water lily,” Magic3D generates a 3D mesh model, complete with colored texture, in about 40 minutes. With modifications, the resulting model can be used in video games or CGI art scenes.

In its academic paper, Nvidia frames Magic3D as a response to DreamFusion, a text-to-3D model that Google researchers announced in September. Similar to how DreamFusion uses a text-to-image model to generate a 2D image that then gets optimized into volumetric NeRF (Neural radiance field) data, Magic3D uses a two-stage process that takes a coarse model generated in low resolution and optimizes it to higher resolution. According to the paper’s authors, the resulting Magic3D method can generate 3D objects two times faster than DreamFusion.

Magic3D can also perform prompt-based editing of 3D meshes. Given a low-resolution 3D model and a base prompt, it is possible to alter the text to change the resulting model. Also, Magic3D’s authors demonstrate preserving the same subject throughout several generations (a concept often called coherence) and applying the style of a 2D image (such as a cubist painting) to a 3D model.

Read 3 remaining paragraphs | Comments

New Meta AI demo writes racist and inaccurate scientific literature, gets pulled

Enlarge / An AI-generated illustration of robots making science. (credit: Ars Technica)

On Tuesday, Meta AI unveiled a demo of Galactica, a large language model designed to “store, combine and reason about scientific knowledge.” While intended to accelerate writing scientific literature, adversarial users running tests found it could also generate realistic nonsense. After several days of ethical criticism, Meta took the demo offline, reports MIT Technology Review.

Large language models (LLMs), such as OpenAI’s GPT-3, learn to write text by studying millions of examples and understanding the statistical relationships between words. As a result, they can author convincing-sounding documents, but those works can also be riddled with falsehoods and potentially harmful stereotypes. Some critics call LLMs “stochastic parrots” for their ability to convincingly spit out text without understanding its meaning.

Enter Galactica, an LLM aimed at writing scientific literature. Its authors trained Galactica on “a large and curated corpus of humanity’s scientific knowledge,” including over 48 million papers, textbooks and lecture notes, scientific websites, and encyclopedias. According to Galactica’s paper, Meta AI researchers believed this purported high-quality data would lead to high-quality output.

Read 6 remaining paragraphs | Comments

Thinking about taking your computer to the repair shop? Be very afraid

Enlarge (credit: Getty Images)

If you’ve ever worried about the privacy of your sensitive data when seeking a computer or phone repair, a new study suggests you have good reason. It found that privacy violations occurred at least 50 percent of the time, not surprisingly with female customers bearing the brunt.

Researchers at University of Guelph in Ontario, Canada, recovered logs from laptops after receiving overnight repairs from 12 commercial shops. The logs showed that technicians from six of the locations had accessed personal data and that two of those shops also copied data onto a personal device. Devices belonging to females were more likely to be snooped on, and that snooping tended to seek more sensitive data, including both sexually revealing and non-sexual pictures, documents, and financial information.

Blown away

“We were blown away by the results,” Hassan Khan, one of the researchers, said in an interview. Especially concerning, he said, was the copying of data, which happened during repairs for one from a male customer and the other from a female. “We thought they would just look at [the data] at most.”

Read 14 remaining paragraphs | Comments

Apple iPhone factory workers clash with police in China

Enlarge / Workers walk outside Hon Hai Group’s Foxconn plant in Shenzhen, China, in 2010. (credit: Getty)

Violent worker protests have erupted at the world’s largest iPhone factory in central China as authorities at the Foxconn plant struggle to contain a COVID-19 outbreak while maintaining production ahead of the peak holiday season.

Workers at the factory in Zhengzhou shared more than a dozen videos that show staff in a standoff with lines of police armed with batons and clad in white protective gear. The videos show police beating workers, with some bleeding from their heads and others limping away from chaotic clashes.

Beijing’s strict zero-COVID regime has posed big challenges for the running of Foxconn’s Zhengzhou plant, which typically staffs more than 200,000 workers on a large campus in the city’s suburbs.

Read 14 remaining paragraphs | Comments

Meta researchers create AI that masters Diplomacy, tricking human players

Enlarge / A screenshot of an online game of Diplomacy, including a running chat dialog, provided by a Cicero researcher. (credit: Meta AI)

On Tuesday, Meta AI announced the development of Cicero, which it claims is the first AI to achieve human-level performance in the strategic board game Diplomacy. It’s a notable achievement because the game requires deep interpersonal negotiation skills, which implies that Cicero has obtained a certain mastery of language necessary to win the game.

Even before Deep Blue beat Garry Kasparov at chess in 1997, board games were a useful measure of AI achievement. In 2015, another barrier fell when AlphaGo defeated Go master Lee Sedol. Both of those games follow a relatively clear set of analytical rules (although Go’s rules are typically simplified for computer AI).

But with Diplomacy, a large portion of the gameplay involves social skills. Players must show empathy, use natural language, and build relationships to win—a difficult task for a computer player. With this in mind, Meta asked, “Can we build more effective and flexible agents that can use language to negotiate, persuade, and work with people to achieve strategic goals similar to the way humans do?”

Read 8 remaining paragraphs | Comments

Scroll to top
Generated by Feedzy