Welcome to Thursday's Inside AI!
We'd like to offer a quick reminder about the benefits of signing up for Premium, which provides access to all Inside AI issues ad-free, along with these recent regularly published features:
- Our Podcast Notes summarizing recent podcasts in the AI sector, such as this interview with Jason Matheny of the U.S. National Security Commission on Artificial Intelligence.
- Our deep dives into the latest reports on AI, including the new AI technologies that made it on Gartner's list of emerging technologies.
- AI Masterclasses featuring expert thought leaders, such as this special feature on the seven ways AI improves business processes.
- Weekly updates on the AI/ML/robotics startups that raised venture capital funding.
...and more.
If you want to start reading this content today, we're offering a 14-day FREE trial. To take advantage of this deal while it still lasts, upgrade to premium!
|
|
|
As Nvidia prepares its $40B takeover of semiconductor company Arm Holding from SoftBank, the Chinese government is concerned that the company’s technology could be politically weaponized against Chinese firms, and may try to block the deal. 95% of chips designed in China use Arm’s technology.
- The deal - considered one of the largest tech acquisitions ever - faces heightened scrutiny due to its magnitude and significance. The $40b purchase will require regulatory approval from the U.S., U.K., EU, and China.
- Chinese tabloid The Global Times published an op-ed this week that says China-based tech companies would "be placed at a big disadvantage in the market" if Nvidia acquires Cambridge-based Arm. It called the deal "disturbing" for the Chinese and European companies that use the company's chip designs.
- Chinese companies on the U.S. blacklist, including a number of AI companies, could risk being cut off from using Arm-based chips, the op-ed said.
- Arm designs chips for smartphone makers like Apple and Samsung, while Huawei subsidiary HiSilicon and other Chinese semiconductor design firms rely on its IP.
- The deal is viewed as a likely boost to Nvidia's data center business, but the company has said it is primarily about AI. The deal "brings together NVIDIA's leading AI computing platform with ARM's vast ecosystem to create the premier computing company for the age of artificial intelligence," Nvidia said.
- The company views GPU-accelerated AI as a huge growth sector with implications in embedded systems and others linked to deep learning, self-driving cars, and more. It could sell its AI and graphics chips to Arm customers as well.
|
|
Snowflake's IPO has implications in the future of AI and ML training as businesses turn more to cloud services to store and analyze data. The cloud database company completed its IPO Tuesday at a price of $120 per share, which was followed by a closing price of $253 yesterday, valuing it at a record $70 billion. Its software architecture ostensibly makes it easier and less expensive to analyze the large datasets that are used to train machine learning algorithms.
More:
- Due to the way it structures its software, Snowflake can challenge cloud powerhouses like Amazon's AWS and Microsoft's Azure through its ability to run high compute AI programs more efficiently.
- Snowflake's computing power can store and analyze data separately and, using a separate system, dedicate more or less resources to each depending on the demand. It charges clients for the exact amount of storage and computing power they require.
- AWS, conversely, doesn't divide its data storage and analysis, so customers might pay for unused computing power to analyze the data.
- Snowflake's $3.4b is the largest amount raised in a U.S. IPO since Uber's $8.1b in May 2019. It has reported increased revenue for the last eight quarters, and revenue increased by 174% YoY in the last fiscal year.

|
|
100 military and defense officials from 13 countries held a virtual conference this week to talk about the use of AI in military practices. The U.S. Pentagon’s Joint AI Center (JAIC) hosted the meeting, which is considered the kickoff of the JAIC's AI Partnership for Defense, a group that aims to forge international cooperation with AI practices.
More:
- The conference included reps from the U.S., Australia, Canada, Denmark, Estonia, Finland, France, Israel, Japan, Norway, the Republic of Korea, Sweden, and the U.K. China and Russia are not involved.
- Their conversations focused on ethical principles and implementation of AI in the defense sector, including the best ways to incorporate them into "the AI delivery pipeline." They plan to talk about responsible AI that doesn't impede humans rights, the JAIC said.
- The JAIC says the forum will help nations dialogue as they move from hardware- to software-centric forces and data-focused AI-ready militaries.
- A strong coalition of AI-ready militaries will help realize the potential of emerging tech and "strengthen global security, deter shared threats, and respond more efficiently together to natural disasters," the JAIC said.
- Mark Beall, the JAIC's head of strategy and policy, said it's not a U.S-led initiative and is trying to be as "inclusive" as possible with like-minded countries. If competitors like China adopted responsible AI, "we’d welcome that," he said.
- The JAIC has communicated with 40-50 other nations that could join, though Beall doesn't see China or Russia joining any time soon. The JAIC wants to hold three "Partnership for Defense" meetings annually.
- The Pentagon approved its first ethical AI principles earlier this year and has started incorporating them into its requests for proposals.
BREAKING DEFENSE
|
|
|
Facebook introduced the research endeavor Project Aria to build AR devices, which will include AI technology. The project doesn't entail actual commercial smart glasses (yet), but is considered a sensor array for future glasses. Through the project, Facebook will research how to develop the software/hardware required for widespread use of AR glasses, which it says are about 5-10 years away.
More:
- Facebook says building glasses that can work with most face sizes and shapes, and creating the accompanying software, will require "several generations of breakthroughs." This includes contextualized AI, enhanced audio and visual, and light enough frames.
- Starting this month, ~100 testers in Seattle and the Bay area will wear a head-worn sensor array to start mapping the real world. Facebook's spatial scan known as LiveMaps will be the foundation for future AR glasses.
- Aria will essentially be a testbed to help Facebook engineers develop the AR ecosystem by "capturing both POV photos and video as well as tracking the wearer’s head and eye movements as well as location data."
- Zuckerberg said that a lot of people are trying to take shortcuts when it comes to building AR glasses by just showing some heads-up information. He said: "I call that 'putting an Apple Watch on your face.'"
- Facebook recently posted many new AI job openings including a Product Manager of AI Applied Research and a Product Design Prototyper of AI Voice.
A version of this story first appeared in today's Inside XR.
|
|
Benjamin Lee
In case you missed it: The Library of Congress unveiled the Newspaper Navigator, an AI-based tool that allows users to search for images from American newspapers from 1789-1963. It uses machine learning to help users search through millions of images based on visual similarity.
More:
- The tool comes from the Library of Congress Innovator in Residence Benjamin Lee, who trained computer algorithms to search through newspaper pages for photographs, illustrations, maps, and more.
- Lee, a Washington University researcher, used an object detection model trained on WW1-era page annotations from the Library's digital newspaper collection.
- To use the navigator, enter a keyword and the AI will find matches that you can download. You can also read accompanying articles and view the full issue of the newspaper.
- The code is available on GitHub.
THE NEXT WEB
|
|
QUICK HITS
- Chinese AI Yitu Internet Technology Co. Ltd., one of the AI firms on the U.S. blacklist, is considering an IPO in Shanghai.
- The U.S. Marshals Service plans to use a smartphone-based facial recognition app to help its agents identify prisoners more quickly during transfer.
- Origami launched Origami Forecast, a machine learning software product that provides energy companies with power generation, demand, and market price forecasts.
- Palo Alto Networks introduced machine learning, analytics, and hardware appliances for its SD-WAN package.
- Public enterprise data storage company Pure Storage will buy Kubernetes-based cloud-native storage/data-management platform Portworx for $370M. (For more news like this, subscribe to Inside Venture Capital).
- Slack is gathering industry leaders and product experts at Slack Frontiers, a free virtual conference dedicated to digital transformation.*
*This is a sponsored post.
Also, we're hiring! Check out our available positions:
|
|
|
|
Beth Duckett is a former news and investigative reporter for The Arizona Republic, who has written for USA Today, American Art Collector, and other publications. A graduate of the Walter Cronkite School of Journalism, she won a First Amendment Award and a Pulitzer Prize nomination for her original reporting on problems within Arizona's pension systems.
|
|
Editor
|
Sheena Vasani is a journalist and UC Berkeley, Dev Bootcamp, and Thinkful alumna who writes Inside Dev and Inside NoCode.
|
|