What’s happened in AI: December 1st-15th

By | December 16, 2017

Exciting two weeks in AI. Highlights include Andrew Ng’s new startup Landing.ai, Black in AI panel at NIPS 2017 and Google announcing its AI made its own AI. Also, check out Venturebeat’s AI weekly update as they do a fantastic job.

Company news / recent developments:

Google’s AI made its own AI, and it’s better than anything ever created by humans

  • AutoML uses a technique called reinforcement learning as well as neural networks to develop the daughter AIs. NASNet, a child AI, was developed and trained to recognize objects in real-time video streams
  • When benchmarked with industry standard validation sets, it was found to be 82.7% accurate at recognizing known objects. This is 1.2% higher than anything seen before and the system is also 4% more efficient than the previous best algorithms

Google is opening a China-based research lab focused on artificial intelligence

  • Like the general talent with Google China, this AI hiring push isn’t a sign that Google will launch new services in China. Although it did make its Google Translate app available in China earlier this year in a rare product move on Chinese soil
  • Dr. Fei-Fei Li, formerly the director of Stanford University’s Artificial Intelligence Lab, was a high-profile arrival when she joined Google one year ago. She will lead the China-based team alongside Jia Li, who was hired from Snap where she had been head of research at the same time as Li

Andrew Ng’s Landing.ai wants to bring artificial intelligence to the manufacturing industry

  • After leaving Baidu earlier this year, it wasn’t quite clear what exactly Ng was up to (except for teaching more classes), but today he announced the launch of Landing.ai, a new startup that focuses on bringing artificial intelligence to the manufacturing industry
  • First strategic partner partner is Foxconn, who they’ve worked with since July

Partnerships:

Honda is working with Chinese AI unicorn SenseTime on self-driving car tech

  • SenseTime, which raised a $410 million funding round this past summer and counts Qualcomm among its investors, is particularly well known for its object recognition technology which has been used by both public and private entities in China
  • Honda’s own AI falls into ‘scene understanding,’ ‘risk prediction,’ and ‘action planning,’ all of which relates to gauging urban motoring scenarios and responding to them in the right way. It said earlier this year that it plans to introduce cars with level 4 automation by 2020

Microsoft looks to make Bing results smarter with new AI powered features and Reddit partnership

  • The updates are focused on bringing more conversational, nuanced answers to users in the “cards” UI being used by competitors like Google. New features starting to roll out today include: Intelligent Search, Intelligent Image Search and Conversational Search
  • One of the areas that they’re looking to one-up Google interestingly is in giving users the opportunity to get multiple viewpoints on some questions, so if you ask Google whether a certain type of food is healthy they may tell you what it’s good for and where it might not be

Nvidia and construction giant Komatsu partner on AI for job site safety

  • This team-up builds on work done by Komatsu previously, as it originally rolled out its SMARTCONSTRUCTION program to build connected, smart worksites starting in 2015. Now, Nvidia’s tech will help Komatsu build 3D visualizations of entire construction sites, with real-time representation of people and machinery working on the ground
  • Drones provided by SkyCatch will offer imaging for visualization, and Jetson will provide image processing and recognition to cameras spread throughout Komatsu’s construction equipment for computing at the edge

Lockheed selects NEC artificial intelligence software to study space data (Space News)

  • The NEC software is called “system invariant analysis technology.” Its engine takes data collected from sensors to learn the behavior of systems — computer networks, power plants, factories and buildings, for example. The systems are taught to automatically detect inconsistencies and recommend fixes
  • The agreement with Lockheed furthers NEC’s efforts to increase its presence in the United States and have its technology integrated in U.S. government projects. NEC recently announced the expansion of its offices in Washington, D.C. It is now a prime contractor for the U.S. Department of Homeland Security’s biometric exit pilot program that uses the company’s facial recognition rapid matching technology

Fundraising:

Commerce platform iZettle raises $47m at a $950m valuation

  • iZettle — the commerce platform based out of Stockholm that competes against companies like Square, Paypal and SumUp to provide card transactions using smartphones and tablets as well as related accounting services — has raised another €40 million ($47 million) as it approaches a $1 billion valuation
  • CEO and co-founder Jacob de Geer told TechCrunch the money will go towards expanding into more markets beyond the 12 where it currently operates in Europe and Latin America
  • Swedish publication Dagens Industri, which reported rumors of the funding yesterday, pegged the valuation at the time at $948 million (8 billion Swedish crowns), and we have confirmed that figure with our own sources

With $10m in new funding, huddly launches its smart GO camera for video conferencing

  • This marks the first time that the Huddly camera is available for direct purchase. Until now, it was only available as part of Google’s Hangouts Meet hardware kits
  • The idea behind the GO, which works with any Windows, Mac and Linux machine, is to put as much intelligence as possible right into the camera (though some of this is still a work in progress) to make video conferences more productive.
  • This means that it’s using on-board computer vision to understand who is in a room, what the lighting looks like and what kind of background noises are possibly affecting the sound

PullRequest pulls in $2.3m seed round led by Google’s new AI fund, Gradient Ventures

  • The Slack Fund, Fika Ventures, Defy Ventures, Lynett Capital, FundersClub and Joe Montana’s Liquid2 Ventures also participated
  • The company has been concentrating on providing code review as a service, a task that often gets lost in today’s high-speed agile development cycles. The company has set up a system of on-demand code reviewers who check for bugs, security issues, coding standards and performance problems

Government / regulatory policy:

New York City moves to establish algorithm-monitoring task force

  • New York City may soon gain a task force dedicated to monitoring the fairness of algorithms used by municipal agencies
  • Formed from experts in automated systems and representatives of groups affected by those systems, it would be responsible for closely examining algorithms in use by the city and making recommendations on how to improve accountability and avoid bias
  • The bill, which doesn’t have a fancy name, has been approved by the city council and is on the Mayor’s desk for signing. The New York division of the ACLU has argued in favor of it

US Congress is trying to define what artificial intelligence actually means (Quartz)

  • A new bill drafted by senator Maria Cantwell asks the Department of Commerce to establish a committee on artificial intelligence to advise the federal government on how AI should be implemented and regulated
  • Passing of the bill would trigger a process in which the secretary of commerce would be required to release guidelines for legislation of AI within a year and a half

Artificial intelligence helps discover new planets (ABC)

  • Machine learning has helped to discover two new planets amidst all the data collected from NASA’s Kepler Space Telescope
  • One of the planets is the circling the Kepler-90 star. It’s the eighth planet found orbiting Kepler-90, which means that solar system hosts the same number of planets as the Milky Way

Events:

XPrize’s $5m AI competition is down to 59 teams (Seeker)

  • The IBM Watson XPrize in artificial intelligence aims to promote technologies that emphasize collaboration between humans and machines
  • A $3 million grand prize, $1 million second-place prize and $500,000 third-place prize will be awarded to the teams that receive the top scores in the competition. Additionally, $500,000 Milestone Prizes will be awarded in the fall of both 2018 and 2019