Inside | Real news, curated by real humans
Inside AI

Inside AI (Aug 13th, 2019)

1. Nvidia says its conversational AI platform has the fastest training record. The GPU maker says it set a world record for training the Google-built Bidirectional Encoder Representations from Transformers (BERT) language model - which it did in 53 minutes on its DGX SuperPOD deep learning server, a split neural network. Nvidia also claims to have the fastest BERT inference, at 2.2 milliseconds, and the largest model trained at 8.3 billion parameters. "Without this kind of technology, it can take weeks to train one of these large language models," said Bryan Catanzaro, Nvidia's VP of applied deep learning research. - ZDNET

  • Email gray
  • Permalink gray

2. Microsoft is recruiting more workers to staff its $34 million R&D center for AI in Taiwan. The software giant chose the East Asian state to grow its AI business, with a focus on research into enterprise software. Specifically, Microsoft is working to expand its AI-powered meeting assistance technology to different industries, says Ken Sun, general manager of Microsoft Taiwan. Employees can tell the virtual assistant bot how many people will participate in a meeting, and the bot will determine the best space; the goal is to use the AI at other administrative offices, medical exam rooms, and PowerPoint slides within the next five years, Sun says. - FORBES

  • Email gray
  • Permalink gray

3. The Indian ride-sharing and taxi company Ola acquired the AI startup Pikup.ai. Earlier this year, Ola announced plans for an advanced tech center in the Bay area, where it plans to develop next-gen technology for autonomous, electric, and connected vehicles. In its acquihire (acquisition plus hiring) of Piku.ai, Ola plans to tap into the startup's AI, computer vision, and sensor fusion technology, its expertise in the industry, and data. Terms of the deal were not disclosed. - FINANCIAL EXPRESS

  • Email gray
  • Permalink gray

4. IBM's Triton AI Supercomputer made its debut at the University of Miami. Nicholas Tsinoremas, director of the University of Miami Center for Computational Science and vice provost for data and research computing, says the $3.7 million system will help the university advance its research and "enable more discoveries" using data science and big data.  “From the purpose-built hardware to the integrated machine learning and deep learning software stack, the IBM technology in Triton represents a new chapter in the way researchers approach data and computation," says David Turek, VP of Exascale Systems for IBM Cognitive Systems. - INSIDE HPC

  • Email gray
  • Permalink gray

5. British artist Anna Ridler collected and photographed 10,000 tulips as part of a dataset and artwork featured at the “AI: More than Human” exhibition in London. Ridler visited the Netherlands to gather the tulips, label each picture (noting its stripes and colors), and then used the dataset to create a 1,614 square-foot installation, comprised of the images, titled Myriad (Tulips). In addition to the artwork, Ridler, a self-taught coder, fed the photos and labels into a machine learning model to create a three-screen video installation. The model is linked to a real-time feed that reacts to cryptocurrency prices. “As the price of Bitcoin goes up, the tulip becomes more stripy,” Ridler says. “As the price of Bitcoin goes down, it starts to become more plain." - BLOOMBERG

  • Email gray
  • Permalink gray

6. Sogou, a Chinese search engine, created AI-based avatars of Chinese authors, which it plans to use to narrate audiobooks in videos. The public company, which raised $585 million in its 2017 IPO to help grow its AI tech, utilized text-to-speech technology and video clips from a conference to develop avatars of te authors Yue Guan and Bu Xin Tian Shang Diao Xian Bing. Since last fall, Sogou's AI news anchors - whose movements are are synthesized via deep learning techniques - have been on the air at China's state-run news agency, Xinhua. - ENGADGET

  • Email gray
  • Permalink gray

7. Google Research’s Brain Team developed a soccer video game where researchers can test and develop better reinforcement algorithms. The Google Research Football Environment - which is available with an open-source license - was based on the publicly available game Gameplay Football, which mimics a full soccer game. They modified it to allow a measure of success for machines, according to the paper, "Google Research Football: A Novel Reinforcement Learning Environment." - MIT TECH REVIEW

  • Email gray
  • Permalink gray

8. Rob Carpenter, Valyant AI‘s CEO and founder, spoke with Observer about his proprietary AI conversational platform for fast food ordering. The enterprise AI, named Holly, has the potential to revolutionize the drive-through experience, he says. As Carpenter explains, fast food restaurants have very high turnover rates - upwards of 200 percent to 300 percent per year - and Holly could fill that gap. Unlike Alexa and other consumers AIs, Holly is not wide but deep, able to juggle multiple items as part of a set menu. Holly can also "process orders faster than kiosks, while typing can be three times slower," Carpenter says. - OBSERVER

  • Email gray
  • Permalink gray

9. Researchers at the University of Massachusetts, Amherst campus, developed a tool that uses machine learning and LIDAR to identify the solar power potential of roofs and sites. The DeepRoof software, which has a “true positive rate” of 91.1 percent, could tell contractors the solar potential of a location using only an address or GPS coordinates. - PV MAGAZINE

  • Email gray
  • Permalink gray

10. InMarket CMO Cameron Peebles says the biggest movements to watch in advertising technology, or adtech, are AI and blockchain. In an interview with Martech Series, Peebles said that blockchain has the ability to "make data more manageable," which would greatly impact the future of adtech. Moreover, he continued to say that AI would make data "more actionable." The jury is still out over which technology will be more revolutionary for the marketing world, but at the moment it is AI that is proving more applicable, with technologies like chatbots already making consumer conversations more fluid. - MARTECH SERIES

This story first appeared in Inside CMO.

  • Email gray
  • Permalink gray

Written and curated by Beth Duckett in Orange County. Beth is a former reporter for The Arizona Republic who has written for USA Today, Get Out magazine and other publications. Follow her tweets about breaking news and other topics in southern California here.

Editor: Kim Lyons (Pittsburgh-based journalist and managing editor at Inside).

  • Email gray
  • Permalink gray

Subscribe to Inside AI

MORE NEWSLETTERS

A hand-picked selection of products, deals, and ways to save money.

Inside Deals

A hand-picked selection of products, deals, and ways to save money.

DAILY
A hand-picked selection of products, deals, and ways to save money.

Inside Deals

DAILY

SUBSCRIBED!

Share via

Digging into the Trump Presidency, issue by issue

Inside Trump

Digging into the Trump Presidency, issue by issue

DAILY
Digging into the Trump Presidency, issue by issue

Inside Trump

DAILY

SUBSCRIBED!

Share via

News, updates, reviews and analysis of industry and consumer trends in the world of streaming

Inside Streaming

News, updates, reviews and analysis of industry and consumer trends in the world of streaming

DAILY
News, updates, reviews and analysis of industry and consumer trends in the world of streaming

Inside Streaming

DAILY

SUBSCRIBED!

Share via

Facebook's mission is to connect the world's people. Ours is to keep an eye on them.

Inside Facebook

Facebook's mission is to connect the world's people. Ours is to keep an eye on them.

WEEKLY
Facebook's mission is to connect the world's people. Ours is to keep an eye on them.

Inside Facebook

WEEKLY

SUBSCRIBED!

Share via