|
|
|
|
Companies Confront Pandemic With Smart Software; Scientists Learn More About Learning
|
|
|
|
|
|
|
|
Welcome back. Airlines, retailers and social-media platforms are building an arsenal of smart-software tools to confront the lasting impact of Covid-19 as the pandemic drags on through the summer. From robotic store managers to pilotless airplanes, the success of these efforts hinges on AI’s ability to be a fast learner, a skill scientists are pushing forward in innovative ways.
|
|
|
|
|
|
|
|
Retail Technology: Intelligent Trend Spotters
|
|
|
|
|
Canadian Tire’s AI system identified a surge in bicycle sales at some stores in March and the company redeployed inventory to meet the demand.
PHOTO: CANADIAN TIRE CORP.
|
|
|
|
Canadian Tire Corp., Canada’s largest retailer, with more than 1,700 locations selling everything from hardware to clothes, leaned on AI to keep up with a pandemic-sparked surge in demand for exercise equipment, patio furniture and other goods coveted by homebound consumers, Vinod Sreeharsha reports.
Tracking sales data. By chance, the company last year rolled out a machine-learning platform that analyzes a variety of internal sales-related data and external environmental data to spot patterns, said Rex Lee, the company's senior vice president and chief information officer.
Managing inventory. Following a sudden upswing in demand for bicycles at its West Coast stores, the company’s AI-enabled systems allowed it to identify the trend quickly and shift inventory to capture more sales, Mr. Lee said.
|
|
|
|
|
|
Facebook: AI-Powered Content Reviews
|
|
|
|
Like most businesses, Facebook Inc. during the pandemic has operated with the majority of employees at home. Unlike most businesses, it has more than a billion of users world-wide. To cope, the social-media giant is using AI to automate a greater share of content moderation and reviews, including natural-language tools that can recognize and flag posts involving suicide or child exploitation, WSJ’s David Uberti reports.
How effective? AI tools helped remove 22.5 million pieces of hate speech posted on the site in the second quarter, up from 9.6 million in the previous quarter, the company says.
What’s next? Facebook says it plans to roll out self-supervised machine-learning technology that can train systems to evaluate images and comments in multiple languages.
|
|
|
|
|
|
|
|
Smart Talk: Self-Learning AI
|
|
|
|
“Instead of saying, how do I change the steering wheel moment to moment, I can just say, ‘I need to go to the store.’ ”
|
|
—Kyunghyun Cho, a professor of computer science and data science at New York University’s Courant Institute of Mathematical Sciences, on the theoretical use of self-taught AI in autonomous vehicles
|
|
|
|
|
|
|
|
Research: Learning How To Learn
|
|
|
AI Gets a Self-Supervised Education
|
|
|
Scientists are pushing the limits of training AI systems to learn by themselves, the way babies do, WSJ’s Steven Rosenbush reports. Through self-supervised learning, as it’s called, AI can train itself without the need for external labels attached to the data, such as being told “this is a cat” to identify other images of cats.
|
|
|
|
How thoughtful? The immediate goal is broader AI that can perform multiple tasks and that could one day lead to artificial general intelligence, or machines with humanlike thinking.
Thinking about what? Current work could lead to the creation of fully autonomous vehicles, virtual tutors for schoolchildren, more effective medical-imaging analysis and the real-time identification of hate speech on Facebook.
|
|
|
|
Airline Industry: AI-Friendly Skies
|
|
|
|
|
Passengers queued at the Delta ticketing counter at Charles de Gaulle airport outside Paris on Thursday as they looked to get ahead of a travel ban on flights from continental Europe to the U.S.
PHOTO: BENOIT TESSIER/REUTERS
|
|
|
|
Autonomous vehicles and facial-recognition scans are in limited use at airports around the world, while other AI capabilities are further off, such as pilotless commercial aircraft, the WSJ reports. The use of smart capabilities is poised to become more widespread as the travel industry contends with the aftermath of Covid-19. Here’s a glimpse at the future of air travel.
Never miss a flight? Smart apps using machine learning to analyze third-party weather information, road traffic data and even security-checkpoint wait times could help passengers catch their flights while minimizing wait time.
Traveling travel bags. Connected suitcases controlled by a traveler’s smartphone will be transporting themselves through the terminal using ultrasonic sensors to avoid objects.
What’s your vector, Victor? Still decades away from being put into practice, the technology that would allow for autonomous commercial jetliners is becoming feasible, though hurdles exist, including training robot pilots to handle contingencies, as when flights are diverted to a different airport.
|
|
|
|
|
|
|
CE controversy U.S. Immigration and Customs Enforcement says it plans to use facial-recognition technology developed by Clearview AI Inc. to investigate child exploitation cases and other online crimes, but not immigration enforcement, an ICE spokesperson told WSJ’s Jared Council. The agency’s use of facial-recognition tools has raised concerns among immigrant advocates. (WSJ)
Faster MRIs. Facebook and NYU Langone Health researchers say a two-year joint project shows that AI can drastically speed up the magnetic-resonance-imaging process, without losing accuracy, by functioning with less data. (WSJ)
U.K. issues data guidance. The U.K. Information Commissioner’s office is warning companies about the risk of violating Europe’s privacy and security regulations by mislabeling personal data in large datasets used for AI technologies. (WSJ)
Drug-tech startup raises funds. Atomwise Inc., which uses AI to help academic and commercial scientists discover new medicines, has raised $123 million to expand and build its own pipeline of experimental drugs. (WSJ)
Startup backer doubles down. Los Altos, Calif.-based Rocketship.vc has raised $100 million to invest in global startups in deals negoatiated online, rather than in person. (WSJ)
Smart tool spots homeless. A new Canadian-made AI-powered program is being used to identify people at risk of becoming chronically homeless, by weighing data points like age, gender, family and shelter history. (CBC.com)
|
|
|
|
|
Amazon is in preliminary talks to invest in U.S. cloud services provider Rackspace Technology Inc., people familiar with the discussion said. (Reuters)
The U.S. is imposing another round of restrictions on China’s Huawei as President Trump renewed accusations that the company’s telecommunications equipment is used for spying. (Associated Press)
House Democrats set a Saturday vote on a bill that would prohibit operational changes to the Postal Service until well after the election and hand $25 billion in additional funding to the agency. (WSJ)
|
|
|
|
|
|
