Welcome to Tuesday's Inside AI! In case you missed it, our latest premium content has included a deep dive into the growth of autonomous vehicles, a roundup of the top AI stories from September, and an AI Masterclass on how Alexa is becoming more intelligent.
Premium subscribers get all of my content and don't have to look at ads. We offer a $10/month plan or the option to be billed $100 annually, and are currently offering a 14-day free trial of premium membership! Click here to sign up!
|
|
|
Clearview CEO Hoan Ton-That
Facial recognition startup Clearview AI will add new compliance features and training to make sure law enforcement uses the technology ethically. One feature would require officers to enter case numbers and crime for each suspect search for better auditing, CEO Hoan Ton-That said during today's Wall Street Journal Tech Live conference.
More:
- Ton-That said individual police departments will still have to enforce rules and monitor police officers who use the software, which matches suspect's photos to a Clearview database made up of public photos scraped from the internet.
- The company has faced scrutiny for what critics say is a violation of privacy laws by using people's photos without their consent. Facebook, Twitter, and LinkedIn have all sent cease-and-desist orders to prevent Clearview from scraping their sites.
- All data that Clearview collects is publicly available, Ton-That said at the conference. He emphasized that the startup no longer provides database access to private companies. He said the technology "is here to stay.”
- About 2,400 law-enforcement agencies, including U.S. Immigration and Customs Enforcement, use Clearview's technology. The U.S. Department of Homeland Security uses Clearview's facial recognition to identify children who are victims of pedophiles, for example. It could still be used to search for undocumented immigrants or whatever the agency chooses to do, Ton-That acknowledged.
WALL STREET JOURNAL
|
|
Adobe released a Photoshop update that comes with the Sky Replacement tool and other AI-powered features, including improved AI edge selection and Neural Filters, which are image editing tools for improving portraits.
More:
- Adobe previewed its Sensei-powered AI Sky Replacement tool for Photoshop, which will enhance background skies in photos, last month. The tool uses AI to separate the sky from the foreground, creating more dramatic scenes.
- The newly announced Neural Filters were created with Nvidia. There are currently eight beta filters, which include skin smoothing, style transfer for applying art styles to image, and smart portrait for altering a subject's age, pose, and facial expressions. They can also colorize black-and-white photos and increase the resolution of certain parts of images.
- Maria Yap, Adobe’s vice president of digital imaging, called Photoshop "the world’s most advanced AI application” through the power of generative adversarial networks, which involves both local and cloud processing.
- Other new features simplify the editing workflow, create patterns, and improve search, help, and contextual actions. All work on devices running Photoshop 22.0.
THE VERGE
|
|
Deep Dive: Why companies fail on AI
A global survey of 3,000 managers, scholars, and executives found that most companies are developing AI but have yet to see significant financial benefits. Only 10% said they've seen a significant financial payout from their investment, according to a new report. The news comes as corporate spending on AI is expected to surpass $50b in 2020, up from $37.5b last year, according to IDC.
What the report found:
- 57% of companies have deployed or piloted their own AI projects, up from 44% in 2018, according to MIT and BCG.
- Six out of 10 have an AI strategy this year, up from four out of 10 in 2018...
A deeper look:
- While there have been more investments in tech and data science, the impact line hasn't changed much.
- Companies generally overspend on AI technology and data scientists but fail to change their business processes that could benefit from AI, which reduces their ability to deliver on AI projects...
|
|
Microsoft is updating its AI that automatically filters out background noise during Teams meetings. The latest AI noise suppression feature, due out next month, is expected to outperform the current algorithm, according to Microsoft.
More:
- The feature works by analyzing a person's audio feed and using trained neural networks to filter out background noise but retain speech.
- Microsoft described it as an update to its existing noise suppression software, which gives people the choice over how much suppression they want through a new "High" setting. Users will be able to switch noise suppression on and off from their Settings menu.
- Zoom also has background noise suppression and low, medium, and high settings.
USA TODAY
|
|
A satellite orbiting the Earth uses AI to automatically identify cloud cover. Intel revealed new details about the launch of PhiSat-1, a small experimental satellite that was launched into sun-synchronous orbit last month.
More:
- PhiSat-1 uses a hyperspectral-thermal camera and inbuilt chip, Intel's Movidius Myriad 2 Vision Processing Unit, to accelerate data processing and minimize data movement. It's the first time the VPU, usually found in consumer devices, has been in space.
- The VPU can handle large amounts of data. It runs local AI software that can identify and remove images where the Earth is obscured, making sure only visible images are sent to scientists. This results in a bandwidth savings of up to 30%, transmitting only the most useful data.
- Startup Ubotica developed the AI software. Intel also worked with the European Space Agency, which along with Arianespace launched the satellite from a rocket along with 45 small satellites for research purposes on Sept. 2.
ZDNET
|
|
|
QUICK HITS
- Google AI launched open-source browser-based toolset “rǝ,” a 3D virtual system that shows how cities looked in the past.
- Role-playing video game "Cyberpunk 2077" will rely on AI to lip sync its dialogue in multiple languages.
- More than 75% of Russians support using facial recognition to identify who came in contact with people infected with coronavirus, according to a study by the Russian Venture Company and the Institute of National Projects.
- A new ML framework could make robots safer in crowds.
- IBM announced revenue of $17.6b for Q3 2020 with a YoY decrease of 2.6%.
- Chargebee is organizing the largest tech summit for change-makers of the SaaS space, featuring champions from HubSpot, Drift, Dropbox, Slack, etc. Join here.*
*This is a sponsored post.
|
|
Tweet of the Day: Microsoft built a machine learning model to track plastic flowing through rivers using autonomous 'interceptor' boats.

|
|
|
|
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.
|
|