Inside | Real news, curated by real humans
Inside AI

Inside AI (Jan 9th, 2020)

1. Several federal lawmakers on a House panel criticized Facebook's recent ban on AI-edited deepfakes, saying the new policy doesn't go far enough to fight the manipulated videos. Facebook's global policy chief, Monika Bickert, addressed questions during Wednesday's hearing of the House Energy and Commerce Subcommittee on Consumer Protection and Commerce. Chairwoman and U.S. Rep. Jan Schakowsky (D-Ill.) started out by calling Facebook's policy "wholly inadequate," noting that a deepfaked video featuring House Speaker Nancy Pelosi appearing drunk has been viewed "millions of times." Under its new policy, Facebook would not ban that video, nor others that are considered "parody or satire." However, Bickert said during testimony that the Pelosi video would still be subject to other company policies addressing misinformation. Facebook has said that only videos generated by AI depicting people saying fictional things would be taken down. - THE HILL

2. Rather than using images of paid models, many companies and organizations will turn to machine-generated images of human faces in the near future (if they haven't already), The Washington Post reports. The rise in AI-generated imagery can accommodate companies looking to promote diversity, since they can be generated quickly and don't involve actual living workers. In the meantime, startups like Argentina-based Icons8 sell subscription packages for fake photos that can be filtered by ethnicity, age, and emotion. However, the Post notes that there are caveats. For example, the fake images are generated by GANs trained on thousands of actual models, who often weren't told ahead of time that they would be included in such a project. AIs have been known to churn out some "nightmarish"-looking faces with mutations, or what Icons8 co-founder Ivan Braun has dubbed "monsters." - WAPO

3. A new report from Juniper Research predicts that more than 800 million devices will have facial recognition hardware by the year 2024. Perhaps unsurprisingly, the hardware is expected to be the fastest-growing type of biometric software on smartphones, thanks to the rise of tech like face ID on newer iPhones. Last year, an estimated 96 million devices had facial recognition. On the software side, an estimated 1.3 billion devices are expected to have that capability by 2024, according to the report, which is titled "Mobile Payment Authentication: Biometrics, Regulation & Forecasts 2019-2024." - HELP NET SECURITY

4. New Zealand startup Soul Machines has raised $40 million to boost its creation of digital avatars. The startup's conversational assistants can animate autonomously and evolve over time, responding to people based on their previous interactions using biometrics and AI. The Auckland-based company uses a cloud-based automation studio that allows brands to "prototype digital humanoids across multiple devices," VentureBeat reports. Meanwhile, Soul Machines’ "Soul Engine" platform can simulate cognitive processes like sensing and attention control. The Series B round was led by Singaporean holding company Temasek, with participation from Lakestar and existing investors Horizons Ventures, University of Auckland Inventors Fund, Salesforce Ventures, and others. - VENTUREBEAT

5. Toyota announced plans for a "city of the future" development, where researchers will test out AI-based technologies such as autonomous vehicles and robot assistants. Announced during this week's Consumer Electronics Show (CES), the "Woven City" project is slated to break ground in 2021 at the bottom of Mount Fuji near Tokyo. It would house 2,000 people as a "living lab" for researchers, scientists and engineers to test emerging tech "in both the virtual and the physical world," Toyota's CEO Akio Toyoda said during CES, adding that "We want to turn artificial intelligence into intelligence amplified." The project is described as fully sustainable, powered only by rooftop solar panels and hydrogen fuel cells. It's a collaboration between Toyota and the Danish architecture firm Bjarke Ingels Group. - CNN

6. Warner Bros. will use AI algorithms to guide how an already-greenlit film is developed over time, according to The Verge. The film studio hired the Los Angeles startup Cinelytic, which launched a machine learning project management system last year that uses AI techniques to predict a film's overall success, among other things. Warner Bros. will use the technology “to guide decision-making at the greenlight stage,” which could include factors like what actors to hire, marketing costs, and when to release a film. Cinelytic has developed and tested out the platform since 2017. - THE VERGE

7. A deepfake video featuring Jim Meskimen doing impressions of Hollywood actors is making its rounds across the internet. Digital artist Sham00k - who claims to create the "best worst deep fakes on the internet - used AI to replace Meskimen's face with that of various actors while he mimics their voices. The video is one of the better deepfakes out there, with great impressions of Robin Williams, Christopher Walken, and George W. Bush. - STUDIO DAILY

Written and curated by Beth Duckett, a former reporter for The Arizona Republic who wrote a book about the solar industry and frequently covers hobby and commercial drones. You can follow her tweets about the latest news in artificial intelligence here.

Edited by Sheena Vasani, Inside Dev editor.

Subscribe to Inside AI