Alphabet Launches New Initiative to Curb Toxic Comments Online
Google’s parent company is working to stop hateful web chatter in time for the EU elections.
With the world so connected, people are communicating more than ever. Being able to hide behind a username gives anonymity, which also lets individuals write things they probably wouldn’t say in person.
Google parent company Alphabet wants to help stop online toxicity and make the Internet a more welcoming place. Its Jigsaw division works on monitoring online conversations and giving people the tools to control their experience. On Wednesday, it is releasing French-language capabilities to monitor online conversations. The update will come just in time for the 2019 European Union parliamentary elections.
“Today’s world is extremely polarized,” Jigsaw’s head of partnerships and business development Patricia Georgiou told Cheddar. “We really believe if people can have non toxic conversations and share different perspectives, maybe they can understand the point of view of the other person and help potentially depolarize the landscape and society.”
Perspective, a free API, will help publications like Le Monde moderate online comments. Chrome extension Tune will let French speakers decide how much “toxicity” they want to see in web chatter.
Harassment and trolling online plagues many digital companies, ranging from media outlets to social media platforms. Both Facebook and Twitter have acknowledged the issues facing their platform, and have added tools to allow users to report inappropriate behavior. Google itself has had to deal with troubling comments, including predatory ones, on YouTube.
Perspective uses artificial intelligence to learn how people talk through public comments, and then evolves as it learns more about the human lexicon. The API launched in 2017 and was previously only available for English and Spanish sites. Several companies including The New York Times, The Guardian, The Economist, and Wikipedia have adopted its technology.
Tune, which uses the Perspective API, allows people to moderate the “volume” of how much noxious chatter they want to see. Both products are available in English and Spanish.
Language can be tricky. For example, the API had to learn that saying “break a leg” is not a violent phrase, but “I want to break your leg” could be a threat. There’s also the question if AI alone can be trusted, since it may be biased by its creators and the information it is seeing. Many platforms, including Google, have increased their reliance on human moderators. The company said it would hire more than 10,000 people to look at questionable material.
While these are all valid concerns, Georgiou pointed out there’s value in working on all methods to improve the online experience. AI can only get better with more people using it and giving feedback. And, if Jigsaw’s actions give other companies ways to make their online comments better, it’s a positive step in making the web a better place.
“If the internet is full of toxicity, people are not going to go to the internet anymore,” Georgiou said. “They’re going to stay offline, or they’re not going to engage as much. It’s really important for the freedom of the internet that it stays a civil space.”