Initiative selects winners that use AI to combat media challenges

بواسطةTaylor Mulcahey
Mar 14, 2019 في Specialized Topics

The Ethics and Governance in AI Initiative has been funding innovation for years, but they were tired of seeing the same subjects apply. So, last September they launched their AI and the News Challenge, and following an open call for proposals, this week they announced the winning projects.

The winners are seven organizations that span the globe, from Chequeado in Argentina to Tattle Civic Technologies in India, each of which has designed a project to tackle a challenge in artificial intelligence (AI) and its effect on the news industry.

“This is exactly how we hoped it would turn out,” said Tim Hwang, who leads the Ethics and Governance in AI Initiative, which is a joint project of the MIT Media Lab and the Harvard’s Berkman Klein Center for Internet & Society.

“We went into it with a real eye on broadening the pool of people, so I think we're really happy with a set that makes people say, ‘oh, those are a bunch of interesting projects that are in a bunch of different places on the map.’"

After the 2016 election in the U.S., Hwang and his team noticed a spike in the number of people interested in the role automation plays in disseminating information online, especially mis/disinformation. But many people, although interested in the topic, still don’t fully understand AI in the technical sense.

Hwang defines AI broadly, as an effort to make machines smart. Today, this manifests itself in a computer science field known as machine learning, in which machines are exposed to large amounts of data in order to train them.

“If you want to train a machine to recognize a cat in a photo, you basically show it lots of photos of cats until it learns what a cat looks like,” Hwang explained.

How this technology influences the news industry, and how the news industry, in turn, influences this technology, has become a major focus for the team in recent years, inspiring the AI in News Open Challenge. They invited proposals that addressed four main challenges: governing the platforms, stopping bad actors, empowering journalism and reimagining AI and news.

What they expected would yield around 100 applications ended up inspiring more than 500. To determine the winners, Hwang and his team assembled reviewers from journalism, academia, technology and business. They ultimately selected seven projects to split the US$750,000 from donors such as the John S. and James L. Knight Foundation, Omidyar Network, LinkedIn co-founder Reid Hoffman, and the William and Flora Hewlett Foundation.

"For us, it's really amazing to get this support because it allows us to put our time into something that we believe is crucial, and no one else is doing it,” said Laura Zommer, executive director and editor in chief of Chequeado, a news organization based in Argentina, and one of the winners.

Chequeado won US$75,000 to launch a series of investigations on the ethics of AI and the algorithms that train the machines, and to produce a guide for other reporters covering these topics.

They developed the idea when Chequeado’s innovation team was creating their own fact-checking bot, Chequeabot, in an attempt to speed up the process. When they came across their own ethical challenges, they looked around the region for information and investigations on the ethics of algorithms.

"We imagine there are others developing technology in the region that have the same challenges or questions that we have,” said Zommer. “Why don't we investigate these issues in the region, and then publish and share it?"

Zommer and her team will investigate this question all around the region, not just in Argentina. To do so, they plan to partner with other journalists and newsrooms, and they will share the results in a way that will be easy to understand and attractive to readers that have very little knowledge of AI in general.

Another winner, The MuckRock Foundation, is working on a very different issue related to news. The nonprofit, based in the U.S., began as a way to help journalists and researchers gain access to large sets of data and documents. What they soon realized, however, is that access is just the beginning.

Michael Morisy, the chief executive of MuckRock, explained that he and his team have developed a project that is not meant to replace journalists, but will “help them scale their work in different ways.”

MuckRock’s winning project, Sidekick, won US$150,000 to use crowdsourcing and machine learning to go through tens of thousands of documents to find meaningful data. Morisy said that there is a lot of excitement from users to be part of the newsgathering business. Sidekick will leverage this enthusiasm, putting users to work sifting through documents in a way that will ultimately train machines to do so without much human interaction at all.

If journalists no longer have to spend significant amounts of time looking through documents themselves, it frees them up to focus on other, more meaningful work. This is the Sidekick project’s goal.

Other winning projects include a reporting series from the Seattle Times, automatic analysis of government records from machine learning company Legal Robot, automatically analyzing and detecting deepfake videos from the Rochester Institute of Technology, and more.

“They end up falling into categories,” said Hwang of the winners. “But we're happy that the set of participants come from very different places and very different backgrounds.”

Readers interested in learning more about the projects and their evolution can follow along on the Ethics and Governance in AI Initiative’s blog.

Main image CC-licensed by Unsplash via Joshua Sortino.