Invited Keynotes

Lisa Gibbs, Associated Press, “The Augmented and Automated Newsroom”

Four years after the Associated Press began automating corporate earnings stories, the agency has made automation and AI an integral strategic initiative. Ongoing projects include real-time video transcription, a news summarization tool and a platform that uses machine learning to verify video posted on social media. AP’s experiences illustrate how news organizations are using automation and AI to break news and assist beat reporting, produce content more efficiently, prevent misinformation and personalize the news for their audiences. This discussion will also explore the resources, workflow, and ethical frameworks needed to take advantage of these technologies and ensure they are deployed in the public interest.

Lisa Gibbs is the director of news partnerships at The Associated Press and the  newsroom’s point person on AP’s automation and artificial intelligence strategy group. This includes working with industry leaders and emerging startups to identify smart applications of automation/AI in media, project development, workflow integration and training. Current projects span a range of technologies, such as image recognition, video transcription and using machine learning to verify information plus surface breaking news stories and trending topics. A native of Miami, Florida, she joined AP in 2014 as global business editor from Money (Time Inc.) and is a previous executive business editor of the Miami Herald. In 2016, she was named Business Journalist of the Year by TalkingBizNews.

 

Brian Hamman, New York Times, “From Designing Boxes to Designing Algorithms: How Programming the News Has Evolved at the New York Times”

Every pixel on a news site represents a choice. Those pixels can be used to inform or delight you, to sell you jewelry, or encourage you to subscribe. Each of those goals, in turn, maps back to different departments within a news organization with competing goals. So how do we decide what to show you? Not long ago we solved this through design. Our screens were divided up into a tapestry of stakeholders with space carved out for advertising, marketing and the different news desks. A complicated negotiation set out the initial layout and then each group could fill in their boxes independently. As our screens and attention have shrunk to the confines of a single scrolling experience on a phone, we can’t rely on design alone. Instead, our editors, advertisers, marketers and engineers must collaborate to encode their intentions into what to show each user in each moment by marrying user experience testing outcomes with design principles, all ruled by the editorial judgment signature to The New York Times. This talk will look at both the New York Times Homescreen and Story page as two case studies of how moving to a single-track experience changed the way we work.

Brian Hamman is a Vice President of Engineering at the New York Times, where he is responsible for all user facing technology across the web and mobile apps. Brian has spent over a decade leading interactive teams in the newsroom and engineering teams in the product organization. He has seen the Times go from separate web and print buildings into one; separate web and print newsrooms into one; and, most recently, separate web and mobile homepages into one.

 

Yphtach Lelkes, University of Pennsylvania, “Projecting Confidence: How the Probabilistic Horse Race Confuses and Demobilizes the Public”

Horse race coverage in American elections has shifted focus from late-breaking polls to sophisticated forecasts that emphasize candidates’ probability of victory. While these probabilistic forecasts may be more accurate than any individual poll, humans are notoriously bad at understanding probabilities. This talk will seek to understand the implications for probabilistic forecasts for political behavior. After demonstrating the prominence of probabilistic forecasts in election coverage, experiments will be presented to show that the public has difficulty reasoning about the probability of a candidate’s victory. Critically, when one candidate is ahead, win-probabilities convey substantially more confidence that she will win compared to vote share estimates. Even more importantly, this talk will show that these impressions of probabilistic forecasts cause people not to vote in a behavioral game that simulates elections. The magnitude of these findings suggests that probabilistic horse race coverage can confuse and demobilize the public, suggesting implications for how journalists use and present forecasts in the media.

Yphtach Lelkes is an Assistant Professor at  the Annenberg School for Communication, University of Pennsylvania. His main focus is on the role of the political information environment in structuring attitudes, with a special interest in the proliferation of digital media. He has also written extensively on affective polarization, or the increasing hostility between Democrats and Republicans. Before joining the University of Pennsylvania, he was faculty at the Amsterdam School of Communication Research. He received his Ph.D. from Stanford University. His research has appeared or is forthcoming in various journals across disciplines, including the American Journal of Political Science, PNAS, the Journal of Politics, the Annual Review of Political Science, and the Journal of Personality and Social Psychology, and has been covered in many news outlets, such as the New York Times, the Washington Post, and the Atlantic.