The 2nd BAIRAL Research Meeting for 2020 Report on “Clarifying Gender Bias in Newsrooms: A collaborative project ‘AIJO’ with eight companies around the world”

Akira Tanaka (Research Assistant of the B’AI Global Forum)

Date: January 27, 2021
Location: Online (Zoom)
Moderator: Akira Tanaka

On January 27, 2021, “BAIRAL,” which is a study session organized by the research assistants of the B’AI Global Forum, invited Yosuke Suzuki (Nikkei America, Inc.) and Issei Mori (University of California, Santa Cruz, an intern at Nikkei America, Inc.), who worked on the planning and development of the “AIJO” project, and invited them to introduce their practice to clarify gender bias in newsrooms by using AI.


In order to encourage the application of artificial intelligence in media organizations, Google has been conducting “JournalismAI” with the London School of Economics and Political Science (LSE) in the UK as the office. Then, as a part of the project, eight news organizations including Nikkei Inc. launched an international project, AIJO, which aims to investigate biases in newsrooms. It aims to ameliorate the problem of various unconscious biases of newsmakers (e.g., gender, age, race) being reflected in headline images, interviewees and articles.


In the first half of the session, they explained what they were able to do and what they were not able to do technically in their actual efforts. They were faced with the difficulty of natural language processing of the text due to the linguistic diversity of the participating companies. Therefore, they decided to conduct the analysis first using headline images, which are often prominently displayed due to their proliferation on social media. During the presentation, they also demonstrated face detection and gender identification using machine learning and showed how they artificially filtered the results. According to them, the analysis tool is available for anyone to use through Github.


Around thirty people, including researchers and journalists, attended the workshop, and some of them asked questions about how guest speakers felt about the actual process of conducting the analysis. We discussed not only the technical issues, but also the dilemmas that arose in developing the analytical tools (for example, the appropriateness of assigning gender to men and women), the issues of gender inequality that the results of the analysis revealed, and how to return the results of the analysis to the news media. Whereas the previous BAIRAL research meeting critically discussed the social significance of artificial intelligence, this time we got a glimpse of the methodological frontier using artificial intelligence to solve social problems.