Algorithmic Accountability
Sources make the news. Reporting that promotes algorithmic accountability depends heavily on seeking out information and perspective from a wide variety of sources. This section provides an introductory guide to the types of sources necessary for high quality reporting on algorithms.
The most ready, accessible sources are those who represent industry. News reporting on AI is “saturated with industry perspectives,” as Mandi Cai and Sachita Nishal of Northwestern University noted in a 2023 presentation on AI literacy for journalism. Industry sources are naturally interested in touting the benefits of new technology and promoting the success of their products.
Negative example: A 2021 Financial Times report on educators using artificial intelligence and machine learning to enable more fair online exams was written “from the perspective of the company selling AI tools,” Princeton University’s Sayash Kapoor and Arvind Narayanan noted. Treating company spokespersons as neutral parties, the article read “more like a PR piece and less like a news story,” illustrating what Kapoor and Narayanan described as the pitfall of “uncritically platforming those with self-interest.”
Be cautious in finding sources via social media.
It probably goes without saying, but contributors’ credibility is not guaranteed by the number of followers they have.
Research backs this up. As Richard Fletcher—currently director of research at the Reuters Institute for the Study of Journalism, University of Oxford—and his colleagues noted in a 2017 article, credibility is better understood in terms of “networked connections to other credible contributors.”
If you find a possible source via social media, vet them outside of social media, for example, by looking up their professional profile online or by reading their published work.
Academic researchers, AI ethicists, and other sources outside industry can offer expert perspectives that are independent of industry interests.
See the section on Informative Frames for more detail.
Positive example: Khari Johnson reported for CalMatters on two public school districts that unsuccessfully introduced AI-powered tools to facilitate teachers’ grading and student communication. Johnson’s report featured a number of independent sources including, for example, a senior technologist at the nonprofit Center for Democracy and Technology; the superintendent of one of the school districts, who advised educators to vet company claims about their AI tools before adopting them; and a leader of the Center for Generative AI and Society at the University of Southern California. The resulting story provided a nuanced, multi-perspective account of the challenges educators face in determining which AI tools are useful and trustworthy.
Negative example: A 2022 New York Times Magazine feature, “A.I. Is Mastering Language. Should We Trust What It Says?” presented substantive concerns expressed by Emily Bender, a professor of linguistics at the University of Washington; but the article dodged the concerns Bender raised by characterizing her and others as “skeptics.” After the story’s publication, Bender posted a detailed response on being “placed into the ‘skeptics’ box” by the Times, which, Bender wrote, “cedes the framing of the debate to the AI boosters.”
