7 November 2014 in Digital Journalism
Every day automated algorithms make decisions that can amplify the power of businesses and governments. Yet as algorithms come to regulate more aspects of our lives, the contours of their power can remain difficult to grasp. This paper studies the notion of algorithmic accountability reporting as a mechanism for elucidating and articulating the power structures, biases, and influences that computational artifacts exercise in society. A framework for algorithmic power based on autonomous decision-making is proffered and motivates specific questions about algorithmic influence. Five cases of algorithmic accountability reporting involving the use of reverse engineering methods in journalism are then studied and analyzed to provide insight into the method and its application in a journalism context. The applicability of transparency policies for algorithms is discussed alongside challenges to implementing algorithmic accountability as a broadly viable investigative method.
27 July 2016 in Digital Journalism
The growing use of difficult-to-parse algorithmic systems in the production of news, from algorithmic curation to automated writing and news bots, problematizes the normative turn toward transparency as a key tenet of journalism ethics. Pragmatic guidelines that facilitate algorithmic transparency are needed. This research presents a focus group study that engaged 50 participants across the news media and academia to discuss case studies of algorithms in news production and elucidate factors that are amenable to disclosure. Results indicate numerous opportunities to disclose information about an algorithmic system across layers such as the data, model, inference, and interface. Findings underscore the deeply entwined roles of human actors in such systems as well as challenges to adoption of algorithmic transparency including the dearth of incentives for organizations and the concern for overwhelming end-users with a surfeit of transparency information.
5 June 2020 in New Media & Society
New media synthesis technologies are rapidly advancing and becoming more accessible, allowing users to make video and audio clips (i.e. deepfakes) of individuals doing and saying things they never did or said. Deepfakes have significant implications for the integrity of many social domains including that of elections. Focusing on the 2020 US presidential election and using an anticipatory approach, this article examines the ethical issues raised by deepfakes and discusses strategies for addressing these issues. Eight hypothetical scenarios are developed and used as the basis for this analysis, which identifies harms to voters who view deepfakes, candidates and campaigns that are the subjects of deepfakes, and threats to electoral integrity. Four potential forms of intervention are discussed with respect to multi-stakeholder responsibility for addressing harms, including education and media literacy, subject defense, verification, and publicity moderation.
15 September 2015 in Digital Journalism
So-called “robot” journalism represents a shift towards the automation of journalistic tasks related to news reporting, writing, curation, and even data analysis. In this paper, we consider the extension of robot journalism to the domain of social platforms and study the use of “news bots”—automated accounts that participate in news and information dissemination on social networks. Such bots present an intriguing development opportunity for news organizations and journalists. In particular, we analyze a sample of existing news bot accounts on Twitter to understand how news bots are currently being used and to examine how using automation and algorithms may change the modern media environment. Based on our analysis, we propose a typology of news bots in the form of a design and editorial decision space that can guide designers in defining the intent, utility, and functionality of future bots. The proposed design space highlights the limits of news bots (e.g., automated commentary and opinion, algorithmic transparency and accountability) and areas where news bots may enable innovation, such as niche and local news.
9 March 2020 in Digital Journalism
Computational news discovery (CND) is a particular application area within computational journalism related to the use of algorithms to orient editorial attention to potentially newsworthy events or information prior to publication. Previous work in this area has been concentrated on prototyping CND tools, which can, for instance, send alerts and leads to journalists about social media events, documents of interest, or salient patterns in streams of data. This article describes a qualitative interview study of journalists as they incorporate CND tools into their practices. Findings provide insights into how CND tools interact with the internal attention economy and sociotechnical gatekeeping processes of the newsroom and how future CND tools might better align with necessary journalistic evaluations of newsworthiness and quality, while ensuring configurability, human agency, and flexible applicability to a wide range of use cases. These findings begin to outline a conceptual framework that can help guide the effective design of future CND tools.