# Refusig AI: How Journalists Avoid Falling for AI's Hype **Tomas Dodds, 2024** [Available here](https://www.oii.ox.ac.uk/news-events/videos/refusing-ai-how-journalists-avoid-falling-for-ais-hype/). ## In short Following a year-long series of interviews with journalists and newsworkers, this presentation draws from the concept of technological refusal to understand how journalists are anticipating and appropriating AI systems in their newsrooms. Refusal can take two forms. On the one hand, it can signal the withdrawal from engaging with certain technologies. “I will not work with AI.” On the other, refusal does not have to be a conversation-ender. As my research shows, journalists are, on an informed basis, refusing to engage with artificial intelligence systems on the terms set by third-party stakeholders. In doing so, journalists are opening the door to have a different conversation: A conversation about how to engage with these systems differently. Journalists are imagining new ways of being with AI systems, auditing these technologies not as individuals but as a well-informed collective. In doing so, journalists’ refusal is an act of design that informs how these technologies are being appropriated in their newsrooms. In this talk, I present results from different research projects that identify not only shortcomings inside newsrooms in the form of knowledge silos and low media literacy but also strategies aimed at generating new opportunities to use refusal as a gatekeeping strategy to improve their work. ## Notas - Ongoing investigation. - **Part 1: Guidelines** - News outlets develop guidelines for the use of AI - Important because of two reasons: - They are aspirational. They signal values and ideas - A symptom of the pressure we are putting on newsrooms. We ask journalists very quickly for their opinion an AI. Some people disagree with this point. - Analysed 37 guidelines from 17 countries. Thematic analysis. - Concentrated in North America and Europe. These are the guidelines they could find at the time. - They did not study how much they are being implemented. - ![[Evolution_IA_Journalism_Guidelines.png|500]]![[AI_Journalism_Guidelines_Topics_Region.png|500]] - **Isomorphism** - Institutional isomorphism, leading to convergence and, ultimately, homogeneity over time. - This phenomenon can be attributed to the rapid emergence of generative AI. - In a neo-institutional model, alignment with the institutional environment is crucial for organisational survival. - These coercive pressures may have contributed to the similatiries observed in these guidelines. The normative professionalisation of journalism has influenced these guidelines to adhere to journalistic norms and values. - **Part 2: Responsible AI** - How are journalists understanding responsable AI? - Journalists know what trust, responsibility, transparency is. They don't need help understanding these concepts. What they don't now is what AI is. - Editors didn't know how journalists were using AI systems. - Some were even using in "on the edge" of legality (e.g. use of data systems). - These took the researchers to topic of knowledge silos - **Knowledge silos** - Organisational settings can both enable and constrain knowledge-sharing. Silos hinder adoption of responsible AI practices in journalism. They create communication gaps between editors, journalists, IT staff, and other stakeholders. They lead to misunderstandings, lack of shared understanding of AI (different language), and other stakeholders. - Types: horizontal (journalists-data scientists), vertical (editor-IT staff), internal (legal-editorial), and external (news organisation-industry). - **Breaking silos** - To address them: understanding silo types. - What to do? - Knowledge-sharing programs can address internal silos. - Industry-wide initiatives are needed to address external silos. - Breaking silos is essential for responsible AI adoption in newsrooms. - **Controlled change** - "We want to be careful. Fast is second" - **Part 3: Refusal** - Hype avoidance - Diversification of professional roles > New ways of covering tech > Hype avoidance - Refusal: informed decision to opt out of a system (AI, Blockchain, etc.) - Individual vs. Group: refusal can happen at both levels. - Refusal as opportunity: when we refuse, we open doors for alternative solutions. - New project: - Cognitive-offloading - In journalistic school, writing headlines was a big skill. Journalists are using ChatGPT to A/B test headlines. --- ### Relacionado - [AI in the News: Reshaping the Information Ecosystem?](https://www.oii.ox.ac.uk/news-events/reports/ai-in-the-news-reshaping-the-information-ecosystem/)