Generative AI and Journalism: From Production to Perception

This new book, Generative AI and Journalism: From Production to Perception, is forthcoming via Routledge in 2026. It is co-authored by T. J. Thomson, Ryan. J. Thomas, and Hannes Cools.

Generative AI has been one of the most-searched-for terms in popular online search engines in recent years and people across the world were—and are—curious about what generative AI is, how it works, its challenges and risks, and its potentials.

Within journalism, this technology is the latest iteration in a long trend of automating news and journalism production, editing, and presentation processes and has been deployed for diverse use cases, from invisible, behind-the-scenes tasks to creating content that audiences view and interact with.

This book takes a global look at generative AI and how it is affecting journalism, those who practice it, and those who consume it. The book represents the first of its kind in the way it pays systematic attention to AI-mediated outputs, industry practices, and audience experiences while also building on empirical data collected across three continents over five years.

---

Advance praise

“In their new book, Thomson, Thomas, and Cools provide a provocative, insightful, and critical examination of the implications of generative artificial intelligence (AI) for journalism. There may be no more disruptive technology in the 21st century than AI, and Generative AI and Journalism delivers much-needed guidance on this topic. With its historical-analytical approach, the book is a must-read for any serious scholar, student, or practitioner of journalism.”

Distinguished Professor John V. Pavlik, School of Communication and Information, Rutgers University

(More endorsements are on their way and will be added here once they are available.)

---

Book Contents

Chapter 1: Historicising GAI’s rise in Journalism

This chapter traces journalism’s evolving relationship with technology in order to situate generative AI (GAI) within a 185‑year history of innovation, anxiety, and adaptation. Rejecting simple utopian or dystopian narratives, it argues that technological change in journalism follows cyclical patterns of hype, experimentation, normalization, and transformation rather than linear progress. Using a historical-analytical approach, the chapter synthesizes scholarship and industry cases across four overlapping phases: pre‑digital automation (1839–1990), digital emergence (1990–2000), datafication (2000–2015), and the pre‑GAI algorithmic turn (2015–2022). It shows how successive technologies – from telegraphy, photography, radio, and early computing to the web, social media platforms, data journalism, machine learning, and large language models – have repeatedly reorganized journalistic labor, formats, business models, and power relations. The chapter identifies enduring tensions around speed versus verification, authenticity and trust, labor displacement and skill reconfiguration, economic imperatives versus professional resistance, and the gradual normalization of disruptive tools. By placing GAI within this longer trajectory, the chapter provides a framework for assessing how contemporary imaginaries, governance debates, and newsroom experiments with generative systems both echo historical precedents and introduce genuinely novel challenges for journalistic independence, accountability, and public trust in democratic societies and media ecosystems.

Chapter 2: Social and Institutional Drivers of Generative AI in Journalism 

This chapter examines how generative artificial intelligence (GAI) is being incorporated into journalism by applying the hierarchy of influences model to identify the broader forces shaping its adoption. Moving beyond technology-centric accounts, it argues that GAI is embedded within a complex environment of social system- and institutional-level dynamics. At the social system level, economic pressures on news organizations, cultural narratives framing AI as inevitable or transformative, evolving legal and regulatory frameworks, and changes in the digital information ecosystem all shape incentives for adoption and perceptions of risk. At the institutional level, relationships with technology companies, platforms, governments, professional associations, and audiences influence how GAI tools are accessed, implemented, and evaluated. The chapter highlights how these interacting forces structure not only whether GAI is adopted, but also how it is used and legitimized within journalism. It concludes that understanding GAI’s role in journalism requires situating technological change within broader contexts of power, inequality, and professional norms, setting up a complementary analysis of newsroom-level dynamics in the following chapter. 

Chapter 3: GAI in the Newsroom: Organisations, Practices, and People 

This chapter examines how generative artificial intelligence (GAI) is incorporated into journalism at the organisational, routines, and individual levels of the hierarchy of influences model. Extending the previous chapter’s analysis of social and institutional forces, it argues that the role of GAI in journalism is shaped through processes within news organisations and the everyday practices of journalists. At the organisational level, ownership structures, business models, leadership priorities, and newsroom cultures influence how AI is framed, resourced, and governed. At the routines level, GAI is typically integrated incrementally into established workflows, supporting tasks such as newsgathering, production, verification, and distribution. At the individual level, journalists’ skills, professional values, motivations, and perceptions of risk and opportunity can shape how these tools are interpreted and used in practice. Together, these interacting forces mediate how external pressures are translated into newsroom strategies and journalistic work. The chapter demonstrates that GAI adoption is not a uniform or deterministic process, but a negotiated outcome shaped by organisational decision-making, routine practice, and individual agency, in addition to forces beyond the newsroom. 

Chapter 4: Evaluating GAI content in journalism from two perspectives 

This chapter focuses on GAI content, particularly in visual and multi-modal form, and some of the characteristics of this content and accompanying antecedents for these attributes. The chapter begins by discussing the rhetoric surrounding GAI in public discourse and how GAI companies work to cultivate an image of GAI as magic, unknowable, and transformative. The chapter continues by discussing evaluation criteria in journalism and training data and their importance to GAI outputs. Next, the chapter discusses GAI biases before exploring how GAI systems evaluate and remediate human-created journalistic content in sensitive contexts. Overall, the chapter uses evaluation as a framework to better understand the limits, biases, and challenges of using AI to generate journalistic content or to describe and remediate it. In articulating these issues, the chapter also points to ways that the technology’s infrastructure, data training materials, and governance can be improved for the benefit of scholars and practitioners alike while concurrently identifying potential paths forward for scholarship and empirical development. 

Chapter 5: Learning from audiences about GAI in journalism 

This chapter examines how audiences perceive and experience GAI in journalism, arguing that public expectations must guide responsible adoption. It distinguishes between two intertwined currents: the production current, where GAI systems enter reporting, editing, and distribution workflows, and the discovery current, where search overviews, and recommendation interfaces increasingly mediate how people encounter news. Drawing on pioneering studies, the chapter identifies a persistent practices–publics gap between how news organisations implement GAI and what audiences understand, and what their information needs are. It maps three core clusters: awareness and literacy (how people recognise and interpret GAI involvement in news), diversity and inequality (how age, and education shape expectations and comfort), and transparency and governance (what kinds of disclosures and oversight are seen as meaningful). Across these domains, audiences express nuanced, task‑specific views: broad acceptance of back‑end assistance such as spelling checks, translation, and summarisation, but deep scepticism toward fully automated articles and artificial anchors. The chapter concludes that GAI strategies in journalism should treat audiences as central stakeholders, aligning use‑case choices, transparency design, and accountability mechanisms with articulated public preferences in order to sustain trust and democratic legitimacy. 

Chapter 6: Theoretical takeaways and future forecasts for the journalistic and scholarly fields 

This chapter begins by reflecting on the first five years of widespread GAI access and on how the technology was perceived and used over this timeframe, broadly, and in journalism, in particular. The chapter situates these past five years within a larger history of research on and practice enabled through technological innovation and muses about which challenges are unique to GAI and which are more longstanding. The chapter continues by summarising key arguments made over the book’s first five chapters. Next, it proposes a future research agenda centred on three key research areas (journalists’ agency and control, audiences’ calibrated trust, and content distribution and authenticity) and ends by detailing three future forecasting scenarios about how GAI might evolve in the next decade. Grounded in rigorous analysis yet written as compelling stories, these scenarios bring potential futures to life in ways that data alone cannot. They offer no consensus view or single forecast; instead, they invite decision-makers to project themselves into a range of challenging but plausible future contexts, using that perspective as a foundation for sharper, more resilient choices today. 

Meet the co-authors: