Author – Stefano Sorrentino
For most of the history of modern journalism, “reading the news” has been understood as an active, often solitary, one to one encounter between a reader and an author’s text. Even in its most passive forms, such as scrolling through a feed and browsing through headlines in the morning, reading implied a degree of direct contact with journalistic output, hence with the words a journalist chose, the framing an editor approved and the story a newsroom decided to publish.
This direct contact can be seen as the last link in the chain of editorial responsibility that over time has given journalism its epistemic legitimacy, foundational to its democratic and civic function.
With the integration of AI in reading experiences, this link is now being questioned and challenged.
From Reading to Asking
Across the journalism industry this paradigm is changing. Beyond news aggregators and external “AI overviews” provided by search engines, even major outlets, from the Washington Post to TIME Magazine, have begun deploying conversational AI interfaces that allow readers to “ask” questions about the news rather than read it (Simon, 2024). The idea is to allow users to obtain an immediate synthesised answer, rather than navigating a complex article or searching through an archive, by simply posing a question in natural language.
However, something important is lost in that interactional pattern. When a reader asks a question to an AI assistant and receives a fluent, confident paragraph in response, they are no longer engaging with journalism but rather with a model’s mediated summary of human-written journalism. What is finally proposed to the reader is a representation of a representation, filtered through parameters that have little to do with the editorial values of any specific newsroom. In this way the journalist’s voice, the deliberate choice of emphasis and the institutional accountability behind the human signature are collapsed into a generated output that feels authoritative because it mimics the hallmarks of journalism and because it appears on the institutional webpage of recognised outlets.
The shift from reading to asking hence goes beyond interface design, as it reconfigures the relationship between news producers and news consumers.
The Substitutive Logic and Its Costs
What characterises most current deployments of AI in reader-facing news platforms is what can be described as a substitutive logic: in this sense, the AI does not help users navigate and read journalism, but rather reads it for them, and then interprets what it found. The reader’s role is reduced to that of a query-issuer, and the journalistic text becomes raw material to be processed rather than a communication to be navigated (Gorwa et al., 2020).
The implications of this substitution are both epistemic and democratic. When interpretive labour is delegated to an AI system, the reader loses the opportunity to encounter the complexity, the ambiguity and the editorial voice embedded in the original text. For this reason, journalism should not be conceived as an information delivery mechanism: most importantly, it is a stance about what matters, made visible by identifiable human actors who can be held accountable for it. A generated summary, instead, approximates that argument, flattening the texture of editorial judgment into a surface of plausible-sounding text (Dodds et al., 2025 & Shin, 2025).
This substitutive model hence ends up relocating the sensemaking process. Studies of chatbot-assisted news consumption have begun to document this dynamic empirically (Komatsu et al., 2020 & Zhang et al., 2025): readers who interact with conversational news interfaces show patterns of increasing interpretive delegation, relying on the system’s framing rather than forming their own judgments through direct contact with reporting.
In this new privately produced, algorithmically curated shape of journalism, the thin line between accessibility and independent, accountable interpretation becomes increasingly difficult to draw, as accessibility is achieved precisely through the erosion of interpretive independence.
An Alternative: Navigation over Substitution
The substitutive paradigm is not, however, the only way to integrate AI into the reader’s experience of news. A different model should be explored, focusing on AI not as a replacement for the journalistic text but as a navigational layer that helps readers move through it more confidently and meaningfully.
In this alternative, AI assistance should operate at the service of the original reporting rather than in place of it. It might help a reader understand who a named person is and why they matter to the story, without rewriting the story itself. It might surface related articles from the same publication’s archive that provide context for a complex ongoing situation, guiding through maps and timelines rather than substituting the act of reading. It might make visible the editorial choices that shaped a piece (e.g. the sources consulted, the framing decisions made, the values that governed what was included and what was not) inviting the reader into a more reflective relationship with the text.
This navigational model assumes that the intrinsic quality of journalism lies also in the editorial voice that shaped it, in the accountability structure that stands behind it, and in the act of engagement that it demands from its reader. In this sense, this kind of AI implementation could make journalism more accessible: helping readers who lack background knowledge orient themselves within complex stories, making archives explorable by people who would never have found them through traditional search, lowering the threshold of engagement for readers who are curious but intimidated.
Reading as a Democratic Practice
At stake, ultimately, is the understanding of reading news as a democratic practice more than a consumer behaviour. The core concept is that journalism has traditionally been an encounter between a voice and an audience, ready to understand and potentially question it (Diakopulos, 2015 & Hoque et al., 2024). It’s important to say that this encounter is never neutral, and journalism has never been perfectly objective. But despite this, it has always been an invitation to engage with a version of reality that someone took responsibility for producing.
As AI systems become the dominant interface through which readers access that information, the question of how those systems are designed becomes inseparable from the question of what journalism is for. The difference between AI that answers questions and AI that helps navigate them is as important as anything else about how the technology is built.
References
Diakopoulos, N. (2015). Algorithmic accountability: Journalistic investigation of computational power structures. Digital Journalism, 3(3), 398–415.
Dodds, T., Yeung, W. N., Mellado, C., & Lima-Santos, M.-F. d. (2025). On Controlled Change: Generative AI’s Impact on Professional Authority in Journalism.
Gorwa, R., Binns, R., & Katzenbach, C. (2020). Algorithmic content moderation: Technical and political challenges in the automation of platform governance. Big Data & Society, 7(1), 2053951719897945.
Hoque, M. N., Mahfuz, A. A., Kindi, M. S., & Hassan, N. (2024). Towards designing a question answering chatbot for online news: Understanding questions and perspectives. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI ’24). Association for Computing Machinery.
Komatsu, T., Gutierrez Lopez, M., Makri, S., Porlezza, C., Cooper, G., MacFarlane, A., & Missaoui, S. (2020). AI should embody our values: Investigating journalistic values to inform AI technology design. In Proceedings of the 11th Nordic Conference on Human-Computer Interaction: Shaping Experiences, Shaping Society, 1–13.
Shin, D. (2025). Automating epistemology: How AI reconfigures truth, authority, and verification. AI & Society.
Simon, F. M. (2024). Artificial Intelligence in the News: How AI Retools, Rationalizes, and Reshapes Journalism and the Public Arena. Tech. rep., Tow Center for Digital Journalism, Columbia University.
Zhang, Y., Nguyen-Le, P.-A., Singh, K., & Gao, G. (2025). The news says, the bot says: How immigrants and locals differ in chatbot-facilitated news reading. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Association for Computing Machinery.
Further Reading:
Leading Newsrooms in the Age of generative AI. Report. European Broadcasting Union, Geneva, 2025 https://www.ebu.ch/guides/open/report/news-report-2025-leading-newsrooms-in-the-age-of-generative-ai
Barthes, Roland. « The pleasure of text », 1982.
Image Attribution
Generated by: ChatGPT
Date: 10 May 2026
Prompt: Prompt:“Generate an image that represents the act of actively reading online news, of creating connections between texts and involved understanding.”