> ## Content Index
> Fetch the complete content index at: https://balaustine.mymagic.page/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI shouldn’t distract journalism into a new trap
- URL: https://balaustine.mymagic.page/ai-shouldnt-distract-journalism-into-a-new-trap/
- Published: 2026-08-21T10:43:16.000Z
- Updated: 2026-08-23T11:25:13.000Z
- Author: VM
- Tags: Analysis, Tech, artificial intelligence, Big AI, Big Tech, business models, Clayton Christensen, deprofessionalisation, editorial authority, Habermasian theory, jobs to be done, journalism, #Migrated-1787484114599, #wp, #wp-post, #Import 2026-08-23 11:22

*The following is a review of* Journalism in the Age of AI: From Acceleration to Reimagination *by Seth C. Lewis, Rodrigo Zamith, and Tomás Dodds, available* [*here*](https://polity-books-backend.prod.politybooks.wiley.host/wp-content/uploads/2026/07/LEWIS-9781509570690-EPDF.pdf?ref=balaustine.mymagic.page)*.*

The authors write, and I agree, that journalism has spent 25 years on a “hamster wheel” of technologies — web, then social media, then mobile — each of which has only forced journalists to produce more and produce it faster. Today, the advent of generative AI promises to do that on steroids, but the authors argue that, by virtue of targeting knowledge work itself rather than just the logistics of production, it also presents journalism with a “constitutive moment”, an opportunity of sorts for journalism to reflect on its real purpose. That is to say, journalism can use the “hamster wheel” to run even faster or, to extend the metaphor, it can reimagine what ‘running’ even means.

Meanwhile, AI is also reshaping political life by increasing access to personalised information, expanding the ability to produce disinformation at scale, imposing a greater cognitive overload, and of course confronting the publics with the specific failures of governance-by-algorithm. Per the authors, the harms of AI for journalism amount to an “epistemic thinness” characterised by deprofessionalisation, infrastructural capture, cognitive atrophy, and the corrosion of democracy’s instruments.

As a way out, they propose using Clayton Christensen’s ‘jobs to be done’, or JTBD, theory, “which asserts that customers don’t simply buy a product or service—they ‘hire’ it to do a ‘job’” ([source](https://online.hbs.edu/blog/post/jobs-to-be-done-examples?ref=balaustine.mymagic.page)). That is, to start from what communities need rather than from what newsrooms are currently designed to produce. And to this end the authors propose four starting objectives for journalism: accessibility, resisting disinformation, micro-relevance, and conversational flexibility, and acknowledging that personalisation also risks “ceding editorial authority” and fragmenting shared understanding. In any case, they conclude, it’s time for journalism to “step off the wheel”.

![](https://balaustine.mymagic.page/content/images/2026/08/screenshot-2026-08-15-at-9-23-55-am.png)

I agreed with a lot of this, especially the [Habermasian frame](https://en.wikipedia.org/wiki/Deliberative%5Fdemocracy?ref=balaustine.mymagic.page), but I also disagreed with four specific bits that also go to the heart of the authors’ proposed objectives.

First, JTBD theory requires a journalist to ask what job a reader is trying to get done, so for instance we give an affected person only what they need to act, rather than a 1,200-word piece with context and analysis. However, a felt need is different from a civic need and JTBD theory doesn’t distinguish between them. That is, a reader asking “is my water safe?” should, according to JTBD, be met with a short piece explaining how to check if your water is clean. However, journalism — as the fourth estate, as a democratic institution that speaks truth to power — should also explain why the water is unsafe to begin with, who’s responsible, and which communities are affected by it. So while JTBD theory can be useful in order to improve delivery and framing, it shouldn’t be used to determine what goes in a journalistic piece to begin with.

Second, journalism shouldn’t tie its self-definition to any external frontier, even implicitly, because if it does, it will end up reacting to changes in that frontier. So while the authors have organised their entire book around the idea of AI as an entity that can catalyse changes in journalism, if the reckoning about journalism’s public purpose is really so overdue — and I admit that it might be — why does it need AI to be the occasion around which it turns? Put another way, journalism should conduct its reimagination based on its own civic mandate first, and treat AI as just another variable to negotiate for now.

Third, the book is right to diagnose the capture of the infrastructure of journalism as the fundamental reason a journalism based on “who’s paying”, rather than “who’s affected”, could only fail. However, the book’s workarounds, including licensing deals with AI companies, all only attempt to get a better deal within the commercial relationship with the so-called “Great Houses” of AI rather than attempt to decouple journalism’s prospects from that relationship altogether. One real-world example of instruments that could facilitate such a decoupling, and which the book also acknowledges, is South Africa’s Competition Commission proposing a 5-10% levy on the platform ad revenue if Google and Meta didn’t comply with the Commission’s remedies to ‘fix’ biases in search results, share audience data, boost locally produced content, and compensate news publishers. Another example is reader-owned news outlets, although I have my misgivings about that in the socio-political milieu prevailing in India.

Finally, the authors’ proposal that journalists aim for micro-relevance and conversational flexibility erect the risk of the problem of a self-fulfilling persona. Specifically, once a newsroom adopts an audience archetype — e.g. that it will write for a 30-year-old female, city-based, college-educated, and salaried citizen — its journalists will start writing to that category instead of mindfully discovering each new story. In the same vein, AI being able to tailor each story according to a reader’s needs or preferences also endangers a shared understanding of important developments, such as a Supreme Court ruling or the results of a corruption investigation, and threatens to create a multitude of one-person publics that ultimately can’t agree on the basic facts of public life.

Overall, the authors are right to invite journalism to step off the wheel but what they’re proposing it do instead just kicks the can down the road.