The writing world I wanted to join disappeared before AI

Watching The Devil Wears Prada 2 made me think less about AI slop and more about the media institutions we had already allowed to disappear.

Watching The Devil Wears Prada 2 affected me more than I expected.

Not because I ever worked at a fashion magazine, or because I think the media world it depicts should be treated as a lost golden age. It could be elitist, exploitative and almost impossible to enter without the right background or connections. The film is hardly an argument for healthy working practices either.

It affected me because I began my career in journalism. At the time, I thought magazine journalism might be where I eventually ended up.

That career already felt as though it was disappearing while I was trying to enter it.

I have spent a lot of time recently reading arguments about AI slop, the declining quality of writing and the jobs generative AI might destroy. Those are legitimate concerns. But the film reminded me that the writing economy did not suddenly become fragile when ChatGPT appeared.

Media had been slowly gutted for years.

The career I thought I was entering

Digital publishing did not arrive as an obvious act of destruction. Much of it looked like progress. Publications could reach larger audiences, new writers could publish without waiting for permission from a small group of gatekeepers and entirely new kinds of work appeared. I benefited from that expansion myself.

But the incentives changed the work. I worked in digital journalism during the period when publishing increasingly became a game of feeding Google. Finding something another publication had reported, rewriting it quickly and getting it online could be easier to defend than spending days finding something genuinely new.

I have written separately about how SEO helped hollow journalism out, and about why slop didn’t arrive with ChatGPT — a decade of it was already being produced slowly and carefully by humans, because it ranked. It was not the only cause: advertising moved to technology platforms, publishers made bad decisions and audiences became accustomed to reading journalism without paying directly for it. But the result was a system that increasingly rewarded distribution and volume while making original reporting harder to fund.

I saw what that did to the craft from inside the system. The part of journalism I found most interesting — talking to people, finding the angle and working out what was actually happening — became harder to justify than producing another version of information that already existed.

I did not leave because I stopped caring about writing. I left partly because the industry was rewarding the part of the job I valued least and making the part I valued most increasingly difficult to sustain.

What disappeared was an institution

What struck me while watching the film was not simply the magazine. It was the publication as an institution.

At its best, a magazine was more than a container for individual pieces of content. It had an identity, a point of view and a group of people developing judgement together. Writers, editors, photographers, designers and specialists learned from one another. Somebody could decide that a subject mattered before search demand or a social algorithm had proved it would perform.

The reality was never as romantic as that description makes it sound. Publications depended on advertisers, access, wealthy owners and cultural hierarchies of their own. Editors could exercise terrible judgement as easily as good judgement. Digital also allowed voices into public life that traditional media had ignored.

Fashion publications also did something more complicated than preserve editorial craft. They helped create the cultural meaning on which the luxury business depended. A Dior bag priced at £3,000 or more is not valued only for its leather, labour or practical function. Craft matters, but so do heritage, scarcity, celebrity, editorial approval and decades of shared agreement that the name means something. That is narrative operating as economic infrastructure. It is another reason not to romanticise the old system: magazines could exercise real judgement while helping turn objects into status.

But replacing human gatekeepers with platforms did not remove gatekeeping. It moved much of it into ranking systems designed around attention, engagement and advertising. The old institutions lost authority without an obviously better set of institutions taking their place.

The loss is therefore not only fewer beautifully edited features or fewer jobs with glamorous magazine titles. It is the disappearance of places that gave people time, support and a public purpose for developing editorial judgement.

Craft does not exist entirely inside an individual writer — the writing was never the whole value it was assumed to hold. It is also produced by commissioning, editing, challenge, apprenticeship, specialist knowledge and the opportunity to do the work repeatedly with people who expect it to improve.

When those conditions disappear, telling individual writers to care more about craft can only achieve so much.

Content marketing inherited some of the people, not the purpose

I moved into content marketing, as many journalists did — one stage in a longer argument I’ve made about that career rather than a clean break from it. It offered a more viable career and still allowed me to research subjects, interview people, construct arguments and explain complicated ideas. Good content marketing can be useful, original and carefully made. I have spent years trying to make it more so.

There is also an irony in the current moment. As generic production becomes easier, companies increasingly need some of the practices publishing weakened: original research, close contact with customers, subject expertise, evidence, interpretation and editors capable of recognising a real story. That is part of what Editorial Intelligence is attempting to explore.

But content marketing is not a replacement for journalism, and I do not want to pretend that it is.

The two forms can share methods while serving different purposes. Content marketing ultimately exists to help an organisation achieve a commercial objective. Journalism, at its best, investigates organisations and institutions on behalf of a public. One can borrow the discipline of the other without inheriting its independence.

The difference becomes clearest when something disappears. If a company closes its blog, its marketing strategy changes. If a town loses its reporters, fewer people consistently watch its council, courts, schools and businesses. If cultural publications close, criticism, discovery and the recording of a period become more dependent on creators, platforms and whatever already has momentum.

That feels more consequential than whether another generic B2B article was drafted with AI.

The AI slop argument is too small

I’ve made a version of this argument before: the slop debate stops at the paragraph — whether a sentence sounds like ChatGPT or Claude, whether someone used an em dash — while the bigger shift happens upstream of it. The institutional version of that shift is what this piece is actually about.

I care whether prose is good. I care about accuracy, rhythm, originality and whether a writer has exercised judgement rather than accepted the first plausible collection of sentences a model produces. The abundance of generic machine-written material is a genuine editorial problem.

But craft is not protected merely by proving that a human wrote something.

Plenty of the material produced during the SEO era was generic, derivative and disposable despite being written manually. Meanwhile, a reporter using AI to search documents, transcribe interviews or interrogate a dataset may be using the technology in service of work that is far more original and socially useful.

The more important questions sit around the text. Who commissioned it? What did they want to find out? What evidence did the writer obtain? Who challenged the argument? Was there time to investigate rather than repackage? Who is paying for that work, and what do they expect in return?

Those are questions about institutions and economics as much as individual craft.

AI did not enter a healthy writing economy. It entered one that had already spent years reducing the economic value of the work it is now capable of imitating. It may accelerate the damage, but blaming AI alone lets publishers, platforms and the previous two decades of decisions escape scrutiny.

I still believe AI can help people do better editorial work. It can reduce some of the mechanical effort that once consumed a journalist’s time and make it easier to connect evidence spread across interviews, documents and datasets.

But a faster research tool does not create the institution willing to fund the research. A better drafting tool does not restore the editor who was made redundant. A model cannot decide that a local council meeting deserves sustained attention when no viable publication remains to send somebody there.

Watching The Devil Wears Prada 2 did not make me want to return to an imaginary golden age of magazines. It reminded me that I once expected to build a career inside a media culture that was already receding — and that some of what disappeared was genuinely valuable.

The writing world I imagined was not destroyed by AI. It was slowly dismantled during the digital era, often under the language of innovation, access and efficiency.

AI might be the next stage of that dismantling. It might also give the people who remain better tools for finding things out. But neither possibility answers the question that matters most: who will sustain the institutions in which serious editorial work can still happen?

That seems more important than another argument about whether a paragraph sounds like ChatGPT or Claude.