91探花

For the past couple of years, we鈥檝e been told, pretty consistently, that AI is coming for writers. Of course, we鈥檝e been told that AI is coming for everybody, but of those who should be most concerned, writers are at the top of the list.

Marketing teams are generating blogs with ChatGPT, startups are replacing copywriters with prompts and LinkedIn is overflowing with people proudly announcing that they wrote an entire ebook in 15 minutes (and others complaining about how much LinkedIn content is AI-generated).

So when Anthropic, one of the world鈥檚 leading AI companies and the creator of Claude, advertised for a Standards Editor with an annual salary reportedly reaching $295,000 (and potentially above), eyebrows were raised.

If AI is getting so good at writing, why is one of the companies building it willing to pay nearly $300,000 for a human editor?

My first thought when reading this news was that this feels immediately contradictory, but, upon further inspection, it seems as though there鈥檚 actually a lot more that meets the eye.

听

Generating Words Isn鈥檛 The Same As Writing

听

One of the biggest misconceptions around generative AI is that producing text and writing are the same thing.

They鈥檙e definitely not. AI is remarkably good at creating grammatically correct sentences, summarising information and producing first drafts. Syntax, essentially.

What it鈥檚 less good at is understanding, however, is nuance, context, judgement and audience expectations. And these are al the things that turn words into effective communication.

As Jonny Murphy-Campbell, AI Expert and Commercial Director at Resolvable, puts it,听鈥淎I is good at generating text but what it鈥檚 missing is human judgment and nuance, and the creative flare of writing that only humans can truly grasp and execute.鈥�

And this distinction is something many businesses are beginning to recognise. The challenge isn鈥檛 just generating content quickly anymore; we鈥檝e got that done. Now, it鈥檚 making sure that content is accurate, trustworthy and actually worth reading.

听

The Age Of AI Content Fatigue

听

Another factor at play here is that people are getting better at spotting AI-generated content (and not simply due to AI detectors, but that鈥檚 a whole seperate discussion). The internet is becoming increasingly crowded with articles, social posts and marketing copy that all sound strangely similar 鈥� it鈥檚 perfectly polished, perfectly structured and also perfectly forgettable.

Maria Levitov, Co-Founder and Managing Director at Snow Hill Advisors, believes we鈥檙e already seeing the consequences of this: 鈥淲hile AI is helping democratise writing, it is also producing major fatigue with content that looks polished on the surface, but lacks originality and often contains major logic gaps once read more carefully.鈥�

For companies like Anthropic, standing out may require exactly the thing AI struggles with most. That is, personality and taste.

That means editors who can identify weak arguments, spot confusing explanations and shape complex ideas into something people genuinely want to engage with.

听

听

Human Oversight Is Becoming More Valuable, Not Less

听

There鈥檚 also a practical reason for hiring experienced editors. As AI becomes embedded into more products and workflows, the risks associated with inaccurate outputs become far larger. Hallucinations, bias and misleading information aren鈥檛 just annoying mistakes anymore, they can damage trust.

Stuart Harvey, CEO of Datactics, argues that human involvement isn鈥檛 optional: 鈥淎nthropic ensuring a human editor remains in the loop of writing is a reminder to businesses that AI should听鈥� in fact, must听鈥� never replace the need for human intervention.鈥�

This is particularly true for AI companies themselves. If your entire business revolves around convincing customers that your technology is reliable, your communications need to be held to an extremely high standard.

听

Are Writers Being Replaced Or Upgraded?

听

The Anthropic job posting may actually tell us less about the end of writing and more about how writing jobs are changing. Rather than replacing humans entirely, AI seems to be shifting where the value sits. Producing a first draft is becoming cheaper and faster, and reviewing, refining and shaping that draft is becoming more important.

As Jeff Watkins, Chief AI Officer at NorthStar Intelligence, says,听鈥淲e may be approaching a point where the more AI allows us to create, the more valuable someone who knows what good looks like becomes.鈥�

And I鈥檇 like to think that鈥檚 the most important conclusion here. The companies building the world鈥檚 most advanced writing tools still appear willing to pay a premium for human judgment, but not because AI can鈥檛 generate words, but because knowing which words matter is still proving surprisingly difficult to automate.

For an industry supposedly watching writers become obsolete, that鈥檚 a pretty interesting development.

听

What Do the Experts Think?

听

• Stuart Harvey: CEO of Datactics
• Jamillah Knowles: Editor and AI Specialist at Persuasion
• Jonny Murphy-Campbell: AI Expert and Commercial Director at Resolvable
• Maria Levitov: Co-Founder and Managing Director at Snow Hill Advisors
• Jeff Watkins: Chief AI Officer at NorthStar Intelligence

听

Stuart Harvey, CEO of Datactics听

听

听

鈥淎I is an extremely useful business tool that is continuing to rise on the agenda of businesses of all sizes across various industries. Anthropic ensuring a human editor remains in the loop of writing is a reminder to businesses that AI should 鈥� in fact, must 鈥� never replace the need for human intervention.鈥�

鈥淢odern AI models are trained on vast amounts of datasets, and their outputs reflect the strengths and weaknesses of that data, which calls for a human in the loop to sit at the intersection of editorial judgement and data governance. AI models need editors who can spot when a model is hallucinating or reproducing bias, particularly to interpret how training data influences model behaviour.鈥�

鈥淭he value of human oversight has increased as AI becomes more capable and integrated into more systems, and the businesses that acknowledge this are the ones that will have a competitive advantage over those with messy data and inaccurate AI systems.鈥�

听

Jamillah Knowles, Editor and AI specialist at Persuasion听

听

听

鈥淚t鈥檚 no surprise that Anthropic is hiring an editor, or that a tech company worth over $960 billion pays this well. As an editor, of course, it hurts me to see a salary this high, knowing that I would not be employed in the UK as an editor for any publication for that much money. As an AI specialist, I can see why Anthropic would want someone with experience and attention to detail.

鈥淭ech comms can often be seen as frothy, having well-written comms with technical knowledge is a smart move as Anthropic now talks to all sorts of stakeholders and generations. Tech hiring writers is not new, Cond茅 Nast staff have gone to Apple and Instagram, News editors have stepped over to Meta and Notion recently took on a name from the New York Times.

鈥淲ords matter, and AI-generated ones are not reliable enough yet for comms, especially when the stakes are high, we鈥檝e all seen too many mistakes from lazy systems already and while AI still fails to process the idea of context, we鈥檒l still see jobs like this coming up.鈥�

听

Jonny Murphy-Campbell,听AI Expert and Commercial Director at Resolvable

听

听

鈥淎I is good at generating text but what it鈥檚 missing is human judgement and nuance, and the creative flare of writing that only humans can truly grasp and execute. AI can never replace writers; even as AI advances and natural language processing becomes more advanced, good writing requires human experience of the world to be able to apply crucial context.

鈥淭hey鈥檙e putting so much money into the editor role because they鈥檙e fully aware of the limitations of AI. Producing the first draft of writing is what AI can do, but actually assessing clarity, tone of voice, accuracy and fact-checking, nuance, and making writing trustworthy requires lots of written experience 鈥� it鈥檚 highly skilled. Anthropic understands fully what AI is capable of and also what it鈥檚 not.

鈥淭he most successful and ethical AI models lean on human experience. For an AI model to run safely and effectively, human oversight and governance is crucial at every stage of the journey, and that鈥檚 where human specialists are essential. Model training is crucial at every stage, which is why editors 鈥� who seem like they鈥檇 be needed at the final stage of the process 鈥� are actually crucial the whole way through.

鈥淎I should always need humans more than humans need AI to run. They鈥檙e supposed to work together with AI as a tool for humans. The most ethical and successful systems of AI prioritise human expertise, so any salary is an investment that reflects the strategic importance of the role.鈥�

听

Maria Levitov, Co-Founder and Managing Director at Snow Hill Advisors听

听

听

鈥淕enerative AI has made writing fast and accessible to everyone 鈥� writing or analytical skills are no longer needed to produce polished social media posts or churn out editorial copy that fits the exact word count specifications. However, AI鈥檚 speed often doesn鈥檛 come with better quality.

鈥淲hile AI is helping democratize writing, it is also producing major fatigue with content that looks polished on the surface, but lacks originality and often contains major logic gaps once read more carefully. This is especially evident on platforms like LinkedIn, where the few posts still written by humans truly stand out in a sea of AI-generated slop that all sounds vaguely the same.

鈥淚t鈥檚 possible that excellent editors employed by AI companies will help train the algorithms and improve them to compete with human writers in the future. However, we are not there yet, so talented editors who help communicate original ideas clearly and effectively are in high demand.鈥�

听

Jeff Watkins, Chief AI Officer at NorthStar Intelligence

听

听

鈥淭he media narrative is that AI is replacing writers, yet Anthropic is hiring a Standards Editor for up to $295,000 a year. So what gives? It feels like a contradiction, but only if we confuse writing with the mere generation of words.

鈥淲riting is not simply the act of placing one word after another. It involves understanding the brief, conducting research, synthesising ideas, choosing the right form, accounting for different markets and audiences, and reviewing the result with care. In my opinion, the real value is often in the edit.

鈥淭he original cut of Star Wars was reportedly so poorly assembled that Brian De Palma took a very dim view of it. Yet in the hands of skilled editors, it became the classic we know today. A great piece of writing is not only about the words themselves. It is about where they go, how they make the reader feel and, crucially, what gets left out.

鈥淗iring writers and editors also does not preclude the use of AI. In reality, there is likely to be an expectation that these people will use it to produce more work, more quickly and to a higher standard.

鈥淎I does not merely 鈥渨rite鈥� for you. It can accelerate research, help review grammar and structure, check material against a house style, and assess whether a piece meets a brief or rubric. The combination of powerful AI tooling and an expert human genuinely in the loop, rather than merely observing from the sidelines, can produce extremely high-quality work.

鈥淧erhaps most importantly, the question of accountability remains a very human affair. Organisations are judged by their public face. AI-generated material that has not been properly sense-checked, tailored to its audience or enriched with genuine human insight creates a poor impression. It can make an organisation appear careless, generic or, worse, as though it does not fully understand what it is saying.

鈥淭here is also a more subtle risk. As AI takes on more of the production process, people may engage less deeply with the underlying subject. If someone has not properly understood or curated the material, they may be unable to spot errors, challenge weak reasoning or recognise when the output simply does not ring true.

鈥淎nthropic is not paying someone nearly $300,000 to check Claude鈥檚 use of em dashes. It is hiring them to protect its brand, voice, and reputation for producing genuine insight. The role will involve translating complex technical material into accurate, timely and accessible content that its audience can understand and perhaps even enjoy.

鈥淎I is making the first draft cheaper and faster for most of us. But that leaves more room for taste, human judgement, regional tailoring and accountability for what we are prepared to put our names to.

鈥淲e may be approaching a point where the more AI allows us to create, the more valuable someone who knows what good looks like becomes.

鈥淭he harder question is what happens to everyone else. Will organisations employ fewer writers and editors, while expecting each of them to oversee a much greater volume of material? As with AI-generated code, will more of us become reviewers of machine output rather than creators in our own right?

鈥淎I may not eliminate writing, but it could change the nature of the work. And in doing so, it risks removing the part many writers find most rewarding: the writing itself.鈥�