Searches for “How to humanize AI content” person grown 943% implicit the past 12 months. 

More radical are trying to “pass the quality check”, either to marque AI contented suck a small little oregon to blag AI contented detection tools.

The process of “humanizing AI content” has go the de facto contented strategy of growth-hungry companies.

I spot stifled writers sprouting up everywhere, asking for proposal connected however champion to navigate this contented headache.

And I cognize conscionable however soul-destroying it tin be, due to the fact that I’ve got the t-shirt.

Here’s wherefore it’s a doomed effort from the start.

We each cognize the large glaring problems with AI contented (and if you don’t, you tin work astir them here), but I don’t judge anthropomorphizing the bot is the solution.

In fact, I’d reason that humanizing the long-form contented churned retired by AI is an inherently flawed process.

The full constituent of utilizing AI is to thrust speed, scale, and efficiency; getting humans to edit, and fundamentally rewrite, long-form AI prose wholly undermines that worth proposition.

By the clip you’ve researched and added successful the “rich” quality element, you whitethorn arsenic good person created the contented from scratch yourself.

And the process of moving “humanized” AI contented done a detector is nonsensical – you’re virtually getting an AI to validate humanity.

I’d walk hours editing the AI contented I’d been saddled with truthful it could walk the quality check, lone to beryllium gaslit into reasoning my contented was AI… by an AI?!

And I’m not the lone victim. In a LinkedIn post, contented advisor Kiran Shahid shared that AI detector tools would lone verify her humanity erstwhile she wrote successful first-person – a benignant of penning that clients are improbable to motion off on.

I person seen akin experiences being shared successful the Women In Tech SEO community…

And connected Reddit…

Having quality writers gussy up AI-generated content, lone to person AI measure their humanity, is not lone bonkers — it’s a flagrant discarded of talent, and frankly a small insulting.

When I was required to “humanize AI”, my workflow looked thing like this:

1. Receive batches of 100% AI contented to edit
2. Make definite each nonfiction was astatine slightest 50% “human” successful AI detection tools
3. Turn each the circles greenish successful the SEO postulation airy instrumentality (pro tip: keyword stuff. Apparently it works wonders in 2024!)

The AI articles sent implicit to maine were truthful disconnected the people I fundamentally rewrote them (but with the added unit of an unrealistic deadline) and, adjacent then, the extremity results were subpar.

When you’re shoehorning successful examples, opinions, and insights which could person shaped the nonfiction if lone they had been included from the beginning, the content’s travel is inevitably going to beryllium compromised.

At this point, I privation to wide thing up. I’m not anti AI. I usage it each time for ideation, structure, outlines, and repurposing – my fig 1 workflow is really to amended the pacing and readability of contented I’ve written successful “flow”. Used similar this, AI tin beryllium incredibly efficient.

But the process of “humanizing AI” — wherever you commencement with 100% AI-generated contented and animate it similar you’re Mary Shelley breathing beingness into Frankenstein’s monster — is 1 of the astir menial tasks and beyond demotivating for writers (despite however chill I’ve conscionable made it sound).

If you’re the 1 enforcing this benignant of strategy, you should besides cognize that it is simply a large mode to haemorrhage talent.

Humanizing AI contented is unsustainable and inefficient, but immoderate companies volition marque you bash it anyway.

If you’re successful this situation, 1 mode you tin marque your beingness a small easier is by grooming AI connected your contented from the outset.

This means, whenever you’re handed a batch of AI generated articles to “humanize”, you tin tally them done your self-trained AI exemplary and get them up to a much passable level of humanity, earlier going backmost successful with your ain edits and additions.

Training an AI exemplary is simply a astute determination due to the fact that it improves the AI strategy itself; otherwise, it’s conscionable laborious manual work, and you mightiness arsenic good person written the contented to begin with.

Tools similar Claude let paid users to make shared “Projects” containing up to 200,000 words of grooming documentation, portion ChatGPT’s customGPTs fto you bid connected up to 20 files, each with a maximum size of 512 MB.

Nishkarsha Kotian, SEO Specialist astatine Channel Engine feeds examples of her marque guidelines to a customGPT, and sets retired rules for the bot to follow.

Similarly, Melissa Popp, Content Strategy Director astatine RicketyRoo, trains her customGPTs connected cognition documentation, and adopts a process of “few changeable prompting” by providing examples of her champion content.

Given that AI tools person a “memory”, you tin besides practise “iterative refinement” portion conversing with them.

In different words, you tin bid them spot by bit, utilizing proposal and examples to amended the wide contented output. Digital PR Expert Gini Dietrich describes AIs arsenic “digital interns” successful this respect.

The happening is, grooming takes time, and adjacent with each the discourse successful the world, if you don’t person a wide imaginativeness of what you privation from your integer intern you tin easy get trapped successful endless punctual loops.

So, grooming an AI exemplary tin lighten the load a little, but it’s acold from foolproof.

Melissa Popp notes that customGPTs lone get her 40-50% of the mode there, and the effort expended successful grooming often undermines the benefits.

It tin beryllium tricky to question “efficiency” strategies similar humanizing AI – particularly if they’re dictated on high.

Often those making the decisions don’t recognize the world of the work, and are lone acrophobic with turning spreadsheet numbers green.

But it’s your occupation to marque that world intolerable to ignore.

In my case, I took these steps:

• Regularly fed backmost connected issues I faced during the process of “humanization” successful meetings
• Tracked the clip I spent “humanizing” AI successful a collaborative spreadsheet to beryllium time-sink
• Outlined the risks of pushing retired subpar AI contented with examples of brands that tanked successful the SERPs owed to manual oregon algorithmic penalties

An alternate I didn’t research is steering stakeholders towards low-involvement AI contented that doesn’t necessitate extended humanization, such as:

• Glossaries
• Product descriptions
• FAQs
• Wikis

If the eventual extremity is traffic, you mightiness besides privation to see proposing non-AI initiatives that person been proven to thrust postulation at scale.

In Ahrefs’ caller probe connected the 50 Bootstrapped SaaS Companies Dominating SEO successful 2024, the brands driving the largest YoY postulation maturation had embraced 3 halfway strategies:

• “free tools” creation
• programmatic content
• content localization

Non-AI projects don’t person to beryllium clip intensive either. Zapier, for example, created thousands of feature-based landing pages programmatically, and past outsourced the leafage transcript to partners during onboarding.

This benignant of traffic-play is acold much apt to fulfill a client/higher-up’s request for quick-win growth, giving you clip backmost to absorption connected your occupation of creating prime content.

Final thoughts

Humanizing AI generated contented is unsustainable and – contrary to what galore radical deliberation – inefficient.

AI should beryllium utilized for ideating, planning, and drafting from the outset – not ghostwriting.

In the words of Moz’s Senior Content Marketing Manager, Chima Mmeje:

AI contented is not the problem. The occupation is erstwhile the contented is truthful poorly written, I tin archer it was AI-generated. AI detectors are trash. If you prosecute a writer and you’re passing contented done a detector arsenic impervious that it’s engaging oregon authoritative, past you should not beryllium moving with writers. Writers, I’m begging you to halt utilizing AI to make full content. Please usage it arsenic a usher to assistance you soma retired ideas, implicit a thought process, oregon adjacent cleanable up your writing, but that archetypal draught should beryllium 100% your work.

If you’re speechmaking this and you person the powerfulness to dictate strategy, bash distant with the soul-destroying process of humanizing AI content.

Trust your humans to constitute contented for humans.