In as we speak’s fast-paced digital age, it’s necessary to acknowledge that synthetic intelligence has revolutionized the best way we create content material. Whereas AI presents quite a few advantages, it’s equally important to make sure that the content material produced stays genuine and resonates with…
Wait there! Wait proper there!
In the event you sniffed out AI’s trademark tone within the first paragraph, you deserve a gold star. In the event you didn’t, it is advisable learn via this whole piece.
AI has grow to be central to advertising, and most manufacturers, together with their writing groups, have jumped on the bandwagon. Good for productiveness? Positively.
However there’s a catch. AI-generated content material is like branded content material itself: When it tries too laborious to sound human, it’s apparent. AI detection instruments aren’t bulletproof, both, as they might flag content material inaccurately in both course, insisting a completely human-written article is AI or vice versa. (Research have discovered these instruments are practically ineffective.)
Whereas AI-generated content material might not damage your search engine optimization or GEO rankings, there’s clear proof that human audiences discover it off-putting. In response to analysis by Knowledge Entry Administration (DAM) firm Bynder, 50% of readers can detect when copy is AI-generated, and 52% discover such content material much less participating. Hootsuite’s 2024 Social Media Shopper examine additionally discovered that almost two-thirds (62%) of customers say they’re much less prone to belief or interact with social posts in the event that they know AI was concerned of their creation.
Supply
So if you wish to attain actual people along with your content material, your copy ought to sound prefer it got here from one.
Listed below are a couple of methods to edit your AI content material so it sounds much less like a corporate-jargon robotic and extra like an precise particular person with a viewpoint.
Frequent Indicators Your Content material Sounds Too A lot Like AI
AI-generated content material is typically so apparent you solely want a look to establish it, like when a single article comprises 55 em-dashes.
Different instances, you would possibly have to learn via the strains to establish telltale indicators like:
Generic or empty introductions
Nicolas Breedlove, CEO at PlaygroundEquipment.com, shares that one of many best giveaways is a scarcity of actual perspective or context. “Earlier variations of AI fashions, like ChatGPT 3.5, usually present content material intros that sound overly inflexible and repetitive. Their outputs additionally lack substance and don’t possess the proper hook to onboard your readers.”
Examine this generic intro for a journey information:
To this human-written model:
Your flight is delayed, your bag is chubby, and the resort simply despatched an electronic mail saying check-in isn’t till 4 pm. Anybody who has ever stepped foot in an airport, bleary-eyed and operating on unhealthy espresso and optimism, is aware of that journey is filled with these little complications. Nevertheless it doesn’t must be. This weblog will present you methods to pack like a professional, skip vacationer traps, and lower your expenses whereas nonetheless having fun with each cease of your journey.
The second is far more personalised and relatable.
Hedging language
Hedging language consists of phrases or phrases that weaken the assertiveness of an announcement and make it much less absolute. Whereas that is nice for tutorial writing, the place researchers wish to keep away from overstating conclusions or implying absolute certainty, utilizing it in your advertising copy can undermine your authority and make your model sound uncertain of itself.
Examples embody:
- It is very important be aware…
- It may be argued…
- Maybe…
- Perhaps…
You may additionally come throughout obscure phrases like “quite a few…” as an alternative of a particular determine.
Claims with out proof
Current knowledge means that even the newest model of ChatGPT (5.0) hallucinates about 1 in 10 instances in its outputs. Even domain-specific AI can’t be totally trusted: In a single Stanford benchmark, authorized fashions obtained issues unsuitable in as much as one-third of their responses.
In advertising writing, hallucinations would possibly embody fabricated quotes (yikes), bogus statistics (oof), or made-up research (simply ask Deloitte why that is problematic.)
So, when you see sentences beginning with:
- “Some analysis exhibits that…”
- “Research counsel that…”
- “Specialists agree that…”
- “It’s broadly identified that…”
So, cease and verify whether or not there’s an actual supply or knowledge level behind any claims AI churns out. Getting caught in an error is embarrassing, and it will probably shortly erode viewers belief.
Monotonous sentence constructions
“You would possibly see tons of formulaic transitions or connectors in a single or consecutive paragraphs, which finally ends up making the entire block really feel flat. This happens [especially often] when utilizing older LLM variations,” says Jeffrey Zhou, CEO and Founding father of Fig Loans.
See this:
AI additionally loves lists of three (Innovate, iterate, and encourage! Create, join, and convert!). These are advantageous sparsely, however in the event you begin seeing them in each part or paragraph, use your editorial judgment to gauge if they are often streamlined or phrased otherwise.
When each sentence size seems to be simply the identical or with as slight a variation as attainable, elevate your whiskers too.
Extreme use of contrastive parallelism
AI content material usually makes use of statements like, “Whereas X is true, Y can also be necessary,” or, “It’s not nearly gross sales; it’s about…”
You’ll generally see this back-to-back in a bit and it mechanically creates a rhythm that feels pressured.
Different examples:
- Social media is greater than a broadcast instrument; it’s a spot to construct relationships.
- It’s not about working more durable; it’s about working smarter.
- It’s much less storytelling and extra story-selling.
- The query isn’t what can AI do for us — it’s what ought to we do with AI.
Particular “tells”/phrases AI loves
You possibly can spot AI’s favourite phrases from a mile away — the linguistic equal of clip artwork. Listed below are a couple of lifeless giveaways you’ve most likely seen 100 instances:
- Delve
- Quick-paced world
- Greater than ever
- Quickly altering panorama
- Image/think about this
- The implications are clear
- The takeaway? or, The end result?
Spot related patterns on this weblog’s intro:
Supply
AI additionally has an affinity for colons in headlines and subheadlines, so edit extreme situations out to be few and much between. Random bolded textual content and oddly capitalized phrases are different formatting quirks that make a bit really feel extra like a machine making an attempt to emphasise significance than a human making a stylistic selection.
Should-Haves on Each Editor’s Anti-AI Guidelines
Enhancing AI-isms out of your content material isn’t so simple as recognizing catchphrases and pulling out parallelisms. Listed below are six issues you could look out for earlier than you hit publish.
1. Change weak verbs and minimize filler
AI usually makes use of padded language and limp verbs — phrases like “helps with,” “is geared toward,” “can be utilized to,” and “serves as.” Past being an AI inform, these flabby constructions are merely lazy writing. Savvy editors have been putting these empty phrases from drafts for many years.
Change these with stronger alternate options and minimize out excesses. As an illustration:
- Helps with → powers
- Is geared toward → delivers
- Works to → improves / strengthens / helps (relying on context)
- Is targeted on → prioritizes / advances
- Performs a job in → shapes / influences
It’s additionally a good suggestion to chop filler like “with a view to,” “in addition to,” or “when it comes to” that bloats sentences with out including which means.
2. Add names, numbers, examples
AI-generated textual content loves to talk in broad claims and sweeping generalities. The quickest strategy to humanize these eye-glazing paragraphs is to get concrete. Hunt down generalisms and personalize them by including names, correct pronouns, or particular examples the place relevant.
Add statistical numerical proof, case research, reviews, infographics — something that confirms your claims.
3. Reality-check, fact-check, fact-check
In the event you’re utilizing instruments like Perplexity or ChatGPT, they are often extremely helpful for surfacing fast references. Typically, these hyperlinks will level you in the proper course — however not all the time. Generative methods are educated to sound assured, to not be right.
That’s why each reality deserves a guide verify. Click on via each supply to verify it results in an actual, credible web page, ideally from a major supply or acknowledged publication. Confirm numbers by cross-referencing them with the unique examine or dataset, not simply the AI abstract. And in the event you see suspiciously excellent phrasing (like a too-good-to-be-true stat or a quote that appears like advertising copy), hint it again to its origin.
4. Enhance use of professional quotes or authentic enter
Weave professional quotes or insights into your content material to make it really feel much less automated, particularly when addressing delicate subjects or providing sensible ideas that require greater than generic recommendation.
To search out sources, you may make the most of platforms like HARO (which relaunched earlier in 2025), and social media platforms, resembling LinkedIn, in addition to skilled communities. You can too reference publicly obtainable quotes from credible people on respected third-party publications in your piece, as long as you cite the unique supply with a hyperlink or by title.
5. Insert native or distinctive particulars
AI content material can really feel prefer it’s addressing everybody and all the pieces. Make it sound extra distinctive by referencing native angles or examples, like these in your trade, geographical location, or related occasions in your area of interest.
Herald examples that your viewers can simply relate to. As an illustration, in the event you’re writing in regards to the state of the SaaS ecosystem, reference an actual milestone (e.g., Figma’s acquisition by Adobe) as an alternative of a obscure “main tech deal.”
These small touches make your piece really feel prefer it was written by somebody who really has experience and is aware of the viewers, versus scraped or regurgitated recommendation from the web.
6. Regulate transitions to sound pure
Extreme transitions give your piece a pre-scripted, bland tone. Guarantee they’re solely used when mandatory and minimize out others. Attempt to restrict formulaic transitions (like “furthermore,” “due to this fact,” or “as well as”) to not more than two each three paragraphs.
You possibly can introduce a mixture of pauses, conjunctions, and brief sentences to make your transitions sound pure and clean.
Enhancing Drafts on the Supply
Earlier than utilizing AI to populate your content material or refine your paragraphs, it is advisable situation it to offer you drafts which are as human as attainable. Listed below are some tricks to make your job simpler in a while within the course of.
Immediate for particular sources, not basic summaries
“AI-assisted drafts are solely as human-like because the prompts you present. In the event you ask ChatGPT to jot down an outline of ‘Generative Engine Optimization,’ it’s going to seemingly sew collectively generic summaries from its coaching knowledge,” says Anna Zhang, Head of Advertising and marketing at U7BUY. “As a substitute, immediate it to take a look at a selected supply, reference an instance, or cite a selected reality.”
The identical precept applies to statistics and case research used to help claims. Instruct your mannequin to retrieve knowledge solely from direct sources, relatively than basic statistical summaries.
Provide voice samples for tone
As soon as your prompts are grounded in actual sources, the following step is to verify the voice feels equally grounded in your model.
“Assist AI perceive your model tone each in textual content and by voice by recording complete samples. You possibly can merely carry your earlier podcasts, weblog samples, and vlogs (if obtainable) and paste them as coaching knowledge, supplied it doesn’t breach knowledge privateness legal guidelines,” advises Leigh McKenzie, group advocate at Visitors Assume Tank.
Different consultant branded particulars, resembling case research, optimistic public opinions, and occasion highlights, will even assist AI create content material that adopts your tone, whether or not playful, like Duolingo, or formal, like Forbes.
Add fashion guidelines: concise verbs, energetic voice, diverse rhythm
Create a mode information that particulars phrases to keep away from, which verbs to make use of as alternate options, most well-liked voice patterns, ultimate rhythms, and really helpful sentence lengths. Then feed it to your AI mannequin and run checks to see how intently it matches your model fashion.
The extra detailed your fashion information is, the higher your drafts will align along with your model’s voice from the beginning. You must also share the information along with your content material crew to allow them to higher immediate AI fashions and guarantee consistency in their very own writing.
Use unfavorable guidelines: no clichés, no hedging, no generic intros
AI instruments like ChatGPT can help you set guardrails round its output. Head to the personalization field and set unfavorable guidelines defending towards clichés, hedging phrases, generic introductions, and repetitive phrases.
These settings are usually not absolute, and the AI mannequin would possibly nonetheless bypass them. Chances are you’ll want to stick the foundations into the chatbox every time earlier than producing new drafts, however the “personalization” function acts as reinforcement.
Construct a Workflow That Helps Human-First Enhancing
Good content material is a results of a well-functioning content material workflow. You possibly can construct one by defining clear editorial processes throughout the outlining, drafting, reviewing, and publishing levels, together with specifying who handles enhancing and compliance.
Your outlining and reviewing levels ought to emphasize human-first enhancing, with a selected concentrate on implementing your checklists and magnificence guides.
You can too undertake AI content material platforms providing devoted human editors, resembling Contently, to make sure your drafts are correct, on-brand, and polished earlier than publication.
AI isn’t going anyplace — however neither is your reader’s intuition for authenticity. Preserve that your north star.
Content material that clicks with algorithms and audiences. See how Contently may also help.
Often Requested Questions (FAQs):
What’s the easiest way to make AI-generated content material sound human?
Edit for specificity. Add actual examples, quotes, and numbers. Differ sentence rhythm, swap out filler verbs, and trim hedging phrases. Be looking out for frequent AI tells and phrases, like “delve” and extreme use of contrastive parallelism or em-dashes.
Ought to I disclose when AI was used to create content material?
Transparency builds belief. If AI helped with analysis or outlining, it’s advantageous to say so, particularly if a human editor verified and refined the ultimate draft.
How can Contently assist with AI-edited content material?
Contently pairs AI-assisted instruments with a world community {of professional} editors who guarantee each piece sounds human, aligns along with your model, and meets editorial requirements earlier than it’s printed.
