What do you think surprised me the most about using AI for business writing?
1. It spews copy in seconds.
2. It can say Forbes published an article about author Marjorie Holmes.
3. It doesn’t always think critically or logically, or connect one idea to another without prompting.
If your answer was, “It can write a cold email like Bugs Bunny,” you get bonus points.
But if you guessed #3, you’re right. I had expected AI to be advanced enough beyond taking orders and make connections I hadn’t considered. Instead, I’ve sometimes found myself explaining to it why its own suggestion contradicts what I just told it.
As Spinal Tap’s Nigel Tufnel once said, “There’s such a fine line between clever and stupid.” And generative AI seems to straddle that line. You might also think of the fine line between “AI-assisted” and “AI-generated,” and wonder if it’s OK to mix both.
If you use AI to write copy, you won’t spontaneously combust like one of the Tap’s drummers (though I can’t confirm that). But based on using generative AI and my 17+ years of professional content creation experience, I’ve learned you should know its limits and how they could limit your marketing.
Because right now, to turn AI output into quality copy, you need solid editing skills. It’s all part of the AI-U or “human in the loop” philosophy.
For professional services firms, the stakes are higher. Clients expect accurate information, and many professions impose standards of care or conduct. Once published, a polished mistake becomes your firm’s error.
AI is useful for behind-the-scenes tasks like outlining, summarizing, or transcribing. It also helps with lower-risk work such as drafting a follow-up email or a social media post.
If you write blogs or other long-form content for a professional services firm, here’s when you should be cautious about using AI.
5 Reasons Not to Rely on AI for Business Writing
1. When you can’t spot bad writing: In AI-generated content, the “i’s” are dotted and the “t’s” are crossed. But that gloss won’t reflect well on you if you don’t recognize well-written copy. You must have the judgment to know where to edit.
By default, without clear directions, generative AI commits many sins “The Elements of Style” warned us about, namely:
- big words
- overlong sentences
- unnecessary words or sentences, especially when it restates what’s already implied
- clichés or AI-isms (aka slop): comparisons like “start”/”stop” or “it’s about…”/”isn’t about…”; “machinations” like “unlock,” “drive,” or “elevate”
Despite what some people believe, good writing uses em dashes — just not too many. Guess where generative AI got that habit.
Beneath the sheen, AI writing also sometimes says little beyond what readers already know.
Example: “Writing engaging newsletter content is essential for keeping your subscribers interested and encouraging them to interact with your brand.”
It’s a clone of your style — more like the android version. The writing resembles you, but lacks your soul — the personality behind your brand — what makes your writing uniquely “you.”
Despite clear prompts, AI is also like that friend who promises not to tell all, but spills anyway. For instance, if you ask AI to write promo for a blog post without giving away everything (known as a curiosity gap), it often spoils the surprise.
Across generative AI platforms, much of the output is similar: writing that reads well, but lacks impact. Knowing the difference between good and bad writing is crucial for creating interesting copy that inspires action.
2. When logic doesn’t make sense: Is the reasoning sound? Does the conclusion follow from the premise?
An AI suggestion can sound reasonable on its own yet lack sense in context or within the bigger picture.
Example: I use AI mainly for editing, which includes checking for missing information.
For a LinkedIn post about avoiding clichés, I told Claude the pointing-finger emoji (👉) is overused. It then suggested the same emoji in later drafts.
Another example: Commenting on an e-newsletter draft promoting my blog about professional services brand messaging, it suggested the same buying triggers apply to service providers and retailers. My blog posits the opposite — that service businesses should be sold differently.
3. When you don’t question the answer: You might have heard that you shouldn’t believe everything you read. Critical thinking questions assumptions and prompts you to double-check claims before accepting them as true.
Logic asks, “Does this conclusion follow?” Critical thinking asks, “Should I believe the premise in the first place?
Example: While editing a post about building credibility with content, Claude suggested I add “(No LLMs used in my client work)” in the call to action to address objections. That opposed my argument about accuracy and honesty, potentially undermining my credibility, as I mainly use LLMs while editing and for blog post images.
4. When you skip the fine print: In professional services fields, accuracy is vital, whether you’re describing tax law, estate planning clauses, or measles symptoms. Generative AI can make things up. Be especially cautious when it doesn’t cite its sources.
When I asked it for quotes, Perplexity, which is usually accurate, would sometimes misattribute or fabricate them entirely.
Example: “Don’t look for love to complete you. Look for someone who complements the wholeness you are already claiming in yourself.” – Atticus
The wrong data doesn’t just look bad — it can create professional and legal risks. Recently, attorneys in Florida and Minnesota have been sanctioned for providing false AI-generated information in legal briefs.
5. When you have questions about the risks of AI use: If you worry that if you use AI, you’ll plagiarize, lose client trust if people notice, or breach client confidentiality, it might not be for you.
- When AI use affects trust: Bynder found that 52 percent of consumers disengage from content they suspect is AI-generated.
A 2025 University of Kansas study found that news releases attributed to a human author were rated more credible and the organization more trustworthy than identical AI-attributed releases.
The researchers suggest AI content creators be transparent about its use, take responsibility for any mistakes, edit carefully, and prepare for reader questions or pushback.
AI can produce surprisingly empathic responses. In a study comparing brief responses from Bing Chat and anonymous online strangers, AI-generated replies made participants feel more heard. But their answers changed when they knew the reply came from AI.
Its empathic communication also isn’t rooted in feelings or experiences.
- When AI use creates privacy or confidentiality risks: Feeding AI nonpublic client data, like buying habits, can breach client confidentiality in certain industries.
- When AI use affects ownership: The U.S. Copyright Office says copyright doesn’t protect purely AI-generated material. An AI-assisted work might still receive protection when a human contributes original expression through creative selection, arrangement, or modification; prompts alone usually don’t establish human authorship.
3 Steps to Take Before You Cut and Paste
1. Check the tool’s current business/privacy terms, data-retention settings, vendor agreement, and your professional obligations.
2. Establish who may use AI, for which tasks, and who signs off before publication.
3. Unless your organization has approved the tool and its data settings, don’t enter confidential client information into a public AI tool, including:
- Personally identifying information (such as a name, address, or date of birth)
- Unpublished strategy
- Contracts
- Medical information
- Financial records
- Proprietary material
An enterprise or business tool might have different data controls and terms.
Should You Use AI For Writing?
AI delivers drafts in seconds. It can’t take responsibility for what those drafts say.
That’s still your job.
For important public-facing content like email campaigns, articles, website copy, or anything else tied closely to your reputation, the right judgment isn’t optional. It’s the quality-control step that protects your brand image.
Let me be the human in your content process, making sure your copy sounds like you to enhance trust and credibility.
Do you use AI for your business writing? If so, how? Share your experience in the comments.
QUOTES
“Where AI was found to come up short is in critical thinking and logic, originality, and contextual judgment, with perception studies showing human‑written content scoring dramatically higher on trust and brand resonance than AI‑generated pieces.”
~ David Deutsch, This Beats Both Human Copy and AI Copy
