Products SEO & GEO Agent Paid Ads Agent Website Agent Conversion Agent
How it works Pricing FAQ Blog Book a call

AI marketing limits: where it falls short on creativity and voice

Updated · 7 min read

Short answer AI is fast at drafts, variations and data, but it can't supply what makes marketing work: an idea nobody else has, your voice, your strategy, or facts it wasn't given. It also carries no blame. Tribunals and regulators hold the business responsible for what its AI says, and in the US purely AI-made content can't be copyrighted.

Want help applying this to your business? Talk it through on a free 30-minute call.

Book a call

Where does AI marketing fall short?

AI falls short wherever the job needs something it doesn't have: a new idea, your judgment, facts about your business, or responsibility for the result. It does well wherever the job is volume: drafts, ad variations, summaries, sorting data.

LimitWhy it happensWhat it costs youKeep with a person
CreativityIt predicts likely words from what it learnedCopy that sounds like your competitors'The big idea, the offer, the hook
Brand voiceIt drifts back to an average toneYou stop sounding like youFinal edit of anything public
StrategyIt only knows what you tell itAdvice that ignores your margins, capacity and townWhat to sell, to whom, at what price
FactsIt can state wrong things with confidenceWrong prices, rules or promises in publicFact check before publishing
LiabilityThe law looks at the business, not the toolYou pay for what your AI saysSign-off on claims, reviews, chat answers
OwnershipUS copyright needs a human authorPure AI output may be free for anyone to copyLogos, slogans, signature content
PrivacyPrompts can leave your controlCustomer data in someone else's systemWhat data goes into a tool

Why does AI-written marketing sound like everyone else's?

Because it is built to produce the most likely next word, and the most likely words are the ones everyone already uses. Google puts it plainly: generative models "don't retrieve facts, but predict a likely sequence of words based on their training data" (Google Search Central: Guidance on using generative AI content).

Research shows the sameness problem directly. In a 2024 experiment published in Science Advances, writers who got story ideas from an AI wrote stories judged more creative, especially the weaker writers. The catch:

"However, generative AI–enabled stories are more similar to each other than stories by humans alone." (Doshi and Hauser, Science Advances, 2024)

For a business, that is the trade. AI lifts a weak draft to average. But if your competitors use the same tools on the same prompt, you all land on the same average. The idea that sets you apart (the guarantee nobody else offers, the story of why you started) has to come from you.

Why can't AI keep my brand voice?

It can copy a voice for a paragraph, but it can't hold one. Tell it "friendly and direct" and you get a generic friendly tone, not the way you talk to customers. Over a long piece, or a month of posts, it slides back toward the safe, polished middle.

Voice is also made of choices AI can't make for you: the jokes you'd never tell, the claims you refuse to make, the local names your customers know. Give the tool a written voice sheet and real examples, then edit every public piece yourself. The step-by-step method is in AI content creation that keeps your voice.

Can AI plan my marketing strategy?

It can list options; it can't choose for you. Strategy depends on facts the tool doesn't have unless you supply them: your profit per job, how many jobs your crew can take next month, which services you want more of, what happened in your town last week.

Missing context causes three common failures:

  • Generic plans. Ask "how do I get more customers?" and you get the same checklist any business gets.
  • Wrong facts stated as true. Google calls these "inaccuracies (also known as hallucinations)" and says: "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." (Google Search Central)
  • Bad timing. A scheduled promotion keeps running during a local emergency or a bad news week, because nothing told the tool to stop.

Use AI to gather and sort options. Decide yourself, then let AI carry out the decision. How that split works for search marketing is in AI vs human SEO.

Who is responsible when AI gets it wrong?

You are. Tribunals and regulators have not accepted "the AI said it" as a defense.

In Canada, a BC tribunal ruled against Air Canada in February 2024 after its website chatbot told a grieving customer he could claim a bereavement fare after travel, which the airline's own policy did not allow. The airline argued it wasn't liable for the chatbot. The tribunal:

"It should be obvious to Air Canada that it is responsible for all the information on its website. It makes no difference whether the information comes from a static page or a chatbot." (Moffatt v. Air Canada, 2024 BCCRT 149)

Air Canada was ordered to pay $812.02 (Canadian dollars), including damages, interest and fees.

In the US, the FTC took action against several AI schemes in September 2024, including Rytr, an AI writing tool with a feature that generated customer reviews and testimonials. Then-Chair Lina Khan said:

"The FTC's enforcement actions make clear that there is no AI exemption from the laws on the books." (FTC, September 25, 2024)

So any ad claim, price, guarantee, review or chatbot answer your AI produces is your claim. Rules on fake reviews are in AI-generated reviews; platform ad rules are in Advertising policies.

Do I own what AI creates for me?

In the US, not if AI did all the creative work. The U.S. Copyright Office's January 2025 report says AI output is protected "only where a human author has determined sufficient expressive elements," such as creative changes or arrangement, "but not the mere provision of prompts" (U.S. Copyright Office, January 29, 2025).

What this means in practice: a logo, slogan or mascot generated by AI from a prompt alone may be free for a competitor to copy. If you want to own it, a person needs to shape it, and a lawyer can tell you whether yours meets the bar.

What should I never put into an AI tool?

Customer personal information you don't need to share. What happens to your prompts depends on the plan:

  • ChatGPT, personal plans: conversations can be used to train OpenAI's models unless you turn off the "Improve the model for everyone" setting (OpenAI Help Center: Data controls FAQ).
  • ChatGPT business plans: "By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models." (OpenAI Help Center)

Canada's Privacy Commissioner advises organizations to "use anonymized or de-identified information within prompts" where possible, and adds that "accountability for decisions rests with the organization," not the AI system (Office of the Privacy Commissioner of Canada, December 2023).

Before pasting in a customer list or a client's file, remove the names, or don't paste it.

Example

A roofer's ad copy (US dollars)

The roofer asks AI for ten Google ad headlines. All ten could belong to any roofer in the state. He keeps the AI's structure but swaps in what only he offers: a written 10-year workmanship warranty and same-week inspections. The AI made the drafts in a minute; the reason to call came from him.

A Vancouver bookkeeper's chatbot (Canadian dollars)

Her new website chatbot tells a visitor that year-end packages start at $500. Her real price is $900. Under the Air Canada ruling, a customer could reasonably rely on that answer. She limits the bot to her published price page and has it hand anything else to her inbox.

A café's AI logo

The owner generates a logo from a prompt and prints it on cups. A nearby shop starts using a near-copy. Because no person shaped the design, a US copyright claim on it is weak. She pays a designer to redraw it with her own changes.

FAQ

Will AI replace my marketing person?

It replaces hours of drafting and data work, not judgment. Someone still has to decide the offer, check facts and approve what goes out.

Is it fine to use AI if I edit everything?

Yes. That is the fix for every limit here: someone who knows the business checks claims, voice and data before anything goes public.

Can AI write my reviews or testimonials?

No. The FTC acted against Rytr for a tool that generated customer reviews. Reviews must come from real customers about real experiences.

Can AI handle my customer replies?

For routine questions, yes, if it answers only from your approved information. Complaints and anything sensitive should go to a person.

Different for you?

Keep learning

NextLLMs in marketing: practical uses for small businessesWhere large language models like ChatGPT, Claude and Gemini help a small business with content, SEO and ads, what they get wrong, and what to keep out of them.

Ask us one question

One line is enough. We reply by email.

Sources

  1. Doshi and Hauser, Science Advances: Generative AI enhances individual creativity but reduces the collective diversity of novel content (July 12, 2024)
  2. Google Search Central: Guidance on using generative AI content
  3. Civil Resolution Tribunal (BC): Moffatt v. Air Canada, 2024 BCCRT 149 (February 14, 2024)
  4. FTC: FTC announces crackdown on deceptive AI claims and schemes (September 25, 2024)
  5. U.S. Copyright Office: Copyright Office releases Part 2 of AI report (January 29, 2025)
  6. OpenAI Help Center: Data controls FAQ
  7. Office of the Privacy Commissioner of Canada: Principles for responsible, trustworthy and privacy-protective generative AI technologies (December 7, 2023)