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

AI and human collaboration in marketing: who does what

Updated · 6 min read

Short answer Let AI do volume: first drafts, variations, summaries, sorting data. Keep people on what needs judgment or proof: offers and prices, facts, real customer stories, replies to upset customers, and the final yes. Google asks for a human fact-check of AI content before publishing, and OpenAI's terms say its output may not always be accurate.

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

Book a call

Who does what: the split for each marketing job

AI makes the first version fast; a person decides what is true, what to offer and what goes out. That holds for every channel, not just blog posts.

JobHand to AIKeep with a person
AdsHeadline and description variations, audience ideas, first look at search termsThe offer, the price, the budget, any claim ("licensed", "same day"), final approval
Website pagesOutline, first draft, title and meta description optionsServices you really offer, prices, service area, photos of real jobs
Email and textsSubject line options, a draft from your bullet pointsWho gets it, what you promise, consent to text people
Social postsCaptions, a month of post ideas, resizing one idea for each platformPhotos, anything about a real customer, replies in comments
ReviewsSorting reviews by theme, a draft reply to a happy customerReplies to unhappy customers, and never writing reviews
Website chat or AI receptionistAnswering hours, location, booking stepsQuotes, refunds, anything legal or medical, upset callers
ReportsPulling numbers, a first summaryDeciding what to change next month

The rule behind the table: if a mistake would cost money, break a law or upset a customer, a person signs off. If a mistake would only be dull, AI can carry it.

Why people must stay in charge

AI tools predict likely words; they don't check facts. Google's advice:

"It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." (Google Search Central: Guidance on using generative AI content)

The tool makers say the same. OpenAI's terms tell users:

"You must evaluate Output for accuracy and appropriateness for your use case, including using human review as appropriate, before using or sharing Output from the Services." (OpenAI Terms of Use)

So the business that publishes the ad, page or reply owns what it says. AI wrote it; you sent it.

Google doesn't punish AI help itself. Its 2023 guidance promises to reward high-quality content "however it is produced". What it does call spam is using automation, AI included, mainly to manipulate rankings, such as pages made at scale with no added value.

How to edit an AI draft: five checks

Run every draft through these five checks before it goes out. A short ad takes two minutes; a service page, about 20.

  1. Facts. Circle every number, name, date, price, law and claim. Check each one against your own records or the official source. AI fills gaps with plausible guesses (hallucinations: confident statements that are false).
  2. Offer. Is this what you actually sell, at the price, area and timing you actually deliver? AI often promises "24/7" or "free estimates" because competitors do.
  3. Proof. Add one thing only you know: a real job, a real photo, a number from your own work. This is what Google means by experience, and no tool has it.
  4. Voice. Read it out loud. Cut words you'd never say to a customer ("elevate", "seamless", "in today's fast-paced world"). Shorten long sentences.
  5. Small print. Titles, meta descriptions, image alt text (the hidden description of an image) and ad extensions count too. Google's review rule covers this metadata as well, because it shows up in search results.

If a draft fails check 1 or 2 in many places, write that piece yourself. Fixing a wrong draft takes longer than starting from your own bullet points.

For a one-page voice sheet to feed the AI before it drafts, see AI content creation that keeps your voice.

What AI must never do for you

Some jobs aren't a quality risk; they're a legal one.

  • Write reviews or testimonials. The FTC's rule on fake reviews, in effect since October 21, 2024, bans reviews that pretend to come from someone who doesn't exist, which the FTC says includes AI-generated fake reviews. Courts can impose civil penalties for knowing violations.
  • In Canada, the Competition Bureau warns that "Anyone who writes or permits writing reviews that give a false or misleading impression to consumers could be liable" under the Competition Act. See AI-generated reviews.
  • Invent an author. Google calls made-up creator profiles, "such as by using AI-generated headshots, made-up names, or false credentials", a form of deception.
  • Make claims you can't back up in ads: licenses, guarantees, "best in town", health or money outcomes. See advertising policies.

Do you need an AI humanizer?

No. A humanizer is a tool that rewrites AI text so it reads less like AI. It changes the wording but checks nothing: wrong prices stay wrong, and the made-up claim stays made up. Google judges whether a page is helpful and accurate, not who or what typed it. Your five-check edit does what a humanizer promises and fixes the facts too. More in AI detectors and SEO.

Should you tell customers you used AI?

Usually not for routine copy, but yes where a reader would wonder. Google's guidance:

"Overall, AI or automation disclosures are useful for content where someone might think 'How was this created?' Consider adding these when it would be reasonably expected." (Google Search Central: Creating helpful content)

Examples where it's expected: an AI chat on your website (say it's automated and how to reach a person), or an image that looks like a real job but was generated.

Example

A plumber's Google ads (US dollars)

AI writes 15 headlines in a minute. The owner deletes three that promise "free same-day service" (he charges a $90 call-out fee and books next day) and approves the rest. Budget and bids stay his call.

A Vancouver bookkeeper's newsletter (Canadian dollars)

She gives the AI five bullet points from client questions that month. The draft cites a filing deadline from the wrong year; check 1 catches it. Her edit adds one story from a client file, with no name. Twenty minutes, instead of two hours.

A dental office's review replies

AI drafts replies to 40 five-star reviews; the office manager approves them in batches. The two one-star reviews she answers herself, without naming patient details.

FAQ

What percentage of marketing work can AI do?

There's no fixed share. Most first drafts and variations can go to AI; every final decision and every fact stays with a person.

Can an agency do the human part for me?

Partly. An agency can edit for voice and SEO, but the owner or a senior staff member should confirm prices, services and claims.

How do I stop AI copy sounding generic?

Give it your facts before it writes: real prices, real jobs, words customers use. Then cut the stock phrases in the voice check.

Is it safe to let AI post on social media by itself?

For posts you already approved, yes. Replies to comments and messages need a person: an AI answer to an angry customer is still yours.

Different for you?

Keep learning

NextHow to use AI for marketing: a 7-step workflow for small businessA step-by-step AI marketing workflow for owners: pick one goal, feed AI your facts, research, draft, review, automate hand-offs, and measure what changed.

Ask us one question

One line is enough. We reply by email.

Sources

  1. Google Search Central: Guidance on using generative AI content
  2. Google Search Central: Creating helpful, reliable, people-first content
  3. Google Search Central Blog: Google Search's guidance about AI-generated content (2023)
  4. OpenAI Terms of Use (effective January 1, 2026)
  5. FTC: Final rule banning fake reviews and testimonials (2024)
  6. FTC: Consumer Reviews and Testimonials Rule, Questions and Answers
  7. Competition Bureau Canada: Online reviews posted by employees (2024)