Direct answer: AI reliably handles marketing production and digestion: first drafts, summarising research, generating options, reformatting content, and synthesising large volumes of text. It cannot reliably supply verified facts, mathematical accuracy, customer empathy, original positioning, or brand judgment. The boundary is jagged, so the only trustworthy map is one you test on your own tasks.

The mistake most marketers make isn't overestimating AI or underestimating it. It's assuming the boundary is intuitive. It isn't, and there's good research on exactly how weird it is.

Why is AI's capability "jagged"?

In 2023, researchers from Harvard, Wharton, MIT and BCG ran a field experiment with over 750 consultants. On tasks inside AI's capability frontier, consultants using AI completed 12.2% more work, 25.1% faster, at over 40% higher quality. On a task designed to sit just outside the frontier, one that looked similar, AI users were 19 percentage points more likely to reach the wrong answer.

The frontier doesn't follow difficulty as humans perceive it. AI can write a competent strategic summary (feels hard) and fail basic arithmetic on your campaign data (feels easy). That's why memorising capability lists is fragile: the useful skill is testing tasks yourself.

What marketing tasks sit inside the frontier?

First drafts of almost anything: emails, posts, briefs, outlines. Summarising and synthesising: reports, transcripts, survey open-ends, review sites. Generating options: ten subject lines, three angles, five structures. Reformatting: blog to social, long to short, one channel to another. Digestion at scale: a year of support tickets into themes in an afternoon.

Common thread: pattern work. That's the machine's home ground.

What should stay human?

Anything needing current verified facts without a checking step (AI states falsehoods fluently). Numbers that matter. Genuine customer empathy: AI has read everything and met no one. Original positioning: AI gravitates to the plausible average, and average positioning is no positioning. Brand judgment and taste. Final decisions with consequences, because accountability doesn't delegate.

How do you map the frontier for your own role?

List eight recurring tasks from your week. For each, run a quick test: give AI the task with a proper brief, twice, and judge honestly against your own standard. Mark each task inside, outside, or "mine to own". You now hold something more valuable than any tool list: a tested capability map for your actual job, ready to re-check as models improve, because the frontier moves.

Sign up to Marketing and AI: Foundations Module 1 Free

FAQ

Can AI do marketing strategy? It can widen and stress-test strategy: generating options, scanning competitors, arguing against your plan. It should not choose the strategy: AI defaults to the statistical average, and trade-offs need a human owner.

Is AI accurate enough for marketing content? For drafts, yes. For publication, only after verification: every factual claim needs a source a human has opened. Fabricated statistics and citations remain a standing failure mode of all current models.

Does AI work for small business marketing? Yes, and often best: the production bottleneck hits small teams hardest, and AI removes it. The same capability map applies at any size.