The three-question creative test
Use Figure 1 to separate the message, the audience's view of its production and the outcome worth measuring. This is Andrew's Intel editorial synthesis, not a validated scoring instrument.
"Made without AI" is provenance, not proof that the work is good or commercially effective.
Neither is an image generator. Both can produce dreadful advertising, although only one currently receives moral credit for doing so.
The practical instruction for managers is simple. Judge creative work on three separate grounds:
- Does it communicate effectively?
- Does the way it was made matter to this particular audience?
- Is there evidence that it changed a useful business outcome?
Do not allow a strong answer to one question to conceal a weak answer to the other two.
That sounds obvious. Yet the increasingly emotional argument about AI-made advertising encourages businesses to judge the tool instead of the work.
A real reaction, wrapped in a larger claim
An ABC News report from Toowoomba, Australia, offers a neat example.
One local creative, who readily says they are not a graphic designer, offered to make hand-drawn posters for small businesses and community groups for $5, or free when necessary. They reportedly received nearly 30 enquiries. A nearby bookshop and wine bar also apologised for using AI-generated promotional material after an author challenged the practice, then adopted an anti-AI policy and reported a positive response. 1
The events are interesting. The implied commercial lesson is much less secure.
Thirty enquiries for almost-free work do not show that the finished posters attracted customers. A favourable reaction to a public apology does not establish higher attendance, increased sales or stronger retention. One author's objection may be commercially important to a bookshop that depends on relationships with authors, but it is not evidence of a general consumer revolt.
The story demonstrates that an anti-AI position can attract attention and goodwill within a particular creative community. It does not demonstrate that conspicuously handmade advertising works better.
The backlash is real, but segmented
Businesses should not dismiss the underlying concern. Some consumers do react negatively to AI-generated advertising, particularly when its use appears deceptive, cheap or indifferent to human creative work.
In a 2026 IAB survey of 505 US Gen Z and Millennial consumers, 39 per cent of Gen Z respondents reported negative feelings towards generative-AI advertising, compared with 20 per cent of Millennials. Advertising executives were also much more optimistic about consumer attitudes than the consumers themselves. 2
But this was not a universal rejection. Forty-five per cent of the surveyed consumers felt positive about AI-generated advertising. Seventy-three per cent said knowing that an advertisement was made with AI would either increase or make no difference to their likelihood of buying. 2
The sensible lesson is not "stop using AI." The IAB's own recommendations are to understand the audience, maintain high creative standards and disclose material AI use consistently. In other words: use judgement.
Experimental research makes the picture more complicated still. A 2026 study of AI disclosure labels found that identifying an image as AI-generated could increase perceived novelty while reducing perceived authenticity. The effects varied with the product and the way consumers processed the advertisement. 3
That research used online experiments, largely with Chinese participants, so it should not be inflated into a universal law. It does, however, puncture the idea that customers have one predictable response to the letters "AI." 3
Human authorship is not quality assurance
Provenance can add value. Handmade packaging, locally produced artwork or a human-only creative policy may signal care, solidarity or membership of a community. For an independent bookshop, craft event or artists' cooperative, that signal might be part of the offer.
It still has to work as communication.
Poor hierarchy does not become clear because it was drawn by hand. Illegible lettering does not become persuasive. A confused message does not gain commercial force from the absence of a prompt. Human work can be warm, distinctive and appallingly ineffective at the same time.
Nor does AI use automatically make advertising generic. Generic work usually reveals generic thinking: a vague brief, weak art direction, no editorial judgement and approval based on speed rather than effect. AI can make that failure cheaper and faster, which is not the same as causing it.
The useful comparison is therefore not AI versus human. It is careless work versus directed work.
Goodwill is not yet a business result
The easiest numbers to collect are often the least useful.
Enquiries from people interested in free creative labour are not customer conversions. Compliments on a handmade poster are not event attendance. Social approval for an ethical policy is not necessarily repeat business. Each may be an encouraging intermediate signal, but none should be reported internally as proof of commercial success.
Before a campaign runs, decide what result would count. That might be bookings, qualified leads, footfall, sales, repeat visits or a measurable change in trust among a defined audience. Then compare the creative options against that result.
If "human-made" is part of the positioning, test that proposition too. Do target customers notice? Do they care? Does disclosure help? Are they willing to choose, recommend or pay more because of it?
The three-question creative test
Use this before approving AI-assisted work or rejecting it on principle.
On a narrow screen, each question is shown as a separate card.
| Review question | What to inspect |
|---|---|
| Does the work communicate? | Can the intended customer quickly understand what is being offered, why it matters and what to do next? Judge hierarchy, clarity, distinctiveness, credibility and fit with the brand. |
| Does provenance matter to this audience? | Would material AI involvement change trust or buying behaviour in this category? A book launch, healthcare advertisement and industrial-parts catalogue may produce very different answers. Decide whether disclosure, limited use or a human-only promise has genuine audience value. |
| What outcome improved? | Specify the behaviour that matters and measure it. Treat enquiries, likes and compliments as signals, not final results, unless attention itself was the objective. |
This test leaves room for a deliberately rough, handmade poster to succeed. Its informality might be exactly right for a community craft evening. It also leaves room for well-directed AI-assisted work to outperform a more expensive human production. The answer belongs to the audience and the evidence, not to the tool's supporters or opponents.
What to remember
There is a genuine commercial signal inside the anti-AI reaction. Authenticity, disclosure and creative provenance matter to some customers, in some categories, some of the time.
But "made by a human" is not a performance metric.
The opposite of AI slop is not human slop. It is good work, made with judgement, for a known audience, and measured against a real objective.
Decision rule: Judge advertising by whether it communicates, whether its production method matters to the intended audience, and whether it changes a meaningful business outcome.
Sources and evidence
The ABC report establishes a local reaction, not campaign performance. The IAB report describes surveyed attitudes in specified US groups; the online experiments study stated attitudes and intentions, not realised purchases.
Source posture: Nearly 30 inquiries were for a $5-or-free poster offer, not customer conversions. The bookshop’s positive response was attributed to its owner and lacks a controlled before-and-after sales measure. IAB findings are survey responses from 505 Gen Z and Millennial US consumers and 104 advertising executives, not a universal audience measure. Shi and Jiang tested disclosure in online experimental settings, largely with Chinese participants; outcomes include attitude and purchase intention, not observed sales. The three-question test is editorial synthesis.
Source pages checked 30 October 2026.
- ABC News, 4 September 2026. The local events and attributed claims appear genuine, but the report supplies no sales, attendance, conversion or comparative performance evidence. Original source.
- IAB and Sonata Insights, 15 January 2026. Survey of 505 US Gen Z and Millennial consumers and 104 advertising executives. Useful evidence of segmented attitudes, not a global population estimate. Original source.
- Shi and Jiang, SAGE Open, 2026. Three online experiments on disclosure, novelty, authenticity and advertising outcomes. Useful evidence of competing effects, with limits on broad cultural generalisation. Original source.