Nobody Wants an AI Spokesperson Who Looks Suspiciously Like Them

Zero. That’s how many hard numbers I could find in the current round of reporting on how people actually want AI used in advertising next year. Plenty of panels, plenty of executive tip lists, plenty of predictions about AI-powered search and chatbots doing heavy lifting for consumer research by 2026. Actual measured consumer sentiment with a figure attached? Not in what’s circulating. The sources I reviewed stop short of answering the question directly.

That gap matters, because the advertising industry is currently building 2026 strategy on top of it.

The one finding that keeps surviving contact with reality

Strip away the conference talk and there’s a single consistent signal: people are fine with AI helping make an ad. They get noticeably less fine when AI replaces the humans in it. The reporting on American attitudes lands on exactly that split — AI as a creative tool is tolerable, AI standing in for someone’s image and voice is not.

This is a narrower permission than most vendors are selling. “Consumers accept AI in advertising” and “consumers accept AI-generated people in advertising” are two completely different claims, and the second one is where the discomfort lives. If your product pitch involves synthetic presenters, cloned voiceover, or a face that never existed reading your script, you are operating on the wrong side of that line.

The generational twist nobody put in the deck

The part I expected to be wrong about: younger adults are less accepting of AI in advertising, not more. The received wisdom in marketing has been that digital-native audiences shrug at automation and older audiences are the holdouts. The reported divide runs the other direction.

Think about what that does to a media plan. The audience your brand is spending the most to reach is the one most likely to clock an AI-assembled spot and hold it against you. Youth-targeted campaigns are typically where agencies push hardest on volume, speed, and cheap variation — which is precisely what generative tools are good at. The efficiency and the audience tolerance are pointing opposite ways.

What this means if you’re buying tools

I review this category for a living, and the market is splitting into two product shapes that get marketed identically:

• Tools that compress the boring parts. Automating repetitive production work, assembling variants, speeding up insight from messy data, handling the fragmented customer journey that marketers are now expected to cover with fewer people. This is the safe side of the consumer permission line.
• Tools that manufacture the human parts. Faces, voices, performances, likenesses. Technically impressive, commercially risky, and sitting right where sentiment turns negative.

Both get sold with the same language about creative output and speed. You have to read the demo, not the landing page. Ask what the tool produces at the final step: an asset a human directed, or a human the machine invented.

The prompt-laundering problem

One detail from the industry chatter stuck with me. In a 2026 ad-creative discussion, the practical advice offered was to take a reference image, ask a model to write a prompt describing a person who looks very similar, and generate from that — specifically so you’re not copying someone directly.

That’s not an ethics framework. That’s a workaround with a rationalization attached. The intent is to get the likeness without the liability, and dressing it up as prompt engineering doesn’t change what’s happening. If a technique exists mainly to create distance between you and a real person’s face, that’s the signal, not the solution. Expect audiences and regulators to read it the same way eventually.

The advice that’s actually holding up

The guidance I keep seeing from people running real marketing organizations is unglamorous and correct: treat AI as a tool that enhances creativity and accelerates insight, not as a replacement for the people doing the work. Keep it human-led. Keep it ethical. It sounds like boilerplate until you notice it’s the only position that matches what consumers reportedly want.

My read for anyone planning next year: use these tools aggressively on the work nobody wants credit for. Research synthesis, variant production, campaign ops, the endless reformatting that eats junior hours. Keep humans visibly in front of the camera and behind the concept. Not because machines can’t fake it — they obviously can — but because the audience you most want to reach appears least willing to forgive it.

And when a vendor tells you the market has already accepted synthetic people in ads, ask them for the number. Based on what’s publicly available right now, they don’t have one either.

🕒 Published: October 6, 2026