Back to the blog
Exclusive9 min read

People Spot an AI Poster at First Glance. You Won't Spot an AI Insight

Michal Krčmář · founder Don Marketer
People Spot an AI Poster at First Glance. You Won't Spot an AI Insight

People Spot an AI Poster at First Glance. You Won't Spot an AI Insight

Can you make marketing decisions based on AI? As a tool, yes. As the referee, no. AI is great for first drafts, research, prototypes and finding holes in your own thinking. But decisions need to rest on data from real people. Research by Prolific (2026) showed that AI-generated synthetic personas drift toward the most common, stereotyped answers, and no method matched the accuracy of a second sample of real humans. And customers notice AI output and punish it: content they believe was made by AI gets rated 48% less trustworthy.

Calculate it yourself - takes 2 minutes

The article continues below - unlock it free as an early adopter.

Exclusive

The rest is for early adopters

Until 20 January 2027 you get free access to every exclusive article in exchange for your contact - as an early adopter you also get 50% off the first year of the subscription once we launch it.

Confirm your email and get 20% off any guide

Privacy policy · Terms & conditions

FAQ

Can I make marketing decisions based on AI?

As a tool, yes. As the referee, no. AI is great for first drafts, research, prototypes and finding weak spots in your reasoning. But decisions should rest on data from real customers. Research by Prolific (2026) found that AI-simulated respondents didn't match the accuracy of a real human sample.

What is AI slop?

Cheap content that AI produces in bulk without human care: generic images, copy and ads. People often recognize it at first glance. NielsenIQ (2024) found that AI-generated ads come across as annoying, boring and confusing and trigger weaker memory activation than traditional ads.

What is insight slop?

Research outputs and reports stitched together from buzzwords, trends and data points without deeper context or practical use. Lore Oxford of Reddit used the term on the WARC Podcast in October 2026. It's dangerous because it looks convincing and budgets get decided based on it.

Do AI synthetic personas work for market research?

Only directionally. According to a Prolific study (2026), demographic personas collapse into the most common, stereotyped answers and roughly triple the error. The authors recommend them for exploratory research, not as a replacement for real respondents.

Do customers mind AI-generated content?

Often, yes. According to Raptive (2025), people rate content they believe is AI-generated 48% less trustworthy, and purchase consideration for advertised products drops 14%. A Washington State University study (2024) found that just using the term "artificial intelligence" in a product description lowers purchase intent.