liveBook · Manning

Ad turing test

As explained in Introduction to Generative AI, Second Edition

An ad Turing test is an evaluation in which people attempt to distinguish advertisements created by humans from advertisements generated by artificial intelligence.

An ad Turing test asks people to distinguish advertisements created by humans from advertisements generated by artificial intelligence. It tests whether machine-produced creative work appears recognizably different from human work. section 6.5

In a 2023 competition, advertising experts classified 10 posters for the same fictional energy drink with 57% accuracy. A 300-person audience reached 53%, close to random guessing. The results suggest that generated advertisements can be difficult to identify by appearance alone. section 6.5

The test also compared practical advertising performance:

  • AI-generated ads averaged 1.83 on System1's emotional-response rating scale, slightly below the 1.9 national average for print advertisements. section 6.5
  • AI teams could produce concepts more quickly and potentially at lower cost, while the competition restricted them from altering the model's output. section 6.5

Buzz energy drink poster

An example uses Stable Diffusion to generate a poster from the prompt “poster for a new energy drink called Buzz.” section 6.5

How Introduction to Generative AI, Second Edition covers it

How deep it goes
Defines the test and reports identification and ad-rating results, but does not explain a formal testing methodology.
Examples
Uses ten fictional energy-drink posters, a Stable Diffusion Buzz prompt, expert judgments, audience judgments, and System1 ratings.
What you can do after
Evaluate whether generated advertisements appear human-made and compare their measured emotional response with print-advertising benchmarks.
How it connects
Links artificial intelligence, large language models, image generation models, advertising, and emotional-response measurement.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Introduction to Generative AI, Second Edition. The section links point to the parts of the book it draws on.