Car Brand Drivers According to AI

For this campaign, we earned links from

Need a Digital PR Campaign? Contact Us

Campaign Idea

While many AI researchers were examining general capabilities, our team took a unique approach and designed a unique experiment to uncover specific biases in image generation systems that were usually based on specific stereotypes.

Our team spotted that image generators were tending to have racial and gender biases towards “typical” drivers for certain car brands when asked to create images of drivers associated with specific car brands.

We discovered immediate news potential when every single AI-generated driver image showed strong gender and racial biases.

The timing aligned perfectly with growing concerns about AI ethics and representation issues in generative systems.

Campaign Research

Our team executed a methodical approach to uncover systematic biases in popular AI systems.

We began the research by prompting ChatGPT to describe what it considered to be the "typical" drivers for 10 iconic car brands, ensuring a wide representation of popular models.

Next, we fed these detailed descriptions into Midjourney, using them to generate visual representations of the drivers as envisioned by the AI.

By the time mainstream media began addressing bias in AI, we had already documented clear and compelling evidence of gender exclusion and racial stereotyping in the generated images, providing proof of these issues in generative AI systems.