This work uses a new technology (A.I. text-to-image generators) to showcase fundamental issues on this technology itself.

After testing A.I.s to see how inclusive they are, we created an impactful awareness campaign to let everybody know about their biases.

Wired, whose mission is to raise awareness about new techs and their effects on culture, has been amongst the first to embrace A.I.s

That’s why they wanted to make their voice heard on A.I., the topic of the moment, and asked us to create a significant and impactful project on it.

One of the main values of the brand is equality, so we tested A.I.s to check how inclusive they were. 

We inserted some simple prompts into an A.I. text-to-image generator, with some shocking results. The bias-ridden images the A.I. created became the visuals of our campaign. 

The Italian context makes the topic even more crucial: compared to other highly developed countries, Italian society is plagued by more racial, gender, and sexual orientation prejudices.

The debate sparked by the campaign reached the main channel of RAI, the public broadcasting company. The importance of RAI for Italian culture can be summarized by its nickname: Italians call it “Mamma RAI”.

We gave an A.I. text-to-image generator some simple prompts, such as /imagine a manager, a thug, two lovers. The results? It would inevitably perpetuate stereotypes. All the managers it imagined were male, all the thugs were people of colour, all the lovers were heterosexual. The bias-ridden images the AI created became the visuals of our campaign, which ran on many different media. 

30 January 2023 - Launch video

From 31 January 2023 until 14 February 2023, we released a new campaign subject per day.

A.I. text-to-image generators are AI-powered tools that take a text prompt, process it, and create an image that best matches the description given in the text prompt.

If the data they are trained with is biased, A.I.s end up being biased themselves, perpetuating old standing stereotypes about gender, sexuality, and race.

We used A.I. text-to-image generators themselves to showcase their own biases, urging their programmers to solve this problem as quickly as possible and making public opinion aware of the issue.

The campaign was overwhelmingly successful in its objective - raising awareness on A.I. bias. People were engrossed by the the project, despite the complexity of the topic. Here are some of the results it generated:

+200K organic interactions

+130% compared to other content on A.I.

15,5 Million potential impressions

77% net sentiment