We are at a pivotal moment for our society, where the AI revolution will impact every aspect of people's lives. Image-generating AIs often perpetuate gender stereotypes, and this can have a significant impact on the way people perceive certain professions. In particular, the underrepresentation of women in tech jobs can be exacerbated by AI-generated images that depict men as the primary workers in the industry, despite 26% of the tech workforce being comprised of women.

Telekom as one of the biggest IT employers in the region is at the forefront of applying and advocating for gender equality norms. They wanted to make their brand attractive for female tech professionals by showing their appreciation for their expertise and using technology to make a positive change in society.

If machines can’t reach the right data out there, what if it was donated?

In an industry-wide effort, Telekom, one of the biggest IT employers in the region, asked the women representing 26% of the tech industry workforce to make a sacrifice:

donate their personal, professional data for the cause in order to build a re-training dataset for image-generating AI models.

The issue of this gender bias lies in how AI models are trained. They feed on data scraped from the public internet, which is not a representative sample of the population, but an ecosystem that strengthens our stereotypes and biases.

Women making up 26% of the tech workforce might be far from equality, but it is a great enough number to provide a massive amount of data that can fill the gaps of AIs’ knowledge. So we invited these women to be a part of the solution.

In the age of super effective algorithms we solve the problem „manually,” not only to provide a practical solution but also to bring to people's attention how many women are affected by this issue in their daily lives.

We launched the data donation initiative on LinkedIn with Ida Urbán, a Hungarian gender equality activist in tech. The first data donation happened live at an industry event for girls and women interested in STEM. It got amplified with a PR, social and influencer campaign on LinkedIn as the main platform.

On our campaign website, women could easily sign in with their LinkedIn profiles and donate their pictures and job titles. We made sure that the women who donate all know that their data will be used ethically, upholding the strictest GDPR standards.

To create the Dataset for Equality, we collaborate with vAIsual, a company that specializes in building ethical datasets for image-generating AI models. The data donation period started in April and will be open until December 30th, 2023. The dataset will be available for free to AI companies through datasetshop.com.

Results are coming in as the campaign was launched on the 23rd of April.

Based on the first wave of donations the campaign is projected to reach the effectiveness target of the dataset well before the end of the given timeframe.

Without any paid media, the data donors’ proud social media posts are providing an industry-wide reach and visibility for the cause.

These posts are the continuous source of organic earned media and generating traffic to the website with a click-through rate well above category benchmarks.