Ctrl.Alt.Img installation

The future is being visualised by artificial intelligence. Many of the images we see every day in the media are increasingly being created by algorithms and image-generation apps. Image banks, big news outlets and social media companies are now turning to technologies like Dall-E 2 to visualise the world around us. This development has an enormous influence on visual culture and our collective consciousness, but what will the world look like through the eyes of AI? And will the average person have any influence on the outcome?  And, if new machine learning tools are being trained on existing datasets with existing social biases, then how can we future-proof AI images to be more inclusive?

Ctrl. Alt.Img is a creative intervention that uses existing AI tools like text-to-image apps to explore different approaches to image making. Our goal is to offer an alternative image database generated by the collective action of the public to broaden the scope of who and what we see in AI-generated images. Ctrl.Alt.Img takes practical steps with our partners to create more inclusivity and nuance in the practice of image making. 

EVERYONE IS INVITED to the Ctrl.Alt.Img interactive installation 2023 and a series of image making workshops taking place in 2024 in Rotterdam.  

Participant Experience in the Interactive Installation 

The installation takes the form of a photo booth, inviting the participant into a private space. The booth is a  colourful wooden structure with a curtain covering the entrance. 

Once inside the booth the participant takes a seat in front of a screen fitted with a camera. The Ctrl.Alt.Img application then captures a photo of the participant for analysis. This image is analysed for gender, age and ethnicity by a machine learning application we have custom built. The results are displayed on the screen in front of the participant.

The participant is then invited to adjust these results for accuracy using the touch screen in front of them. In our initial tests this moment has proven to be quite provocative as the machine’s bias and very simple reading of the participant becomes evident. In the next step the participant can also add context details to their description that are important to them in the way they identify themselves, such as “wearing a headscarf”, “migrant”, “teacher”, “mother”.

The adjusted data is then fed back into our machine learning application which generates a photorealistic image of the participant using DALL-E 2.

Various snippets of footage are shown to participants throughout the experience, revealing the thoughts and ideas about identity and representation from people we interviewed on the topic.

Ultimately, the participant is then presented with a print out of three images on one photosheet - 1.) a photo of themselves 2.) a DALL-E 2 image based on the data of image 1  3.) another DAll-E image this time based on the adjusted data.

affect lab is collaborating with documentary house  Prospektor, photographer Cigdem Yuksel and creative technologist Babusi Nyoni (Triple Black) to realise Ctrl. Alt. Img.

Background 

For the past 3 years affect lab and Prospektor have been working on separate creative and technology projects that focus on representing minority groups.  Affect lab’s project How To Tell a Herstory is a unique platform for telling stories about migrant women, using machine learning, the family photo album and community-based workshops. At the same time, Prospektor has been working on a project titled Moslima  initiated by photographer Cigdem Yuksel, that addresses the stereotypical image of Muslim women in the media and how this can be creatively addressed. Both projects have given us deep insights and experience in the current biases and opportunities that are embedded in visual archives. Creative technologist Babusi Nyoni’s work includes the development of the storytelling tool for How To Tell a Herstory and the creation of the first machine learning model that can accurately identify black girl hairstyles in a browser. Affect lab, Prospektor and Babusi Nyoni are now joining forces to develop Ctrl.Alt.Img to address the future of image making.