The robot or the brain? Building a classifier for visual news frames of Artificial Intelligence

Automated method for detecting visual frames in online images of AI.

Artificial intelligence (AI) is often portrayed with brains or human-like robots in media or user-generated online images. These images are powerful because they grab readers’ attention, stick in their memory, and appeal to their emotions. As a result, the visualizations of AI can influence how people think about this new and rapidly evolving technology. These visual frames put AI in a specific light and can shape people’s understanding of it. This project was designed to study and describe the visual frames contained in online images of AI.

What are we trying to achieve and how?  

Since the number of images online is large and growing, an automated method is necessary to analyse them. While commercial solutions typically focus on recognizing objects in images, this project’s goal was to automatically detect a more abstract concept: the portrayed meaning of the image, or the visual frame. To do this, communication scientists and data scientists worked together to develop a custom-trained image classifier that can automatically categorize images of AI into different visual frames such as “thinking machine” or “collaborative AI”.

The software can be used by academics and practitioners alike who want to explore how AI is depicted in user-generated, stock, or news images, and who want to understand which visualizations of this new technology are dominant online.

Curious to learn more about how research software can support research?

Our team of experts are involved in a wide variety of projects across all scientific domains. Take a look at some of the other projects we’re involved in to learn how our research software supports research.


Logo Vrije Universiteit Amsterdam