Human or AI? : Can you recognise AI-generated faces?
The revolutionary study « Human or AI », initiated by ContentDetector.AI, has revealed captivating insights into people’s ability to distinguish between real images and images generated by artificial intelligence (AI). This analysis focuses on public perception and awareness of DeepFakes, an increasingly relevant topic in our digital world.
The main objective of this research was to assess the ability of everyday users to differentiate AI-generated faces from those of real humans. This is part of a wider effort to understand public perception of AI-generated visual content.
For this study, ContentDetector.AI carefully selected a set of 40 images, comprising 20 authentic human faces and 20 faces generated by an AI image generation model. The study involved a social game where participants were confronted with five images and had to identify which were real and which were generated by AI.
The results of the study were as follows:
- Of the 3,380 participants, only 2.9% were able to correctly identify all five images.
- 12.5% were able to correctly identify four or five out of five images.
These results underline the significant progress made by AI technology in creating images that are close to reality, making it difficult for the general public to differentiate between deepfakes and real images. Note also that the game attracted 16,900 responses in just two weeks.
These results clearly demonstrate the need to develop advanced AI detection tools and laws to mark AI-generated content, in order to combat the proliferation of deepfakes in social media.
The « Human or AI » study offers an essential perspective on the challenges posed by the distinction between real and AI-generated content. It highlights the importance of staying informed and adapted to the rapid advances in technology in our interconnected world.
