In a never before seen feat, AI has been able to generate images based on what a particular person likes or finds attractive.
This has been possible by researchers at the University of Helsinki and the University of Copenhagen.
Researchers were curious to know if a computer could identify the facial features of a person that a particular individual finds attractive. To do this, they first collected brain signals from participants and fuse them with a brain-computer interface that would create the artificial matches, just like they do on popular dating apps like Tinder.
Just like the dating app, users were shown faces created by GANs or general adversarial neural networks. However, here instead of swiping left or right, the users were asked to focus on the fact that they were watching, while the brain’s response was being collected and monitored through its electric signals with the help of an electroencephalograph.
With this, the brain-computer interface gained an opinion on attractiveness, based on the user’s preference and generated an entirely new face (using GANs) that took pointers from what it learnt from the previous faces.
Also Read: Why the Cybersecurity Tech Industry Needs to Address the Gender Gap
These newly generated portraits were then shown to the users to see if the model world and the result was rather surprising. The new faces that the AI-generated matched the preferences of the user with 80 per cent accuracy.
Senior Researcher and Docent Michiel Spapé from the Department of Psychology and Logopedics at the University of Helsinki, explains, ‘The study demonstrates that we are capable of generating images that match personal preference by connecting an artificial neural network to brain responses. Succeeding in assessing attractiveness is especially significant, as this is such a poignant, psychological property of the stimuli.’
They added, ‘Computer vision has thus far been very successful at categorising images based on objective patterns. By bringing in brain responses to the mix, we show it is possible to detect and generate images based on psychological properties, like the personal taste.’
