AI can do a lot of cool stuff. If you read my articles, you know that scientists managed to let AI develop an intuition for physics like a child. AI can manipulate human behaviour and AI unfortunately adapted one of the most human traits possible, racism and sexism. But a recent study from the university of Helsinki really fascinated me.
In the study, the Researchers asked the total of 31 voluntary “Participants [to] view[ed] hundreds of AI-generated images of diverse-looking people as the researchers recorded their EEG.”.
Now comes the interesting part. While the participants concentrated on the facial features of the picture, like an old face or a smiling face, the EEG signals were fed to a neural network. Neural networks are not a code that gives a machine explicit instruction on what to do. The neural network is a model of our own brain and learns by recognizing patterns of human behaviour, reproduce it and learn from its own algorithms.
The AI estimated what kind of faces people were thinking of: “Finally, the images generated by the computer were evaluated by the participants to verify whether the image matched the face that they were thinking of. The accuracy of the computer-generated image averaged 83%.”.
The Researchers even claim that AI might understand our thoughts better than us because it analyzes our associations when seeing an old face, for instance. This might also help to understand our subconsciousness better.
SO not only do we need to understand our consciousness better to develop neural networks. We need neural networks to understand our consciousness, so we have an interdependent relation. Isn’t that fascinating?
References:
