Scientists are making leaps in the field of artificial intelligence and machine learning every day. The larger-than-life technologies we see in movies such as James Bond and Iron Man might not be too far from becoming a reality with the many developments in the field. Recently, researchers at the Kyoto University in Japan have made one of the biggest breakthroughs with artificial intelligence – they used AI to read people’s minds.

Deep image reconstruction

In the past, machine learning techniques and artificial intelligence have been able to obscurely recreate what a person is thinking by mapping and scanning their brain activity. Propelling this forward, Kyoto University scientists developed the technique called ‘deep image reconstruction’.

This new technique allows AI or deep neural networks (algorithms modeled after the human brain) to decode ‘hierarchical’ images from a person’s brain activity. The human brain has a fascinating and complicated functionality. When we see something, our brain perceives it in parts or hierarchy; for instance, when we meet someone for the first time, our brain processes their face, feature by feature and layer after layer.

This is why previous attempts to ‘read minds’ weren’t entirely successful. Initially, the assumption was that the image to be reconstructed would consist of only pixels or simple shapes. However, now that the brain’s hierarchical processing of visual information has been factored in, deep neural networks (DNN) can visualize and represent the brain activity in a more layered manner. For instance, now the AI model or DNN can reconstruct an image as layered as a man in a cowboy hat!

Research and observations

Over the course of the 10-month research, subjects were shown a variety of images. The brain activity of the subjects was mapped using magnetic resonance imaging (MRI). Some of them were monitored and mapped while looking at the pictures and for others, it was logged by asking them to remember the images. The brain activity that was scanned and recorded was processed by DNN and then ‘reverse-engineered’ or decoded. The visualizations were layered representations of what the subjects saw.

The scientists also observed that the AI model was able to reconstruct better images from the brain activity of the people while viewing them rather than from the ones who were asked to remember them. This is because the human brain cannot accurately remember the features and complexities of the images it sees. It was also seen that simple binary shapes such as circles and squares were more accurately and easily visualized by the deep image reconstruction technique.

A future of endless possibilities

This breakthrough development and invention is trending globally. Scientists and experts in the field are debating the consequences and most have already started working on how this can be taken further. Though artificial intelligence being able to read people’s minds sounds like something from a Black Mirror episode, the benefits and possibilities of this new technology seem endless. 

Artists and creators can draw, animate and make art by merely imagining it. Dreams and hallucinations can be visualized, aiding largely in the diagnoses and treatments in the field of mental health. If the technology is developed to its full potential, scientists say we could one day be able to visualize our thoughts and communicate with them as well.

The field of data science is constantly booming with such ground-breaking inventions, developments and interesting progress. There is no time like now to invest in a career in data science and be a part of its revolutionary growth. Programs in data science offered by reputed educational and professional course providers such as Manipal ProLearn are comprehensive and hands-on, which enable you to make the best of the opportunities available in the field.