Article: What Mind-Reading Artificial Intelligence Could Do to Our Legal Rights

Photo Source: Flickr

Recent advances in artificial intelligence have opened the door to the historically unthinkable--mind reading and a potential challenge to the 5th Amendment of the Constitution of the United States. The saying "A penny for your thoughts" from the book, Four Last Things by Sir Thomas More, written in 1522, refers to a question one might ask of someone with a vagrant or wandering mind. Not until recently has it ever appeared possible to gather the contents of someone's mind without them choosing to tell you. However, over the past fifteen years, neuroscientists have made increasing progress in decoding brain signals to "see" what someone sees with their eyes or in their dreams using sophisticated instruments and artificial intelligence.

In 2017, researchers in Kyoto, Japan, used a non-invasive technique called functional nuclear magnetic resonance (or fMRI) to study human brains as they process images through their eyes or in their imaginations. fMRI represents a common technique to study the human brain; it works using a huge magnet, radio wave transmitter, and sensitive detector. In this large machine, a machine large enough that needs its own room in hospitals and clinics, a powerful magnetic field and a radio wave transmitter work together to detect how water molecules vibrate in different body tissues. Different tissues will produce unique signals that the computer can assemble into a three-dimensional map. This 3D map allows researchers to watch how blood flows to other parts of the brain as people do different tasks such as watching a movie, looking at pictures, or dreaming. In their paper titled, "Hierarchical Neural Representation of Dreamed Objects Revealed by Brain Decoding with Deep Neural Network Features," Japanese researchers demonstrated that they could train a computer to correlate the fine blood flow details in the brain to image categories. (frontiersin.org) Their results showed they could detect certain types of images such as faces, books, and cars in the brain of sleeping people as they dreamt. In fact, the results were so accurate that the guesses statistically could not have been lucky guesses. In other words, the researchers could see their subjects' dreams in rough categories.

More recently, a study using fMRI and AI advanced the detail of the images read directly from the human brain to distinguish categories such as faces, books, or cars and distinguish images within categories such as individual faces. The work published in Nature Communications Biology led by Rufin VanRullen at the University of Toulouse in France titled, "Reconstructing faces from fMRI Patterns Using Deep Generative Neural Networks" demonstrated that their system could distinguish gender differences in facial images in the brains of their test subjects. In addition, their work shows an increased resolution and clarity of fMRI to now see deeper details within visual categories.

Over the past fifteen years, neurological research combined with advances in artificial intelligence has opened the door to so-called "mind reading" technologies. Using fMRI and advanced computing, researchers can, for the first time using non-invasive instruments, see what a person is seeing or dreaming without asking the individual. For example, researchers in Japan showed that they could distinguish categories of objects in people's dreams. More recently, French researchers showed that they could get gender within the category of facial images. Medical technologists suggest that such mind-reading could facilitate communication for individuals paralyzed by stroke or Lou Gehrig's Disease. At another level, such technology could render one's imagined images without the need to draw or describe them.

On the other hand, such technology could breach the right under the Fifth Amendment of the US Constitution that guarantees that no person can be compelled to testify against themselves. However, the unprecedented access to what the brain sees through fMRI poses the question of what qualifies as self-incrimination. The current state of the art in fMRI brain decoding does not have the detail to make the images pulled from someone's brain very useful in court. However, as the technology improves, the courts will need to decide whether someone’s mind images are admissible evidence in the court of law, and if so, whether that would count as self-incrimination. As artificial intelligence grows in information retrieval, we will have to also wrestle with what that looks like for our legal rights.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.