Artificial Intelligence and Brain Science: Background information when reading The Last Murder at the End of the World

Artificial Intelligence and Brain Science

This article relates to The Last Murder at the End of the World

In The Last Murder at the End of the World, a small group of people have survived the deadly fog that destroyed mankind. These survivors have managed to create a peaceful, productive society on their small island, benefiting from the sense of community bestowed by Abi. Abi is a mysterious intelligence that is part of the minds of all the islanders; she can not only read everyone's thoughts and communicate with them, but also exert control over the islanders' minds and bodies as needed. Although mind control in this sense doesn't exist in the real world, recent advances in artificial intelligence (AI) have shown how this technology can be used to help scientists better understand the brain.

The process through which the brain takes input from the eyes to create a mental image is not fully understood by current science. However, a recent study from scientists at Japan's Osaka University used a new AI algorithm (a set of instructions used in calculations and operations) to read brain scans and recreate an image of what the scanned individual was seeing. Four study participants viewed 10,000 images of common items such as clocks, teddy bears and airplanes while their brains were being scanned using functional magnetic resonance imaging (fMRI). These scans mapped the blood flow to active regions of the brain and were then used to train the algorithm—named Stable Diffusion—to recognize certain scans as related to specific images.

Once trained, the algorithm was tested using additional scans, and although Stable Diffusion could initially recreate the layout and perspective of the image based on the scans, it couldn't recreate objects and instead placed abstract figures into the image. This changed, however, once photo captions were included during the training phase. The text and visual information together made it easier for the algorithm to decipher the brain scans in the testing phase and create the image more accurately. Although Stable Diffusion currently only works with the scans from which it was trained, there is hope that future iterations could be used on unscanned brains, with conceivable applications including recording human dreams, studying how animals understand the world and even communicating with people with paralysis or in comas.

Other AI-based technology has been used as a form of non-invasive mind reading. Researchers at the University of Texas used fMRI data to translate brain activity from participants into a continuous stream of text. Three individuals listened to podcasts for sixteen hours while undergoing fMRI scanning, and the scans were used to train an AI-based decoder built using a large-language model (an AI construct that looks at wording in a high number of examples of language and then predicts the next word or phrase). The individuals then listened to a story or imagined their own story, and the decoder was tested using the corresponding scans. The decoder was able to reconstruct the thoughts with a high degree of accuracy, although due to an inherent time lag between the thought and the duration of the measured blood flow, the outcome only gave the gist of the original material rather than exact words or phrases. The custom nature of the models also limited the decoder's utility in the short term, but researchers nonetheless hope that additional research could allow this technology to be used for speech restoration in unscanned individuals who have suffered a stroke or are suffering from neuron diseases such as Alzheimer's.

Other AI-based brain technologies are less academically based. Launched in 2016, Elon Musk's Neuralink is an AI-enabled chip that, once implanted in the brain, could record and potentially stimulate brain activity. Musk hopes to use this technology to allow humans to "join" supercomputers and allow users to play games or send messages with only their thoughts, but he also envisions his chips could allow individuals who have lost the use of their limbs to control their phone or computers just by thinking. Additionally, scientists hypothesize that if Neuralink could stimulate brain activity, it could increase mobility in paralyzed individuals or help blind patients regain their sight. Neuralink was implanted in the first human subject in January 2024, so additional research is needed to determine whether it has broad medical uses or if it is more of a science fiction novelty.

Although Abi in The Last Murder at the End of the World is a far more advanced form of AI than anything that currently exists, AI-based technology is a quickly growing field. Despite many ethical concerns about its broader use, AI's potential to restore communication in individuals affected by disease or injury could well improve both individual lives and society as a whole.

Artificial intelligence re-creations of images based on brain scans (bottom row) match the layout, perspective, and contents of the actual photos seen by study participants (top row), courtesy of Science.

Filed under Medicine, Science and Tech

This article relates to The Last Murder at the End of the World. It first ran in the June 5, 2024 issue of BookBrowse Recommends.