Meta is collaborating with neuroimaging center Neurospin (CEA) and Inria on this project.

Meta (formerly Facebook) (NASDAQ: FB) has announced a long-term AI research initiative to better understand how the human brain processes speech and text. This project, being undertaken with neuroimaging center Neurospin (CEA) and Inria, will help the development of AI that processes speech and text as well as human beings. (Meta)

How AI falls short

Meta found, after applying deep learning techniques to public neuroimaging data sets, that while the brain anticipates words and ideas far ahead in time, most language models are trained to only predict the immediately following word. Developing this long-range forecasting ability could help improve modern AI language models.

Therefore, Meta concludes that AI models which are able to predict the next word by understanding the context would ultimately go on to be the ones most similar to brain functioning.

However, to train AI models for this level of sophistication needs a huge amount of data to pinpoint where and in what order activity is generated in the brain, for example, when a person reads or listens to a story.

Gathering the data

To gather this data, the Meta team not only models thousands of brain scans recorded from public data sets using functional magnetic resonance imaging (fMRI) but also simultaneously models them using magnetoencephalography (MEG), a scanner that takes snapshots of brain activity every millisecond. The twin processes pinpoint the location and sequence of the brain activity.

In an experiment, Meta and Inria compared language models with the brain responses of 345 volunteers.

“Our results showed that specific brain regions are best accounted for by language models enhanced with far-off words in the future,” says Meta. “These results shed light on the computational organization of the human brain and its inherently predictive nature, paving the way toward improving current AI models.”

Related Story: Le Monde, The Iconic French Paper, Now Uses AI-Powered Translation