PolyU researches find AI are easier to train the same way as humans process languange

The study also gave fresh insights on how brains process full discourse.

Researchers from the Hong Kong Polytechnic University (PolyU) found that large language models used for GenAI can be easier to train using the same way as humans process language.

The team led by Li Ping, dean of the Faculty of Humanities and Sin Wai Kin Foundation Professor in Humanities and Technology at PolyU, used next sentence prediction (NSP) task and studied the correlation between the model's data and brain activation.

According to PolyU, NSP simulates one central process of discourse-level comprehension in the human brain to evaluate if a pair of sentences is coherent.

The researchers trained two models, one with NSP and word prediction, and the other only with word prediction. 
 
Functional magnetic resonance imaging (fMRI) data were also collected from people reading connected or disconnected sentences. 

The experts examined how closely the patterns from each model matched up with the brain patterns from the fMRI data.

“The model with NSP matched human brain activity in multiple areas much better than the model trained only on word prediction. Its mechanism also nicely maps onto established neural models of human discourse comprehension,” PolyU said. 

There were also new insights found about how brains process conversations. Specifically, the study showed that parts of the right side of the brain, not just the left, helped understand longer discourse.

Furthermore, the model trained with NSP could better predict how fast someone reads. This shows that simulating discourse comprehension through NSP helped AI understand humans better, PolyU said.

ALSO READ: PolyU's fungus-based fat replacer targets obesity, elevates gut health

“Our findings suggest that diverse learning tasks such as NSP can improve LLMs to be more human-like and potentially closer to human intelligence,” Li said. 

"More importantly, the findings show how neurocognitive researchers can leverage LLMs to study higher-level language mechanisms of our brain. They also promote interaction and collaboration between researchers in the fields of AI and neurocognition, which will lead to future studies on AI-informed brain studies as well as brain-inspired AI,” he added.