Human-Centered, Non-Invasive Brain-to-AI Interfacing (B2AI): A Proof-of-Concept Feasibility Case Study in Augmented Cognition

Source abstract

Brain-to-AI Interfacing (B2AI) offers a novel human-centered neurotechnology framework for augmented cognition. In this proof-of-concept feasibility case study, we demonstrate a working non-invasive B2AI prototype supporting bidirectional communication between a user and a large language model (LLM) through EEG-based input and TMS-based feedback. In a virtual navigation task, we found that use of the B2AI system was associated with higher path efficiency in a single-user case study. Additionally, we found that the average time per step was higher when the system was active, likely reflecting the time required to initiate the EEG command, query the AI module, receive TMS feedback, and execute the instructed movement. Together, these findings suggest that the B2AI system was technically feasible and may support more efficient task performance under constrained experimental conditions. Future work and fully powered experiments will be needed to determine the generalizability, usability, safety, and application value of the system, but this case study suggests that current technology supports the use of B2AI as an experimental tool for augmented cognition research.