A Motivational Model of BCI-Controlled Heuristic Search

Area of Science:

• Neuroscience
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Autonomous AI systems require human supervision for safety and ethical alignment.
• Brain-Computer Interfaces (BCI) offer a potential pathway for direct human influence on AI.

Purpose of the Study:

• To propose and validate a framework for human supervision of AI using BCI.
• To integrate cognitive neuroscience findings with AI algorithms for enhanced human-AI interaction.

Main Methods:

• Developing a framework utilizing BCI to influence AI heuristic search algorithms.
• Employing prefrontal asymmetry, measured via functional near-infrared spectroscopy (fNIRS), as a motivational signal.
• Modifying heuristic functions in AI based on BCI-derived user disposition.

Main Results:

• Demonstrated a proof-of-concept experiment controlling AI computation progression using fNIRS-based prefrontal asymmetry.
• Successfully mapped user disposition signals to AI heuristic search parameters.
• Showcased the feasibility of a motivational BCI approach for AI supervision.

Conclusions:

• The proposed framework enables effective human supervision of autonomous AI systems through BCI.
• This approach enhances AI control and alignment by incorporating human motivational states.
• Future research can expand on this BCI-driven human-AI collaboration model.