Towards AI-Powered Applications: The Development of a Personalised LLM for HRI and HCI

Area of Science:

• Artificial Intelligence
• Human-Computer Interaction
• Robotics

Background:

• Current large language models (LLMs) face challenges in dynamic user adaptation.
• Existing research often overlooks user-specific contexts and environmental evolution.
• There's a need for adaptable AI in human-robot and human-computer interaction.

Purpose of the Study:

• To propose a novel Personalised Large Language Model (PLLM) agent.
• To enhance the integration and adaptation of LLMs in human-robot interaction (HRI) and human-computer interaction (HCI).
• To address the gap in dynamic user and environmental adaptation for LLMs.

Main Methods:

• Developed a methodology for personalising LLMs using domain-specific data.
• Utilized the NeuroSense EEG dataset for testing and validation.
• Enabled personalised data interpretation for AI adaptability.

Main Results:

• Demonstrated the usability of the PLLM agent in real-world HRI scenarios.
• Showcased applicability across diverse domains like healthcare, education, and assistive technologies.
• Validated the approach for personalized data interpretation and AI adaptability.

Conclusions:

• The proposed PLLM agent represents a significant advancement in AI adaptability and personalization.
• This work contributes to user-centric AI applications and addresses ethical considerations like generalisability and data privacy.
• The findings support the potential of PLLMs to offer substantial benefits across various fields.