Embedding Emotional Intelligence in AGI for Enhanced Cognition
Year: 2025
Type: article
Abstract: Embedding emotional intelligence into Artificial General Intelligence (AGI) enhances its cognitive versatility beyond mere emotion recognition. This study develops a multimodal AGI prototype that integrates facial expression data (CK+ dataset) and speech emotion cues (RAVDESS dataset) to inform reasoning and adaptability across diverse tasks. A transformer-based architecture fuses visual and auditory embeddings, feeding them into a reasoning module enhanced by a fine-tuned large language model (LLM) to generate emotionally tailored responses. The AGI assists users in problem-solving or independently resolves scenarios, adapting to emotional contexts. Evaluation targets >85% emotion recognition accuracy and 20% higher task success rates—measured as user or AGI performance improvements—over emotion-agnostic baselines, plus user satisfaction (>4/5) from 50 evaluators. Early simulations suggest emotional integration boosts contextual reasoning by up to 25%. This work pioneers an affective AGI framework, leveraging emotional cues to enhance cognition, with applications spanning education, healthcare, conflict resolution, and ethical AI, and insights into emotion’s role in intelligence.
Authors: Vishal Narender Punjabi, Balajee Asish Brahmandam, Srinath Chandramohan, Independent Researcher, USA
Language: English
Topic: Emotion and Mood Recognition
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W4409979119