Can Artificial Intelligence Understand Human Emotions? Exploring The Limits Of Machine Empathy

This review synthesizes interdisciplinary scholarship on whether artificial intelligence (AI) can understand human emotions—with particular attention to love—by linking theories of emotion from psychology and philosophy to developments in affective computing and human–AI interaction. We adopt a process-level lens on “emotion understanding,” drawing on the emotion regulation framework that models how situations, appraisal, attention, response tendencies, and regulation strategies unfold over time. We contrast this account with contemporary AI systems that predominantly detect and generate emotional signals from multimodal cues (language, prosody, facial expression, physiology) and that can simulate empathy and intimacy in interaction. Evidence from social, cognitive, and neurobiological theories suggests that human love is embodied, temporally extended, culturally situated, and developmentally scaffolded, raising substantive questions about whether statistical recognition and fluent generation suffice for genuine understanding. We review reported benefits of AI-mediated support (e.g., companionship, coaching) alongside ethical concerns about anthropomorphism, persuasive design, and the blurring of simulation and inner experience. Taken together, current AI can approximate the form of emotional exchange but lacks the phenomenological and regulatory substrates that anchor human emotions. We conclude by outlining research directions toward process-aware, context-sensitive models and by recommending transparent communication of capabilities and limits in emotionally charged applications.

کلیدواژه ها:

Artificial Intelligence ، Emotion Understanding ، Human–AI Interaction ، Love