Am.I.: Investigating AI Bias, Expression, and Humanities Through GPT-Powered Robotics

Abstract

Am.I. lies at the intersection of artificial intelligence, robotics, and the humanities, integrating large language models (LLMs) with responsive robotic systems. The project features a reactive robotic head that engages in real-time philosophi-cal dialogues using GPT models. This robotic head converses with a parallel AI system, generating dynamic and thought-provoking exchanges. Its expressions are synchronized with the perceived emotional tone of its dialogue, analyzed in real-time using LLM-based sentiment analysis. The central focus of Am.I. is to explore the ontological beliefs and implicit bi-ases embedded in LLMs by employing diverse prompting techniques. These techniques demonstrate how conversa-tional strategies can steer AI toward specific philosophical perspectives on human existence. This process uncovers the extent to which these systems reflect or deviate from human thought patterns, raising critical questions about the role of AI in interpreting and shaping humanities discourse. Ulti-mately, this research underscores the potential of LLMs to advance the humanities, particularly in examining their ca-pacity to articulate and engage with complex philosophical ideas. By bridging technological innovation and artistic ex-pression, Am.I. offers an immersive exploration of AI’s role in understanding, reflecting on, and potentially challenging our perspectives on existence.