This study reveals which text features make humans attribute consciousness to LLM-based AI systems, with significant implications for AI development and deployment.

Key findings:

• Metacognitive self-reflection and subjective expressiveness most strongly influence perceptions of AI consciousness
• Human responses clustered into groups with different attribution patterns of consciousness to AI
• Features like logical reasoning and knowledge had less impact than expected

Medical relevance: Understanding how humans perceive AI consciousness has important implications for medical applications, including patient interactions with AI systems, clinician trust in AI tools, and ethical deployment of AI in healthcare settings where perceived consciousness might affect treatment decisions.

Identifying Features that Shape Perceived Consciousness in Large Language Model-based AI: A Quantitative Study of Human Responses