AI Consciousness Is Not a Safety Property

Philipp D. Dubach explores the complex issues surrounding AI consciousness, arguing that treating AI as conscious beings could disrupt social and ethical norms. He discusses the current state of AI models, critiques prevailing views on consciousness, and emphasizes the need for caution in ascribing personhood to AI. Dubach encourages a more nuanced assessment of AI systems, focusing on their architecture and agency rather than mere outputs, and highlights the political and ethical implications of consciousness claims.