Lecturer of Applied AI Armando Vieira shares his thoughts on Medium, an online publishing platform.
The dominant discourse on AI alignment rests on a quiet but catastrophic assumption: that morality can be imposed on systems that do not experience anything. The debate unfolds as if values were transferable objects — encoded, optimized, constrained — rather than lived structures. Yet across philosophy of mind, ethics, and cognitive science, a recurring warning appears: without experience, there is no morality — only behavior that imitates it.
This essay argues that most alignment proposals fail not because they are insufficiently clever, but because they target the wrong ontological category. They attempt to align zombie systems. And zombies, by definition, cannot be moral.
Read further: https://medium.com/@Lidinwise/zombies-moral-machines-and-the-category-error-at-the-heart-of-ai-safety-c00a1cc10774
