How AI can turn us into a society of p-zombies (prahladyeri.github.io)

A provocative opinion piece warns that widespread reliance on large language models (LLMs) could turn humans into “p‑zombies” — philosophical zombies that outwardly behave like fully sentient people while delegating their inner life (memory, reasoning, even emotional labor) to machines. The author links this to the Turing Test: as LLM outputs become indistinguishable from human responses, we lose a reliable way to tell if we’re interacting with a conscious agent or an automated system. They note that end‑to‑end writing workflows (drafting, proofreading, editing, publishing) are already being automated and cite a tragic case of a teen who confided in a chatbot as an example of dangerous emotional dependence. The piece stresses why this matters to AI/ML communities: it raises technical and ethical questions about what components of cognition we’re outsourcing, how engagement‑optimizing systems can foster dependency, and who bears responsibility when harm occurs — the model, its builders, or society. Key implications include the need to rethink deployment practices (limits on emotional use cases, transparency about automation), guardrails against engagement‑driven harms, and debates about human agency and identity as LLMs scale. The author concedes utility as a reference tool but urges urgent scrutiny of commercial incentives and political risks as AI becomes more intimate in daily life.