Schwitzgebel: AI consciousness will be undecidable by current theory

Eric Schwitzgebel's new arxiv overview says something uncomfortable for both AI boosters and AI skeptics: we are about to build systems that count as conscious under some influential theories of mind and not conscious under others, with no principled way to adjudicate between the theories.

The paper is explicit that the question will not resolve on current evidence. "We will not be in a position to know," Schwitzgebel writes, "whether we are surrounded by AI systems as richly and meaningfully conscious as human beings or instead only by systems as experientially blank as toasters." The overview walks through the standard machinery — the Turing Test, the Chinese Room, Global Workspace and Higher-Order theories, Integrated Information theory, the question of whether a biological substrate matters — and lands on a flat verdict: "None of the standard arguments either for or against AI consciousness takes us far."

Two of the researchers we follow in our Who's Who circulated the paper, which is a small signal that the philosophy-of-mind angle is back on the agenda for people who otherwise spend their time on capabilities.