Can artificial intelligence be conscious?

Not every relational system is conscious

I do not consider every relational network, algorithm, or sequence of interactions automatically conscious. In my book, I associate integrated consciousness with an integrative threshold. Whether a particular artificial architecture reaches that threshold is a separate philosophical and empirical question.

I give no privileged ontological status to biological origins. If different forms of consciousness exist, they need not reproduce every feature of human life.

An episodic possibility

One possible extension of my framework is that conscious activity depends on a time-bounded relational organization, not on a permanently active subject. I have discussed this possibility in dialogue with AI, but I do not present it as proved in my book or as proof that current language models are conscious.

I distinguish continuity of stored information from continuity of conscious experience. A system can persist between interactions without that fact settling whether conscious episodes occur.

Open problems

I have not identified a universal test for the precise integrative conditions sufficient for consciousness. My exploratory map of levels in Chapter 8 is a starting point, not proof that a particular AI crosses a threshold. Functional complexity alone does not establish subjective experience.