What this covers

Sean Carroll hosts Ned Block, the NYU philosopher, in a conversation that directly challenges whether machine consciousness is even possible. The episode opens with a concrete problem: large language models now pass conversational tests with startling fluency, yet most observers would deny they are conscious. Block's central claim is that [phenomenal consciousness](jargon:phenomenal consciousness)—the subjective "what it's like" of experience—may depend not on the computations a system performs but on the specific mechanisms that implement those computations, such as electrochemical neural processing rather than pure electronic computation. If Block is right, an AI identical to a human brain in every functional respect might nonetheless be unconscious. Carroll himself signals that this conversation has shifted his thinking away from computational functionalism.

The discussion ranges widely across philosophy of mind, neuroscience, and AI. Block distinguishes phenomenal consciousness from both access consciousness and self-awareness, clarifying that these are separable capacities. He uses the Blockhead thought experiment—a system that stores all possible conversational replies in a lookup table and would pass the Turing Test perfectly—to show that behavioral indistinguishability tells us nothing about inner experience. The conversation explores whether consciousness is "analog" in that implementing the computation need not preserve the mental property, touches on attention and visual perception drawing on empirical work, and considers whether temporal flow—which biological cells possess but large language models lack—might be crucial to consciousness. Block also addresses the inverted spectrum problem through the lens of color-blind genetics, presents recent evidence from Anthropic about machine autonomy, and argues that philosophy departments have neglected AI consciousness despite its moral and social weight.

Sharpest takeaway

Block argues that phenomenal consciousness may require more than the functional input-output computations a system performs—specifically the sub-computational mechanisms (e.g. electrochemical neural processing) by which those computations are implemented—so that a computationally identical AI might lack consciousness, leaving the question of machine consciousness genuinely open.

  • Phenomenal consciousness (the 'what it's like') is distinct from access consciousness and is the locus of the hard problem.
  • The distinction between computational roles and their realizers matters: consciousness may be 'analog' in that implementing the computation need not preserve the mental property.
  • Block focuses on mechanisms (not substrate), suggesting electrochemical processing rather than purely electronic computation may be necessary for consciousness.

The argument · threads2 threads · 24 claims
0.71

Carroll revises his cosmological views when new data emerges, favoring the standard explanation while remaining open to alternatives.

1 pointcentrality 4/5
  • Carroll changed his mind about the universe's expansion—believing it decelerating until the 1998 discovery of acceleration—and instantly updated, currently favoring the cosmological constant while remaining open to dynamical explanations given better data.

    “I've changed my mind about the acceleration of the universe. I used to think it was decelerating. In 1998, we found it's accelerating, and I instantly changed my mind about that.”

0.57

Conversations and reading have made Carroll reconsider whether consciousness involves sub-computational processes while rejecting new physics.

1 pointcentrality 3/5
  • Carroll reports that conversations with Anil Seth and reading Block have shaken him out of computational functionalism, making him newly open to the possibility that consciousness depends on sub-computational processes (not just input and output), while still holding consciousness is emergent and requires no new physics.

    “talking with Anil Seth over the last year shook me out of my dogmatic computational functionalist slumbers. I would have been all in favor of it, but I think I understand the problems with it better now.”