If digital computers are conscious, they are conscious at the hardware level
If digital computers can be conscious, some argue, that consciousness must arise from the physical hardware itself — a view that nudges toward panpsychism, the idea that some form of experience pervades all matter. Commenters debate whether current AI systems like LLMs could ever be conscious given their architecture, lack of memory and embodiment, and whether consciousness is even definable or testable in a way that distinguishes humans, animals, and machines. The thread also raises ethical stakes: how we answer these questions affects whether we treat advanced AI as mere tools, potential moral patients, or something in between.
Overall reaction to the article & panpsychism
- Many find the piece unusually careful for this topic and intend to reread; others see it as “typical philosophy” and likely unhelpful.
- The author’s panpsychist leanings trigger both interest and skepticism. Some see panpsychism as the only way to make sense of consciousness; others say it’s unfalsifiable, folk-psychological, or ideologically motivated.
- Several note the article assumes qualia are physically realized, which some participants dispute or treat as rhetorically suspect.
What is consciousness? Competing intuitions
- No shared definition emerges. Proposed views:
- Consciousness as basic “experiencing” not tied to a specific process.
- Consciousness as awareness of qualia vs. awareness of external reality.
- Illusionist takes that “consciousness” is a misleading construct emerging from reasoning systems.
- Disagreement over whether qualia are real, well-defined, or even needed.
- Some stress that without a rigorous definition, arguing about machine consciousness is premature.
LLMs, AI, and degrees of consciousness
- One camp: current LLMs are absolutely not conscious—no intrinsic motivation, memory, homeostasis, persistent state, or stake in outcomes; they are probabilistic token predictors and closer to thermostats than minds.
- Another camp: we don’t understand human consciousness either, so asserting LLMs have “no form” of it is unwarranted. Future architectures (or scaled-up systems with memory, embodiment, and continuous operation) might plausibly be conscious.
- Debate over whether “reasoning” in LLMs is genuine or just marketing; some argue behavior is the only accessible test.
- Several note that human uniqueness is overprotected by shifting goalposts; terms like “technoconscious” are suggested to sidestep semantics.
Hardware, substrate, and simulation
- Some argue if consciousness is computational, substrate shouldn’t matter; others think digital error-correction and discretization may suppress precisely the holistic physical effects that matter for phenomenal experience.
- Discussion of whether full universe simulations are even possible, analogies to quines and self-expanding archives, and whether computation alone can ever capture consciousness.
Ethics, moral status, and “tiling the universe”
- Concerns raised about a utilitarian vision of filling the cosmos with happy simulated minds.
- Others reply the immediate ethical priority is to avoid creating vast amounts of suffering, not maximization of “good experiences.”
- “Moral patienthood” for LLMs is viewed by some as misplaced, especially given their lack of free will or clear inner life.
Embodiment, memory, and internal state
- Several propose that consciousness requires embodiment, rich sensors, and continuous internal state, not just text I/O.
- Experiments are described where LLMs are given dream-like processes and machine-load-dependent “emotions,” but a persistent, self-referential internal state is still seen as missing.
- Debate over whether externalized or engineered memory in LLMs is fundamentally different from human memory rewiring in the brain remains unresolved.