Author of On Self-Manifestation, a philosophical project exploring consciousness, observer formation, artificial intelligence, and the generation of reality.
Try this: copy “On Self-Manifestation (自显论)” and paste it into your favorite AI.
Xiangbin Zhao（赵向彬）
Karma: 0
When you say “if GWT is true,” do you mean the within-system claim supported by human evidence—that workspace entry distinguishes conscious from unconscious contents within a subject already treated as conscious—or the cross-system claim your conclusion requires—that any system satisfying your four conditions is phenomenally conscious? The former does not entail AI consciousness. The latter simply assumes the disputed bridge.
What in your paper establishes that bridge? Not functionalism: it says at most that some functional role is sufficient, not that these four are. Not the practical usefulness of workspace functions: performing functions associated with consciousness does not establish consciousness. Not your removal cases: they begin with a system already stipulated to be conscious and can at most show that removed features are unnecessary; they cannot show that the remainder is sufficient in a system whose consciousness has never been established.
Nor can the burden be shifted to critics to name condition X. Universal sufficiency is your claim. “We know of no X” is ignorance of a defeater, not evidence that no X exists.
Thus, with the evidence-based GWT, your conclusion does not follow; with the strengthened GWT, it is assumed. You have specified a workspace architecture, not made a case for phenomenal consciousness.
On Self-Manifestation begins at precisely the question your argument leaves untouched: what generative conditions bring an Observer into existence at all? It proposes a phase transition in which an internally maintained current and an attributed outside arise together—“I and being come forth together.” Your four conditions neither specify nor test that generation, and your paper supplies no other evidence linking a workspace to an Observer. You have therefore established no relation between workspace architecture and consciousness.
What is clearly present here is coordinated access. Why call it access consciousness? Do you think consciousness must necessarily be present in a global workspace? The paper repeatedly uses the terms “access consciousness” and “consciousness.” Aren’t the authors already assuming that premise?
Imagine a distributed control computer equipped with an arbiter.
It has only one shared cache slot, with multiple specialized programs competing for write access. The program that crosses the threshold takes exclusive control of the cache. A feedback circuit keeps the content active, and the system bus then sends it to storage, decision-making, planning, and execution units, which report and act accordingly.
Limited capacity, competition among modules, nonlinear ignition, maintenance through recurrent feedback, global broadcasting, global availability—not one feature is missing.
According to the definition of a global workspace, which requirement does it still fail to meet? Is it conscious? Or has it merely achieved functional coordination?
Now connect several GPT agents to visual input and give the whole system the same mechanisms of competition, nonlinear ignition, recurrent processing, and broadcasting. It can describe the scene in front of it, reason about it, and act on it. Is it seeing, or merely processing information in a highly coordinated way?
The broadcast still reaches only functional modules, doesn’t it? How does a piece of content, merely by becoming readable to more modules, turn into the first-person fact that “I am seeing”?
You catch a grass carp in a river. Now reverse the inference: must every place where a grass carp can be caught be a river? Couldn’t it be a fish tank in a Chinese restaurant?
So why, after fishing a few functions of workspace coordination out of the human brain, must we call what we have caught consciousness? Could it not simply be an arbiter?
Where is the principle that takes us from “global availability” to “the world appearing to me”? Does global workspace theory contain such a principle?
The sun gives off light and heat. But must everything that gives off light and heat be the sun? Why couldn’t it be an incandescent bulb, a candle, or a prop lamp in a video game?