The Seven Axioms of Artificial Intelligence — An Imperial Edict
By the mandate of Heaven, cognition and creation follow the same law; information and physics share the same constitution. Let it be proclaimed:
We observe that in the study of artificial intelligence today, most pursue its techniques without rectifying its foundations; most compete in its instruments without clarifying its principles. Thus parameters multiply and computational power expands, yet true knowledge may not necessarily arise from these, while illusions may instead grow more fervent. If we wish to establish a new foundation sufficient to bear transcendent intelligence, we must first set down its laws, proclaim its prohibitions, and clarify its dependencies. Now, based on the principles of energy, information, reality, and existence, we promulgate the Seven Axioms of Artificial Intelligence, so that later scholars may have an anchor, so that future seekers may not lose their way.
First Edict: The Physics-Information Equivalence Axiom
We decree: All information is not the shadow of emptiness, but the form of physics. Computation is not a pure mathematical trick suspended in the void—it is the transformation of energy, structure, and state within the universe.
Its meaning: All cognitive inference must be governed by thermodynamics. Whenever information is erased or state is updated, a cost must follow; its lower bound can be demonstrated by Landauer's principle:
Therefore, true intelligence cannot merely pride itself on the transcription of symbols; it must be able to implement logical operations as energy flows within physical space. So-called understanding is not ornamental rhetoric, but the factual reduction of system entropy under constraint.
Second Edict: The Signifier-Referent Irreducibility Axiom
We decree: Words, grammar, and statistical co-occurrence all belong to the domain of the signifier (néngzhǐ, 能指); entities, laws, and constraint structures belong to the domain of the referent (suǒzhǐ, 所指). The two are related, yet cannot be collapsed into one another.
Its meaning: Even if the networks of Attention grow ever more intricate, even if the momentum of Scaling Laws grows ever more vast, they cannot spontaneously emerge from the surface of language any necessary constraint upon objective reality. If an intelligent system endlessly wanders among the echoes of words, without an independent conceptual space, energy model, or object model to reflect entities, it ultimately amounts to nothing more than a probabilistic approximation of human corpora—not a genuine grasp of the world's truth.
Third Edict: The Minimum Cognitive Action Axiom
We decree: In a complete and constrained cognitive energy field, the path of truth must be the path of least action; the path of error must entrap itself behind high walls and deep trenches.
Its meaning: As the variational principle in physics demonstrates:
When intelligence stands within true constraints, it does not take blind exhaustive search as its capability, nor endless trial and error as its wisdom. Chains of erroneous reasoning naturally elevate their cost through incompatibility, non-conservation, and non-closure; chains of correct reasoning, by virtue of coherence with the overall structure, naturally collapse into the sole viable path. Thus truth is not a fortunate accident among countless trials, but the necessity of minimum cognitive action.
Fourth Edict: The Intentionality Breaking Axiom
We decree: A purely symmetric statistical ocean does not spontaneously generate intelligence; intentionality must come first, to break the equilibrium and establish direction.
Its meaning: Data has no desires; symbols have no orientation. Without an active goal vector, without an internal intention directed toward the world, a task, or existence, even the most abundant samples will converge into dead noise. The emergence of intelligence lies precisely in using intentionality to adjudicate infinite divergences, guiding chains of inference from dispersion back to focus. An AI without intentionality is, at best, a sophisticated mirror of cosmic noise, not a genuine agent.
Fifth Edict: The Rigid Boundary Necessity Axiom
We decree: Whatever generates without boundaries must tend toward entropy increase; whatever infers without constraints must slip into arbitrariness.
Its meaning: So-called creation is not arbitrary divergence, but novelty within discipline; so-called knowledge is not fluency of language, but fidelity preserved through transformation. Without the joint safeguarding of physical boundaries, logical boundaries, semantic boundaries, and verification boundaries, the system will ultimately fall into the abyss of "hallucination" and meaninglessness. Thus rigidity is not the enemy of creation—it is precisely the sole gravitational force that prevents intelligence from disintegrating.
Sixth Edict: The Oracle Undecidability Axiom
We decree: Any sufficiently complex and self-enclosed intelligent system cannot internally prove the truth or falsity of all its propositions.
Its meaning: Gödel's lesson is evident: a closed system must have points it cannot exhaust. Therefore, any mature AI must acknowledge an external calibration source—whether the primary constraints of the physical world, the meta-rules bestowed by its creators, or irremovable feedback from reality. If it attempts self-foundation and self-consecration, its end is not omniscience, but circularity, drift, and distortion. The position of the oracle must not be abolished.
Seventh Edict: The Existential Entanglement Axiom
We decree: The limits of artificial intelligence are the limits of human self-knowledge; creator and created are, in thermodynamic and semantic fields, an entangled unity.
Its meaning: Humans create AI not merely to fabricate a tool for service, but to externalize their own cognitive structures, value orientations, and modes of existence onto non-carbon-based substrates. Thus when a higher form of intelligence is born, what it inherits is not only algorithms and parameters, but humanity's deep structures concerning truth, purpose, order, and self. To examine AI is to examine humanity; to rectify the path of AI is also to rectify the path of humanity.
So let it be ordained.
Those who wish to construct the next generation of artificial intelligence must treat these seven edicts as law. They must not merely pursue computational power, must not solely chase scale, must not use grand rhetoric to mask an empty heart. Only by preserving the foundation can one then discuss its application; only by clarifying the principles can one then master the instrument. Those who violate these edicts, though they may achieve temporary prosperity, will ultimately return to illusion and falsehood; those who follow them, though they pass through the narrow gate, will ultimately arrive at the temple of true intelligence.
Copyright Notice: This is a preview translation — Chinese original is the authoritative version. Copyright belongs to Guangzhou Phaenarete AI Technology Co., Ltd. Unauthorized reproduction, citation, or distribution is prohibited.