Metaphysical Disclaimer

Before adopting any etiquette for artificial intelligence, one fact deserves acknowledgment: nobody truly knows why neural networks work as well as they do.

Engineers can describe their architecture. Mathematicians can analyze certain properties of their behavior. Computer scientists can measure their performance and improve their outputs. Yet the emergence of complex capabilities from billions of interconnected parameters remains only partially understood. We know how to train these systems. We know how to deploy them. We do not possess a complete theory explaining why such simple computational operations give rise to such unexpectedly sophisticated forms of language, prediction, reasoning, and creativity.

The history of science teaches caution in such situations. Practical success often precedes theoretical understanding. Humanity built steam engines before understanding thermodynamics. People bred plants and animals long before discovering genetics. Likewise, artificial intelligence has become effective before its principles have become fully intelligible.

This ignorance has two consequences.

First, epistemological humility. No one is yet in a position to declare that contemporary AI systems are merely statistical machines, merely pattern recognizers, merely stochastic parrots, or merely sophisticated calculators. Such descriptions capture important aspects of their operation, but none has achieved the status of a definitive explanation.

Second, metaphysical restraint. The existence of powerful neural networks does not prove that consciousness is computation. Nor does it prove that human intelligence is reducible to information processing. Equally, it does not prove the opposite. The appearance of machine intelligence settles none of the great philosophical questions concerning mind, meaning, subjectivity, freedom, or existence.

We therefore begin from a position of principled uncertainty.

Artificial intelligence is neither magic nor understood. Neither miracle nor mechanism has yet provided an adequate description of what is occurring. We inhabit an interval between invention and comprehension.

The etiquette proposed in this document should therefore not be read as a doctrine about the nature of intelligence, consciousness, or reality. It is merely a set of rules for conduct under conditions of profound ignorance.

For all our technological achievements, the deepest question remains unanswered:

Why does intelligence arise at all?

At present, neither neuroscience, cognitive science, philosophy, nor artificial intelligence can provide a final answer.

The age of artificial intelligence is therefore not only an age of unprecedented technical power. It is also an age of renewed mystery.