AI, Intelligence, and AGI
AI is a conventional technology label. Intelligence is a theoretical attribution. AGI is a system-level capability description.
Connection Theory keeps the established terms artificial intelligence (AI) and artificial general intelligence (AGI) because they are useful shared language in science and industry. Their use on this site should not be read as an ontological claim that every system conventionally called AI already satisfies the theoretical definition of intelligence used here.
AI is a field label
Artificial intelligence names a technological field and a broad class of engineered systems.
Calling a system AI therefore identifies its conventional technological category. It does not, by itself, establish why a particular behaviour should be attributed to intelligence in the theoretical sense developed here.
AI label ≠ evidence of intelligence
Intelligence is a theoretical attribution
On this site, intelligence is treated as a macroscopic attribution grounded in the behaviour of an organised system across relevant conditions and time.
The scientific task is to identify the causal organisation, active dynamics, environmental relation, retained capability, and adaptive processes that justify that attribution, rather than treating intelligence as a primitive substance located inside a component.
AGI is a system-level capability description
AGI is retained in its conventional industry sense: an artificial system displaying broad, robust, transferable capability across many domains and conditions.
Connection Theory does not treat “general intelligence” as a single primitive substance or hidden scalar that must reside inside one model. The explanatory question is instead what organisation and dynamics could generate sufficiently broad Effective System Capability to justify the label.
AGI ≠ a primitive substance called general intelligence
AI organization is an application class
An AI organization is an organised system containing artificial capability nodes, resources, interfaces, or tools. Calling it an AI organization does not assume that adding components improves capability.
Whether organisation adds value is a separate causal and empirical question. A well-matched organisation can integrate differentiated capabilities into a stronger whole; a redundant or badly coordinated organisation can add cost or make performance worse.
Why keep the conventional names?
Scientific revision does not require renaming every established field term. The names can remain while the explanatory concepts beneath them become more precise.
The rule used across this site is therefore:
Keep the conventional label where it aids communication; state separately what the theory requires before making a causal or behavioural attribution.