Status Updates From Inteligencia Artificial

A truly intelligent AGI would possess functional consciousness. For example, it would focus on (pay attention to, be aware of ) different things at different times. A human-level system would be able also to deliberate, and self-reflect. It could generate creative ideas, and even deliberately evaluate them. Without those capacities, it couldn't generate seemingly intelligent performance.

Radical novelty in AI does indeed require outside influences, because it's true that a program cannot surpass its possibility space. But these influences needn't be physical. A GA system connected to the Internet might evolve fundamental novelties by interacting with a virtual world.

From about 1960, however, an intellectual schism developed. Broadly speaking, those interested in life stayed in cybernetics, and those interested in mind turned to symbolic computing. The network enthusiasts were interested in both brain and mind, of course. But they studied associative learning in general, not specific semantic content or reasoning, so fell within cybernetics rather than symbolic AI.