AAIA – The Architecture of Real Riflective Intelligence Appendix 2 - Technical foundations

AAIA – The Architecture of Autocognitive Artificial Intelligence. Abstract. The contemporary debate on artificial consciousness and cognitive architectures revolves-each in its own way-around the same fundamental question: how far can we push the simulation of mental functions without betraying their meaning? Recent work can be broadly grouped into three major families of approaches: Strong neurobiological models, which treat the human brain as a binding benchmark for any hypothesis of artificial consciousness. Cognitive and metacognitive architectures, which aim to enhance systems’ control, explainability, and adaptive capabilities without necessarily committing to the question of “consciousness”. Conceptual and skeptical positions, which deny the very possibility of artificial consciousness or warn against the linguistic illusions created by current systems. AAIA is explicitly conceived after this tripartition, not in opposition to it. It takes the best of each tradition as given and shifts the focus of the question: rather than asking whether a machine can be conscious in a human sense, it asks how to design a machine endowed with an internal, operative, teleological reflexivity-a “real reflective” capability-able to evaluate itself, its own actions, and its coherence over time, in line with how a newborn behaves, according to developmental psychology.