Rethinking Artificial Intelligence

The modern school of Artificial Intelligence was originally expected to provide a full working model of intelligence as a set of procedures. Scholars implemented these procedures over time to conceptualize the notion of an intelligent machine. Computer scientists rushed to implement working models that would allegedly reach beyond many limits. Perhaps the most debilitating act was equating what is efficient in procedures to what is artificialized in intelligence. Equally debilitating was interpreting the speed of arithmetic calculations as a quantifier: it led to teams being interpreting speed and accuracy as reflections of intelligence. In order to reach an artificial form of intelligence that is faithful to the amalgam of biological, physical and chemical that it seeks to imitate; scholars of AI must reach a deeper synthesis of its integrative nature, leading to the creation of many artificial synthetic forms of Intelligence, instead of a single vision of intelligence that simply focuses on matching the performance of the human brain. Having said that, we can clearly concur that most of the AI Modern School's limitations have been discovered and are well-documented and known to the AI community. Our aim is to discuss a number of these issues, particularly the limits previously described. We avow that these limits emerged from epistemological misunderstandings on the perceived meanings of intelligence itself, leading to the limits imposed in the current interpretations of AI. Future work in AI, or alternatively coined Synthetic Intelligence, must revisit fundamental assumptions about the nature of the brain, cognition, computing, and intelligence. Synthetic Intelligence focuses on the phenomena such as intelligence and consciousness, and mapping them to the physics of the brain and models of brain processes at each of its multiple levels. It is the ‘stack’ of brain subsystems at multiple levels, from cortical down to molecular, joined by a common thread, that make up a mind. What we need are mathematically described mechanisms and information structures to integrate computational discourse analysis, value systems, mapping of cognitive structures to neuron interactions and to the molecular mechanisms of such interactions. The key to this discovery will be the study of emergence of intelligence and consciousness in engineered systems - implemented in silico or in vitro.