AI Alignment Foundation from First Principles: Cybernetic, Mechatronic and Engineering Aspects

“AI Alignment” is a moniker describing a community and associated literature of computer science researchers, interdisciplinary thinkers, and economic actors engaged with the frontier of AI development. The Ordi- nary Language word “aligning”, as a verb, typically has some correspon- dence to the patterns of discourse in the field, e.g., aligning technology to some model of the world or desiderata of a stakeholder. However, to this date, no formalism and attempt to define and axiomatize the term and its research program from First Principles has been comprehensively attempted, let alone appearing in consensus among the community asso- ciated with the memetics of “AI Alignment”. We present a foundation for enabling good faith epistemology in studying “AI Alignment”, based upon the substrates composing AI operation. The foundations are based on Badiou’s radical enshrinement of mathematical equation systems as ontology itself, as motivated by the intention to cleave apart culturally contextual terminology obfuscating research in the topic. We hope this can present a solid foundation that facilitates greater epistemic virtue in the grand research program of aligning AI, at large.