Intelligence as Selection: Ashby, Beer, and the Design of Viable Autonomous Vehicles

Purpose — This paper applies Ashby's definition of intelligence as appropriate selection bounded by information, together with Beer's Viable System Model (VSM), to Autonomous Vehicle (AV) architecture. It argues that what the industry frames as a deficit of general cognitive capability reduces, in Ashby's formalism, to a deficit of information in the system's regulatory organisation.

Approach — A derivation chain from Ashby's intelligence bound through the Law of Requisite Variety and the Good Regulator theorem to Beer's VSM establishes the structural conditions for autonomous system viability. The VSM's five systems are instantiated in the autonomous vehicle domain, and documented engineering decisions and system failures are diagnosed against this mapping. A structured search of open bibliographic sources found no prior application of these frameworks to autonomous vehicle architecture, and identified a three-way disconnect between Cybernetics, the IEEE Intelligent Control tradition, and the contemporary autonomous vehicle field.

Findings — The autonomous vehicle literature contains no formal, control-theoretic definition of intelligence despite the word's ubiquity. AV failures map onto identifiable VSM pathologies — notably absent S3* (independent audit) and degraded S4 (reduced sensor variety). Architectures that persist converge empirically on the VSM structure. The claim that full self-driving requires AGI dissolves when intelligence is formalised as appropriate selection bounded by information capacity.

Originality — The paper establishes a Cybernetic diagnostic framework for autonomous vehicle architecture, identifies a literature gap spanning three disciplines, and demonstrates that the "AGI requirement" for autonomous driving is a category error: what is required is not a cognitive breakthrough but sufficient information, received and processed, to achieve appropriate selection across the driving environment's variety.