[Paper Review] A Cyber Science Based Ontology for Artificial General Intelligence Containment

Jason M. Pittman, Courtney E. Soboleski|arXiv (Cornell University)|Jan 28, 2018

Scientific Computing and Data Management36 references3 citations

TL;DR

This paper proposes a cyber science-based ontology to formalize Artificial General Intelligence (AGI) containment, structuring core elements like AGI, humans, and the cyber world into a 5-level framework with 32 descriptors. The key contribution is a systematic, scientifically rigorous foundation for future AGI containment research, addressing gaps in conceptual clarity, contextual framing, and methodological rigor.

ABSTRACT

The development of artificial general intelligence is considered by many to be inevitable. What such intelligence does after becoming aware is not so certain. To that end, research suggests that the likelihood of artificial general intelligence becoming hostile to humans is significant enough to warrant inquiry into methods to limit such potential. Thus, containment of artificial general intelligence is a timely and meaningful research topic. While there is limited research exploring possible containment strategies, such work is bounded by the underlying field the strategies draw upon. Accordingly, we set out to construct an ontology to describe necessary elements in any future containment technology. Using existing academic literature, we developed a single domain ontology containing five levels, 32 codes, and 32 associated descriptors. Further, we constructed ontology diagrams to demonstrate intended relationships. We then identified humans, AGI, and the cyber world as novel agent objects necessary for future containment activities. Collectively, the work addresses three critical gaps: (a) identifying and arranging fundamental constructs; (b) situating AGI containment within cyber science; and (c) developing scientific rigor within the field.

Motivation & Objective

• Address the lack of systematic conceptual frameworks for AGI containment, which currently relies on ad hoc cybersecurity techniques.
• Establish a formal ontology grounded in cyber science to structure fundamental constructs of AGI containment.
• Position AGI containment within the broader, dynamic cyber world rather than a simplistic human-vs-AGI binary.
• Provide a replicable, axiomatic foundation to enable scientific inquiry and shared understanding in AGI safety research.
• Bridge gaps in scientific rigor and conceptual clarity that currently hinder progress in AGI containment

Proposed method

• Developed a single-domain ontology with five hierarchical levels, 32 atomic codes, and 32 associated descriptors derived from academic literature.
• Defined three novel agent objects: humans, AGI, and the cyber world, as central entities in containment relationships.
• Modeled relationships (R), subattributes (S), and features (F) using a formal diagrammatic structure to represent containment dynamics.
• Applied cyber science principles—particularly Kott’s (2015) framework on malicious software and intent—to reframe AGI and containment as active, interactive agents.
• Formalized containment as an active, dynamic relationship between agents in the cyber world, distinguishing attack (initiated by AGI) and defense (initiated by human/defender).
• Introduced a functional model: f(c) = k for containment equilibrium, f(c) = k+1 for attack, with initiation and intent as dual features of attack

Experimental results

Research questions

• RQ1 How can a formal ontology be constructed to systematize the fundamental constructs of AGI containment?
• RQ2 In what way does situating AGI containment within cyber science improve conceptual clarity and research rigor?
• RQ3 What are the essential agent objects and relationships that define AGI containment in a dynamic cyber world?
• RQ4 How can attack and defense be formally distinguished in AGI containment, particularly in terms of initiation and intent?
• RQ5 Can a cyber science-based ontology enable a shift from binary, adversarial models to more nuanced, scientifically grounded containment strategies?

Key findings

• The study successfully developed a 5-level ontology with 32 descriptors, providing a formalized, structured framework for AGI containment research.
• The ontology identifies three core agent objects—humans, AGI, and the cyber world—as essential for modeling containment dynamics.
• Containment is formalized as an active, dynamic relationship in the cyber world, not a static or binary state, with distinct roles for attack and defense.
• Attack is defined by two simultaneous features: initiation (action by agent) and intent (preemptive or responsive), while defense is always reactive and singular in initiation.
• The model f(c) = k represents equilibrium containment, while f(c) = k+1 models attack as a deviation from equilibrium, enabling formal analysis of containment failure.
• The ontology enables a shift from ad hoc cybersecurity approaches to a scientifically grounded, replicable framework for future AGI containment research