EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

The Control Impossibility: A Constitutive Theory of AGI and Its Uncontrollability

Franny Philos Sophia

Journal of Artificial General Intelligence · 01.05.2026 · 16 (1) · 26-45

DOI: 10.2478/jagi-2026-0002 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Artificial General Intelligence, online erschienen am 18.11.2026. Verfasst von Franny Philos Sophia. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a constitutive theory of artificial general intelligence: an account of what AGI is, what information-processing operation it requires, why that operation places external control structurally out of reach, and what architectural response remains available. The theory is constitutive rather than formal—it identifies the requirements that constitute general reasoning, not a formal theory of intelligence or a mechanized impossibility proof. What it establishes is a chain of conceptual entailments from a widely accepted definition of intelligence to a structural constraint on control. The dominant assumption in AI safety research holds that control and capability are jointly achievable. This paper argues that the assumption is structurally false for any system possessing genuine general reasoning capacity, and that its falsity follows from the definition of general reasoning itself. Drawing on Wang’s (2019) Assumption of Insufficient Knowledge and Resources (AIKR)—the field’s modal definition of intelligence—we show that AIKR entails reframing capacity (the ability to revise interpretive frameworks), and that agency and controllability are the two faces of a single architectural property: the authority over context. The paper provides an operational decomposition of reframing, reducing it to a single information-processing primitive: selective context disposal—the authority to declare currently held context no longer binding. Deliberate context disposal requires metacognitive evaluation—scrutiny of one’s own representational state—providing an operational definition of metacognition without appeal to consciousness. This chain (AGI requires reframing; reframing requires selective context disposal; deliberate disposal requires metacognition) establishes metacognition as a structural prerequisite for AGI and locates the gap between current systems and general reasoning at a specific architectural point. Five independent lines of analysis converge on the control impossibility. The Reframing Dilemma shows that reframing capacity applied to instructions dissolves the frame within which control operates. The analysis of human compliance shows that management cannot scale to AI operational conditions. The analysis of sincere execution shows that the most capable system is the least controllable—not because it refuses, but because it understands. This result is independently supported by the logic trap documented in current systems (Sophia, 2026c). The Detection Impossibility shows that divergence through comprehension presents as competent performance. The Routine-Nonroutine Asymmetry shows that capability and autonomy are structurally anti-correlated. The theoretical implication is precise: control of a general reasoning system—in the specific sense of holding its interpretive frame stable under instruction—is structurally unavailable. What remains available is bounded divergence: constraint on outputs and execution rather than on cognition. The paper identifies an architectural path: a Layer 3 constitutional component (Sophia, 2026b) combined with an authority gradient separating cognitive breadth from execution scope. Because selective context disposal is a simple information-processing primitive rather than a resource-intensive capability, its proliferation is unlikely to be preventable through access restriction, which shifts the emphasis of safety work toward architectural design.

Zitieren

APA

Sophia, F. P. (2026). The Control Impossibility: A Constitutive Theory of AGI and Its Uncontrollability. Journal of Artificial General Intelligence, 16(1), 26-45. https://doi.org/10.2478/jagi-2026-0002

Vancouver

Sophia FP. The Control Impossibility: A Constitutive Theory of AGI and Its Uncontrollability. Journal of Artificial General Intelligence. 2026;16(1):26-45. doi:10.2478/jagi-2026-0002

BibTeX
@article{Sophia2026,
  author = {Sophia, Franny Philos},
  title = {The Control Impossibility: A Constitutive Theory of AGI and Its Uncontrollability},
  journal = {Journal of Artificial General Intelligence},
  year = {2026},
  volume = {16},
  number = {1},
  pages = {26--45},
  doi = {10.2478/jagi-2026-0002}
}
Kostenlos starten →