jagged intelligence

Compiled with AI assistance and selectively reviewed. Please consult the cited passages; the books remain the authoritative source. About this companion

Editorial synthesis · provisional

Definition

Jagged intelligence is RB3’s description of uneven competence across tasks and milieus, where impressive symbolic performance can coexist with difficulty adapting under unfamiliar friction. The proposal preserves this unevenness as a problem of task and world coupling rather than proof that intelligence is absent, a fixed universal capability score or a verdict settling artificial general intelligence.

Explanation

The local treatment acknowledges competence while questioning what happens when archive or interface success encounters novelty. It does not infer from a failure that all performance is empty imitation. AB’s autobiographical example is retained as a reported experience in the manuscript, not an independently replicated benchmark result.

The ARC discussion concerns novel adaptation and is explicitly not a final court of judgment on intelligence. Its native references remain source apparatus rather than a new survey of present model capabilities. The proposed entry adds no ranking, percentage improvement or generalized claim about every available model.

Friction is useful only when it connects activity with relevant correction. It can also merely obstruct, so more difficulty does not automatically improve a system. The source’s contrast is between different couplings and consequences, preserving the need to evaluate what the specific task and milieu actually demand.

Scope and attribution

Uneven task adaptation differs from absent intelligence, a final AGI verdict or a belief that all friction helps.

AB — The Actual Book

Source-verified occurrence; see the printed-page references below.

Occurrence locators: 151; Notes: 245.

RB3 — The Real Book, Volume III: Murmurations

The developed account belongs to RB3 in the cited local scope, with antecedents and qualifications retained.

Occurrence locators: 160, 161.