Ceci n'est pas une pipe: AI systems as semantic abstractions

2026-07-10

AI Summary

Current AI systems often produce outputs mistakenly treated as factual, yet lack clear definitions and verification mechanisms for semantic correctness. This work proposes a tripartite analytical framework grounded in formal semantics and knowledge representation theory, which systematically distinguishes domain knowledge, authoritative reference sources, and the system’s currently available information. The framework provides the first precise characterization of typical semantic failure modes—such as unwarranted extrapolation, unsupported assertions, and mismatches between claimed knowledge and cited sources. By establishing verifiable semantic norms and an evaluative lexicon, it enables rigorous assessment of the reasonableness boundaries governing AI-generated content, citations, tool invocations, and real-world interventions.

Abstract

An AI system's output is not the fact or world state it appears to describe, but rather an engineered representation. We propose a semantic framework to describe AI systems, to be able to examine the correctness of such representations. To do so, we distinguish what is justified by accepted domain knowledge, what reference sources say, and what the system can currently use. This allows us to give precise definitions to common failures: extrapolation, refuted or unsupported assertion, sources versus knowledge mismatch, stale or refuted source, added hypotheses, unsupported use... We hope our framework gives a useful vocabulary for specifying and checking AI systems whose outputs, citations, tool calls, and world-changing actions must be justified by reliable claims and explicit authority rather than apparent fluency.

Problem

• AI systems
• semantic abstractions
• representation correctness
• justified claims
• authority

Innovation

• semantic framework
• AI justification
• knowledge representation
• source reliability
• assertion validation