Why LLMs Are Not a Higher Level of Abstraction in Computing

Summary

The article argues against the popular claim that Large Language Models (LLMs) represent a "higher level of abstraction" in computing. The author contends that LLMs are fundamentally different from traditional programming abstractions (binary → assembly → C → higher-level languages), which are deterministic, composable, and predictable. LLMs, by contrast, are probabilistic, non-deterministic, and lack the composability and predictability that define true abstractions. The piece uses the Alan Perlis quote to illustrate that different representations serve different purposes and are not simply hierarchical.