Why LLMs Are Not Like Compilers or Power Tools
blainehansen · hn · 2026-07-20
A long essay arguing that LLMs are not good analogies for compilers or power tools.
The core point is that those analogies overstate reliability and precision. Compilers transform code deterministically under a well-defined language spec; power tools extend human capability in a predictable physical way. LLMs, by contrast, are probabilistic systems that can be persuasive while still being wrong, so treating them like classic engineering tools can lead to bad expectations about correctness, safety, and workflow design.
The post pushes readers to think about LLMs as a different kind of technology altogether: useful, but fundamentally dependent on validation, supervision, and context.
More from AGI Musings
- ControlAI CEO says an international ban on superintelligence is needed to avert extinction risk — zetalyrae · 2026-07-22
- Gary Marcus says LLMs still cannot really do math on their own — GaryMarcus · 2026-07-22
- Gary Marcus says LLM math skills are like knowing only a car’s engine size — GaryMarcus · 2026-07-22
- AI may make digital work infinitely leveraged while offline life gets more human — illscience · 2026-07-22
- Better AI math could save researchers time by killing false conjectures earlier — prateekj · 2026-07-22
- AI’s economic forecasts are split by nearly a quadrillion dollars by 2035 — bittingthembits · 2026-07-22