The cuts

John Searle
Syntax is never sufficient for semantics. A room that manipulates Chinese symbols perfectly still understands nothing.

Split 'behaves as if it understands' from 'understands' — and gave the split a scene everyone could argue about.

Minds, Brains, and Programs1980
Hans-Georg Gadamer
Understanding is a fusion of horizons — never presuppositionless, always a dialogue between the interpreter and a living tradition.

Framed understanding as participatory and historical two decades before AI made the question urgent — the tradition Searle's critics keep rediscovering.

Truth and Method1960
Geoffrey Hinton
Understanding is compression. To predict the next word well enough, you must model the world that produced it.

Made 'it's just statistics' a position that has to be argued for, not assumed.

The risks and promise of AI, according to Geoffrey Hinton (60 Minutes interview)2023
Emily Bender
A system trained only on form has no access to meaning. Fluency is not grounding.

Named the modern failure mode — the stochastic parrot — and forced the grounding question into the LLM era.

Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data2020

The tensions

Where it sits

Upstream · downstream

Where this lives on the site

Theme
Language
Support
© funclosure 2025