Cognitive Information Theory
Structural Limits of Objective-Function AI through Irreducible Entropy, Causal Ambiguity, and Sequential Distribution Drift
.png)
Abstract
Contemporary AI assumes accuracy improves monotonically with scale. We challenge this through Cognitive Information Theory (CIT), which decomposes cognitive information into a reducible component (structural complexity) and two irreducible ones (channel entropy and causal ambiguity). For systems governed by a scalar objective (the class Sof ), CIT proves that when either irreducible component is non-zero, no amount of data, capacity, or calibration guarantees error-free predictions.
Read the paper on squinting pipelines
Download the paper to learn more.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Please enter a valid email and try again.