Stationary Assumptions, Kalman State-Space, and the Error of Statistical AI
For Norbert Wiener, knowing was filtering: separating a stationary signal from corrupting noise. Contemporary AI inherits this premise, mistaking the statistical suppression of error for genuine understanding and judgment.
From Comparative Mimicry Heuristic to the Ideology of Human Replacement
Turing's imitation game became a shorthand for machine intelligence. Its reception often collapses a question about observable performance into claims about understanding, then into a design preference for replacing people.
Deconstructing Accelerationist Fatalism through Vital Normativity
Modern discourse oscillates between viewing technology as a neutral tool or as an autonomous destiny. Both stances mask human architectural choices; genuine technical emancipation lies in Intelligence Augmentation as vital extension.
Channel Capacity, Semantic Neutrality, and the Limits of Statistical Fluency
In 1948, Claude Shannon separated transmission from semantics. With Large Language Models, treating statistical fluency as conscious understanding repeats an old error.