Kolmogorov complexity, AIXI and undecidability: why intelligence lives in problems compute can't swallow

JensHonack · x · 2026-11-18

A theoretical thread connecting Kolmogorov complexity (intelligence via the AIXI framework), MDL (complex structures via epiplexity), and the Second Recursion Theorem (self-referential, self-replicating programs via functional equivalence). The author argues their shared incomputability and undecidability is no coincidence: intelligence and consciousness inevitably reflect undecidable problems, and only by confronting them can life pour in infinite compute. Scaling laws, he suggests, should be read this way — invest compute in problems compute itself can never swallow.

Original post →

More from AGI Musings

AGI Musings channel →