Large Language Models Understand Language
Claim: Large language models genuinely understand and comprehend human language rather than merely predicting tokens through statistical pattern matching.
Bullshit Meter
Pure ontological fiction, excluding real psychological effects.
Physics vs. mind loop
Does it work in the physical world, or only through belief and behaviour?
Works physically, independent of belief.
Bio & cult hazard
6.9%Risk of abandoning medicine and of getting trapped in a closed group.
Topological fingerprint
The structural profile of the claim on five axes.
- Empirical rigor
- 51.3%
- Magic & dogma
- 6.6%
- Rhetorical fog
- 20.1%
- Conspiracy
- 3.3%
- Psychosocial
- 16.1%
Ockham's guillotine
What remains after cutting away the unfalsifiable superstructure.
Scam & grift
0.9%How much it works as a money-extraction funnel.
Fog & Barnum
40.4%Protection from refutation through vagueness and stolen jargon.
Energy, frequency, vibration, quantum alignment, natural harmony, ancient wisdom, holistic balance…
- Barnum effect11.0%
- Motte-and-bailey41.8%
- Stolen jargon19.0%
Guru roulette
29.6%Does the result depend on who performs it?
Repeatability between operators: 70.4%
- Special gift needed9.8%
- Contradicting experts17.7%
- "Skeptics block it"2.1%
Fossil & dogma clock
44.8%Does the field correct its errors, or is it frozen?
- 🚀Active science51.0%
- 🩹Degenerative patching39.8%
- 🎭Marketing chameleon6.1%
- 🦴Fossil in amber3.1%
Tin-foil hat reactor
3.0%How big a global conspiracy the claim needs.
Paranoia level 1/5 · No conspiracy needed