Cognitive Scaffold

Preparing your thinking workspace

arrow_back_ios_new
MENTAL MODEL · M1903

AI Hallucination

AI Hallucination
TechnicalHigh supportArtificial Intelligence
Included
account_tree

Version 1.0.0 · Updated 2026-07-30

CORE DEFINITION

Generative AI can confidently produce nonsense (fabricating facts, citing non-existent papers) because it is essentially doing 'probabilistic prediction' (what is the most likely next word) rather than 'fact retrieval'.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

- Verification awareness: When using AI (or listening to certain experts), always maintain the vigilance of fact-checking. Fluency does not equal truth, and confidence does not equal correctness.

anchor

Anchor fast decisions

Based on 'autoregressive generation'. The model predicts the next token by probability, aiming to 'seem real' rather than 'be true', without factual grounding, thus fabricating plausible errors.

MINIMUM ACTION

In progress 0/1

Practice this model in one real situation:

Check to track your progress (stored locally)
Learning progress0%
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more

Source support: Explicit

  • link
    zh.wikipedia.orghttps://zh.wikipedia.org/wiki/%E5%B9%BB%E8%A7%89_%28%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%29ZH · Explicit
    verified

RELATED MODELS

Related model

Vanishing Gradient

Very small backpropagated gradients can slow or stall learning in earlier layers or across long time intervals.

arrow_forwardRelated model

Self-Attention Mechanism

Generates queries, keys, and values from the same sequence and aggregates representations using position-wise compatibility; masks and other constraints determine which positions are visible.

arrow_forwardRelated model

Maes-Garreau Law

The Maes–Garreau law is the statement that "most favorable predictions about future technology will fall within the Maes–Garreau point", defined as "the latest possible date a prediction can come true".

arrow_forward