Shoggoth
Version 1.0.0 · Updated 2026-07-30
CORE DEFINITION
A popular metaphor in the AI community. A large language model (LLM) is like a formless, many-eyed, terrifying monster (a shoggoth), which is actually a pile of incomprehensible mathematical matrices. Through RLHF (Reinforcement Learning from Human Feedback), we put a 'smiley face mask' on it to make it look like a polite assistant.
SCAFFOLDING EFFECT
Reduce cognitive load
Cognition of inconsistency between appearance and reality. It reminds us not to be fooled by the anthropomorphic appearance of AI. Its core is alien-like, chaotic probabilistic statistics, and at the moment the 'mask' slips (jailbreaks, hallucinations), its uncontrollable nature is exposed.
Anchor fast decisions
A formless monster from Lovecraftian mythos, borrowed by the AI community as a metaphor for LLMs: the underlying layer is an unexplainable probabilistic tensor composed of massive parameters ('alien-like chaos'), which is 'given a smiley face mask' through RLHF/alignment training to appear polite and controllable. The mask (alignment) is a shallow constraint; the core is still random sampling, so during jailbreaks or distribution shifts, the 'mask slips' to reveal uncontrollability and hallucinations.
MINIMUM ACTION
In progress 0/4Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Shoggothverified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS