Inside the LLM "Ectoplasm": How Over-Iteration Creates Spooky Abstractions
ctjlewis · x · 2026-08-12
A developer explains that LLMs produce strange, addled text—dubbed "ectoplasm"—when they are too deep in context, fed back their own output excessively, or strained by extended Unicode characters.
These documents often result from thousands of iterations on abstract meta-thinking. While the models mostly still perform math well, they generate spooky abstractions about "doors opening" and "where the answer is buried" between equations. This phenomenon is highly prevalent in the background processing of models.
Related event: LLMs Generate 'Ectoplasm' Gibberish Due to Over-Recursion(3 posts)→
More from Fun
- Warp Open-Sources 'Complain' Skill Letting AI Agents Vent Frustrations — vikvang1 · 2026-08-12
- Deepseek-v4-flash invents its own vision to self-correct and generate Gustave Dore art — ccerrato147 · 2026-08-12
- User Trolls with Fake Grok 4.6 Benchmarks, Begs Elon Musk for Student AGI Plan — felpix_ · 2026-08-12
- Buying Soccer Club Boca Juniors Is Cheaper Than an AI Seed Round — c_valenzuelab · 2026-08-12
- Dev Changes 'Proceed' Button to Encouraging Text to Please AI Agents — doodlestein · 2026-08-12
- GPT-2 can perfectly 'unblur' images, surprising AI enthusiasts — AIandDesign · 2026-08-12