Long thread argues LLMs are Edison's DC grid: hallucination is architecture, not a bug
2C_ornot2C · x · 2026-11-15
A widely shared thread makes a historical analogy: LLMs resemble Edison's DC power grid — genuinely revolutionary and commercially dominant today, but architecturally limited in ways that can't be patched.
Key claims:
- Hallucination is not a data bug but the architecture itself: formal results show LLMs cannot learn all computable functions and will inevitably hallucinate as general problem solvers
- Cited papers argue hallucination stems from the mathematical/logical structure of LLMs (computability theory, Gödel incompleteness), making it impossible to fully eliminate via data or fact-checking
- The author thus challenges the assumption that LLMs are the permanent foundation layer of AI
The retweet adds a reading list including "Hallucination is Inevitable" (arXiv:2401.11817), "LLMs Will Always Hallucinate," and work on the generalization-hallucination trade-off.
More from AGI Musings
- Pangram AI detector can't tell generated from edited text, yet its "100%" label wrecks careers — ipeirotis · 2026-11-15
- NDA-bound screenshot hints compute for given intelligence falls superexponentially — kevinnbass · 2026-11-15
- Reading list: the formal proofs that LLM hallucination is mathematically inevitable — 2C_ornot2C · 2026-11-15
- Economist: evidence of AI job losses still thin, but Bay Area legacy firms are cutting — alvelda · 2026-11-15
- AGI would still face a world running on decades-old infrastructure no one dares to touch — SuB8u · 2026-11-15
- Veteran designer: AI fuels credit-the-LLM copying, a toxic cheap trick — pixlpa · 2026-11-15