Unverified50% confidenceOpinionExact time
The "stochastic parrots" camp argues that LLMs have never been exposed to the real world that linguistic symbols refer to, and merely stitch together text that looks plausible in a statistical sense.
1
Sources
50%
Confidence
Long-term
Relevance
10/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGoogle internally once believed that large language models were not real things at all, feeling that no one would actually want to use them.66% similarVerifiedText generated by large language models tends to have low perplexity, with word choices that are overly consistent with probabilistic expectations, lacking the unexpectedness of human writing.65% similarVerifiedThe stochastic parrot hypothesis, proposed by Bender et al. in 2021, questions whether large language models truly understand the meaning of language or merely reproduce statistical patterns in training data.64% similarUnverifiedThe author argues that large language models can arrange words in grammatically and semantically convincing ways, yet do not hold any opinions.64% similarUnverifiedA recent study found that large language models do not consistently follow Bayesian rules, and their probabilistic beliefs exhibit internal inconsistencies.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/997594API
curl https://kongchang.com/api/v1/knowledge/claims/997594MCP
get_claim(id=997594)