Emily Bender
Computational Linguist
About
Emily M. Bender is an American computational linguist at the University of Washington whose work sharpened the central objection to LLM understanding. Her 2020 'octopus test' (with Alexander Koller) argued that a system trained only on linguistic form has no access to meaning; the 2021 'Stochastic Parrots' paper (with Timnit Gebru and colleagues) named the risks of ever-larger language models and became one of the most cited critiques in the field. She brings a linguist's discipline to questions the AI industry often treats as settled: what language is, what meaning requires, and who bears the costs when the two are confused.
Key Contributions
- Co-authored 'Climbing towards NLU' (2020), arguing form alone cannot yield meaning — the octopus test
- Co-authored 'On the Dangers of Stochastic Parrots' (2021), the era-defining critique of scale
- Coined vocabulary the debate now runs on, from stochastic parrots to form vs. meaning
- Advocates for linguistic rigor and documentation practices (Bender Rule, data statements) in NLP
- Co-authored The AI Con (2025), a book-length case against AI hype
Questions they sharpened View the streams
Papers & Publications
Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data
2020With Alexander Koller, the octopus test: a system trained only on form has no path to meaning.
Read PaperOn the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
2021The era-defining critique of ever-larger language models and the risks of scale.
Read PaperConnections
Blaise Agüera y Arcas
DebatedVP & Fellow, Google
In December 2021 Agüera y Arcas published 'Do large language models understand us?', arguing that statistics do amount to understanding in any falsifiable sense; Bender answered in January with a line-by-line rebuttal that rejected both the claim and his analogies between language models and blind and Deafblind people. It is the live version of the argument her octopus test had staged in the abstract: one side holds that meaning requires grounding outside the text, the other that understanding was always relational and that denying it to models is a double standard. This site's question 'Does an LLM actually understand?' carries both of their cuts.
medium.com · medium.com · aclanthology.org
John Searle
KindredPhilosopher of Mind & Language
An interpretive pairing across forty years: Searle's Chinese Room and Bender's octopus both ask whether a system that manipulates symbols flawlessly has any purchase on what they mean. The difference is where each gets its force. Searle argues a priori, about intentionality and what a program could never in principle possess; Bender argues as a linguist about what a corpus of pure form can and cannot supply. One says the room could never understand; the other says nobody has shown where the meaning would get in.
Geoffrey Hinton
In contrastAI Pioneer & Researcher
They hold opposite cuts on two of this site's questions, 'What is understanding?' and 'Does an LLM actually understand?'. Hinton's answer is that predicting the next word well enough forces real compression of the world that produced it — understanding is what good compression amounts to. Bender's is that a system trained only on form never had access to meaning, and that fluency is precisely what makes the confusion so easy. Nothing in the observed behaviour separates them; what separates them is what each will accept as evidence.
Gary Marcus
KindredCognitive Scientist, AI Critic & Author
The two most persistent critics of the LLM era, arguing from disciplines that barely overlap. Marcus comes from cognitive science and objects that these systems have no robust abstraction and no causal model — a claim about architecture, and an implicit request for a better machine. Bender comes from linguistics and objects that form without grounding was never a route to meaning, and that the costs of pretending otherwise fall on people. They are filed together as skeptics, but only one of them is asking for a fix.
Melanie Mitchell
KindredProfessor, Santa Fe Institute
Two of the field's most careful skeptics about LLM understanding, operating at different altitudes. Bender is a pole of the debate — form without grounding cannot be meaning, full stop; Mitchell is its cartographer, whose PNAS survey with David Krakauer mapped both sides without planting a flag. Read Bender to feel the argument's force, Mitchell to see its shape.
