A recent essay by Stefan Wolfram argues the success of large language models is evidence for your theory of language. He says their means are otherwise too limited to be successful. But you say this is partly true, but partly misleading. Large language models have a fundamental property which demonstrates that they cannot tell you anything about language and thought. They work just as well for impossible languages as for possible languages. So therefore, in principle, it's telling you nothing about language. You can take a smartest chimpanzee or the dogs under my desk, they can listen to this noise forever. That's a fundamental property of humans built in.
Noam Chomsky joins Tyler to discuss why Noam and Wilhelm von Humboldt have similar views on language and liberty, good and bad evolutionary approaches to language, what he thinks Stephen Wolfram gets wrong about LLMs, whether he’s optimistic about the future, what he thinks of Thomas Schelling, the legacy of the 1960s-era left libertarians, the development trajectories of Nicaragua and Cuba, why he still answers every email, what he’s been most wrong about, and more.
Read a full transcript enhanced with helpful links, or watch the full video.
Recorded February 27th, 2023
Other ways to connect
Photo credit: Duncan Rawlinson - Duncan.co