Large language models (LLMs) and linguistics both deal with human languages, but the relationship between the two fields is unclear. While some researchers argue that LLMs provide valuable insights into language learning, others believe they offer little contribution to linguistics. The rise of natural language processing and LLMs has sparked renewed interest in computational models of language, with applications ranging from machine translation to vaccine design. However, the debate continues on whether LLMs truly understand language or merely mimic it, and whether their cost is justified.
The field of linguistics is being influenced by the development of powerful tools like LLMs, which can generate text that is easily mistaken for human-written content. However, while LLMs excel at generating meaningful and coherent text, they lack functional competence, which involves world knowledge and pragmatics. The future may see LLMs diverging even further from human cognition, but there is hope that they will be utilized effectively in future linguistic research.