Large Language Models for Linguists

Large Language Models for Linguists: This book bridges the gap between linguistics and technology. It provides an in-depth education on large language models for linguists and demonstrates how AI is being taught today using language as data without the nuances of language. It gives a playbook for linguists to get involved in teaching AI in a way that preserves the world's diverse linguistic and cultural heritage. Key Features: ? Minimum Prerequisites: No prior knowledge of technology or coding is required, making this accessible to anyone interested in pivoting into AI for languages. ? Demystifies Large Language Models (LLMs) and the technologies needed to build AI for languages including: Natural Language Processing, NERs, hallucinations, fine-tuning and custom-LLM for languages. ? Brings an extensive list of diverse, global real-world research, language datasets and LLM models for world languages covering: Portuguese, Vietnamese, Yoruba, Tamil, Hindi, Swahili, Spanish, Korean, Arabic, Italian, Multicultural London English, Indigenous languages, Sign Languages and more. ? Highlights how historical trauma, colonialism, migration, and resilience are embedded in language, shares well researched history and books used to train AI, and explains why AI must be built with an understanding and respect for this complexity. ? Provides a practical playbook for linguists and cultural stewards to build datasets and benchmarks that honor the essence of their communities with diverse language examples showcasing how vocabulary carries cultural meaning of places, history, traditions and the identity of people. ? Dynamic Online Resources: Additional resources for linguists to find language datasets and a community to collaborate to career pivot to build AI for your language. (LLMLinguist.org) Ideal for those in humanities looking to understand Large Language Model (LLM) and AI looking for a practical guide to influence AI's linguistic future, and for computational linguistics and technologists aiming to create language models that reflect-rather than erase-the richness of the world's voices.