Essay.10.Explanability.in.the.Era.of.Large.language.models
Keywords: XAI, LLM
Abstract: The advent of Large Language Models (LLMs) has revolutionized natural lan- guage processing, exemplified by models like ChatGPT, llama2, and Gemini. This paper explores the capabilities and challenges posed by LLMs, considering the implications of the scaling law and recent investigations into emergent abilities. Investigating explainability in LLMs, we delve into research by Schaeffer et al. and considerations from "On the Dangers of Stochastic Parrots," highlighting bi- ases, static dataset constraints, and interpretability issues. While LLMs excel in benchmarks, questions persist about their real-world efficacy. We advocate for in- terdisciplinary collaboration, emphasizing advancements in model interpretability, fairness, and responsible deployment for unlocking the full potential of LLMs.
Submission Number: 213
Loading