This position paper proposes a data-centric viewpoint of AI research, focusing on large language models (LLMs). We start by making the key observation that data is instrumental in the developmental (e.g., pretraining and fine-tuning) and inferential stages (e.g., in-context learning) of LLMs, and yet it receives disproportionally low attention from the research community. We identify four specific scenarios centered around data, covering data-centric benchmarks and data curation, data attribution, knowledge transfer, and inference contextualization. In each scenario, we underscore the importance of data, highlight promising research directions, and articulate the potential impacts on the research community and, where applicable, the society as a whole. For instance, we advocate for a suite of data-centric benchmarks tailored to the scale and complexity of data for LLMs. These benchmarks can be used to develop new data curation methods and document research efforts and results, which can help promote openness and transparency in AI and LLM research.
@article{arxiv.2406.14473,
title = {Data-Centric AI in the Age of Large Language Models},
author = {Xinyi Xu and Zhaoxuan Wu and Rui Qiao and Arun Verma and Yao Shu and Jingtan Wang and Xinyuan Niu and Zhenfeng He and Jiangwei Chen and Zijian Zhou and Gregory Kang Ruey Lau and Hieu Dao and Lucas Agussurja and Rachael Hwee Ling Sim and Xiaoqiang Lin and Wenyang Hu and Zhongxiang Dai and Pang Wei Koh and Bryan Kian Hsiang Low},
journal= {arXiv preprint arXiv:2406.14473},
year = {2024}
}