Towards AI-assisted Academic Writing

TL;DR

The paper addresses the challenge of AI‑assisted academic writing by proposing two live in‑editor capabilities: contextual citation recommendation and automatic introduction generation. It implements these via a pipeline that combines BibTeX, OpenAlex, and SPECTER2 embeddings within a Retrieval Augmented Generation framework, plus a three‑step prompt chain to produce structured introductions. Quantitative evaluation on context‑rich citation retrieval using $P@k$ and $MRR$ metrics and qualitative/linguistic evaluation on 14 NBER introductions (with $ROUGE$ and entailment analysis) demonstrate the approach's ability to surface relevant citations and generate coherent, claim‑aware introductions. Qualitative interviews with researchers reveal workflow needs and design considerations, underscoring the potential for AI writing assistants to augment, rather than replace, scholarly authors and outlining a rich space for future improvements in interpretability and multilingual support.

Abstract

We present components of an AI-assisted academic writing system including citation recommendation and introduction writing. The system recommends citations by considering the user's current document context to provide relevant suggestions. It generates introductions in a structured fashion, situating the contributions of the research relative to prior work. We demonstrate the effectiveness of the components through quantitative evaluations. Finally, the paper presents qualitative research exploring how researchers incorporate citations into their writing workflows. Our findings indicate that there is demand for precise AI-assisted writing systems and simple, effective methods for meeting those needs.

Figures (3)

• Figure 1: Flowchart from paper and citations to written introduction.
• Figure 2: Distribution of ROUGE-1 scores for the generated introductions.
• Figure 3: Distribution of scores for whether claims from the generated introductions entail the original introduction based on an LLM.