Can AI Express Sentiment Like Humans in Research Abstracts? Implications for Academic Publishing
Learned Publishing · 2027-01 · 40 (1)
DOI: 10.1002/leap.2126 ↗At a glance
Journal article in Learned Publishing, published online on 2026-11-20. Written by Ying Wei and 1 co-authors. 43 references. The full text is freely available (open access, CC BY-NC-ND 4.0).
Abstract (original language)
Sentiment is widely examined in academic and non‐academic contexts. However, with the advancement of generative artificial intelligence, no study has compared linguistic positivity in AI‐generated and human‐written academic texts. Adopting sentiment analysis, this study compares the sentiment scores of 420 research abstracts published in top journals across 14 hard and soft disciplines with 420 abstracts generated by DeepSeek. Our findings show that DeepSeek‐generated abstracts and human‐written journal abstracts show no significant difference in terms of their sentiment distribution. This pattern is observed across hard and the majority of the soft disciplines. These results indicate that AI‐generated abstracts closely approximate human‐written abstracts in linguistic positivity, as shown in lexicon‐based sentiment scores. Given the findings, this study envisions human‐AI collaboration in future knowledge production and calls for adjustments to current frameworks of academic evaluation.
Cite
APA
Wei, Y., & Yao, M. (2027). Can AI Express Sentiment Like Humans in Research Abstracts? Implications for Academic Publishing. Learned Publishing, 40(1). https://doi.org/10.1002/leap.2126
Vancouver
Wei Y, Yao M. Can AI Express Sentiment Like Humans in Research Abstracts? Implications for Academic Publishing. Learned Publishing. 2027;40(1). doi:10.1002/leap.2126
BibTeX
@article{Wei2027,
author = {Wei, Ying and Yao, Mingxin},
title = {Can AI Express Sentiment Like Humans in Research Abstracts? Implications for Academic Publishing},
journal = {Learned Publishing},
year = {2027},
volume = {40},
number = {1},
doi = {10.1002/leap.2126}
}
EUVIMED