Generative AI's Impact on Academic Writing: Boon or Bane?

Generative AI has significantly impacted academic writing, increasing productivity but raising concerns over quality. While non-native English speakers benefit from improved language skills, AI-generated texts often disguise lower-quality research. Diverse sources are now more accessible, but critical evaluations of scholarly work remain essential.

Adelaide, Dec 19 (The Conversation) - Over the past three years, the influence of generative artificial intelligence (AI) on society has been profound, particularly in human writing. Large language models like ChatGPT, trained on diverse textual data, can now produce intricate and high-quality texts autonomously.

AI's widespread application has led to the hyperproduction of low-quality outputs, or 'AI slop,' generated with minimal human effort. This phenomenon raises questions about AI's role in education, work, and culture, as well as its impact on scientific writing. According to a fresh study from UC Berkeley and Cornell University, appearing in Science, it appears that 'scientific AI slop' is prevailing.

The study analyzed over a million preprint abstracts from 2018 to 2024, exploring AI's link to productivity, manuscript quality, and literature diversity. Results showed a dramatic increase in preprint production following AI adoption, especially among non-native English speakers. Articles written with AI employed more complex language, often at the expense of quality. Diverse sources have become more accessible due to AI-driven search engines, underscoring the need for critical evaluations in academia.