Overview
- Frontiers’ new whitepaper reports that 53% of reviewers use AI tools, based on a global survey of 1,645 researchers with especially high uptake among early-career scholars and in regions such as China and parts of Africa.
- Among AI users, 59% rely on it to draft reviewer reports, 29% to summarize manuscripts or check references, and 28% to flag potential misconduct such as plagiarism and image duplication.
- Frontiers has launched an in-house AI platform for its journals and is urging sector-wide updates to publishing rules to reflect current practice.
- Publishers permit only limited AI assistance, require disclosure, and prohibit uploading unpublished manuscripts to third-party chatbots to protect confidentiality and intellectual property.
- Recent tests of large language models, including a GPT-5 experiment, show they can produce polished review-style text but often miss constructive, accurate critique, reinforcing the need for human oversight.