[Approximate Reading Time : 4 mins]
Artificial intelligence (AI) is gradually making its presence felt in all spheres of life, and the field of academic and scientific publishing is no exception. In recent times, the speed at which academic manuscripts are being submitted for publication has increased significantly. The volumes are also substantially high, and this is proving to be a strain on journal editors. They are under immense pressure to find peer reviewers quickly to evaluate the submissions before publication.
Statistics indicate that manuscript submissions to peer-reviewed journals are increasing by 6 percent annually. It is estimated that over 15 million hours are spent on the review of manuscripts that were rejected and submitted once again to other journals.
In the normal course of things, when an article is submitted, the editorial staff will give it a quick look to determine if it’s suitable for the journal. At this time, they will run a plagiarism check and also see whether the structure of the article matches with the tone of the journal. While this may seem a simple process that takes just a few minutes, the magnitude of effort involved becomes starkly clear when you multiply this time by hundreds or thousands of submissions.
After this comes the task of finding peer reviewers—something that is becoming tougher with each year. This is primarily because peer reviewing is a mundane and, not to forget, time-consuming task. This is where AI comes in.
How Does It Work?
Experiments are going on to determine if the power of AI can be harnessed to screen and evaluate these manuscripts, thus reducing the burden on human peer reviewers. We can use AI in peer reviewing to simplify the process for human reviewers and make it more enjoyable. This can be done by using AI solutions to perform initial screenings and then matching the manuscripts with the right reviewers. Since individual reviewers will have their own opinions and conclusions about each paper, AI can help calibrate reviewer scores. Similarly, AI can be used to pinpoint missing data and invalid conclusions. This will help reviewers identify problem areas quickly and complete the review process faster.
Ethical Challenges
One major concern with respect to AI peer reviewing is that the likelihood of preexisting biases being reinforced is much higher. For instance, the rejection rate of papers submitted by authors from countries that have historically been on the sidelines with respect to scientific literature is likely to be higher with AI. This is because AI tools will rely largely on the biases exhibited by prior human reviewers and might not take into account the enhanced quality of submissions from these countries over the past few years.
A combination of AI and human reviewing looks to be an acceptable solution for now, until the tools for automated peer review are further refined.
Amnet can help you with the print and digital publishing for all your research writing and academic publishing needs. We cater to various segments, including university presses and university libraries.
Sources:
1.
https://blogs.lse.ac.uk/impactofsocialsciences/2021/05/17/can-ai-be-used-ethically-to-assist-peer-review/.
2. https://thepublicationplan.com/2021/05/14/will-machines-replace-peer-reviewers-the-role-of-artificial-intelligence-in-scientific-publishing/.
3. https://www.scholarcy.com/how-reviewers-can-use-ai-right-now-to-make-peer-review-easier/.
