[Approximate Reading Time : 4 mins]
Artificial intelligence (AI) now has real-world application in every industry. The publishing industry is no exception. Research publishing requires much manpower throughout its workflow. With an increasing influx of manuscripts every year, it is one of the potential fields where AI can make a positive difference.
Use of AI
The ultimate purpose of AI is to replicate human thinking without bias. Simple features like speech or text recognition use AI for processing. Currently, AI only does simple plagiarism, bibliography, quality, spelling, and grammar checks. Journals screen and scrutinize research works with the help of peer review administrators and peer reviewers, but this is becoming very tedious. So AI-based review assistance has been developed and put to use by some publishers to perform basic checks that ensure quality. In this way AI can considerably lessen the work of reviewers.
AI in Peer Review
Peer review is the main filter that journals use to test the legitimacy and credibility of research work. The difficulty lies in choosing the right reviewers and also in having reviewers divide their time between their regular work and review. Besides this, reviewers are usually overworked, and some of their review decisions may be biased, which then requires further human intervention. AI can mitigate this by providing suggestions to reviewers based on their previous works, comments, and interactions. AI can also check identities, keywords, and conflicts of interest; predict impact factors; and manage correspondences, invitations, reminders, and queries. These tasks consume more time and energy when manually done.
Pros and Cons
When AI manages and selects the content, unnecessary tensions between authors, reviewers, and editors can be avoided. Human bias can also be removed to some extent. It can perform large-scale checks within seconds. Text and data mining can point out bad reporting and statistics and fabricated data. If AI is deployed in open access publishing, full automation of the workflow is possible. This will require readers to decide the relevance of the publication.
But AI is still not efficient at identifying and rejecting fake reviews and comments. Some manuscripts that are not up to journal standards may thus seem publishable. So AI should be studied to figure out the tasks it can perform autonomously.
Future of AI in Peer Review
The peer review of scientific research cannot entirely rely on AI. Human oversight is required to make the final decision on every task. What AI can do, however, is to sufficiently reduce the current bottlenecks in the publishing workflow. We are yet to witness the advantages of AI in publishing at its full potential. Someday peer review studies and advancing technology will develop new solutions and allow us to reap many benefits in this regard.
Amnet is an end-to-end publishing service provider whose technabled solutions have simplified the publishing workflows of many organizations. EnableOA (https://enableoa.amnet.com/) is an open-source publishing platform that provides a host of products, applications, data analytics, and editorial services under one roof.