The COVID-19 pandemic not only accelerated digitalization and connectivity around the world, for the publishing industry it also shone a spotlight on how research is evaluated and disseminated as a public good. At Frontiers our Open Science Publishing Platform was certainly put to the test with thousands of coronavirus-related submissions flooding in from early 2020 onwards which required rapid but thorough peer review, while many of the researchers best suited to evaluating them were working on the frontline. Technology made this possible and we argue that it is time for scientific publishing to embrace what has traditionally been considered a paradox: quality at scale.<br/><br/>Frontiers’ platform is developed in house and continuously evolves with feedback to optimally support authors, reviewers, and editors. Artificial Intelligence (AI) can provide unique tools to our users to ensure both the quality submissions and the peer review process while the volume of articles grows. Our AI module, called AIRA (Artificial Intelligence Review Assistant), has become a cornerstone of our platform and serves two functions; first, to enable numerous quality checks, such as language rating, image manipulation detection, and conflicts of interest, upon submission of a manuscript, second, to find the best matching reviewers for a manuscript. The real benefits of AI can however only be unleashed when there is an efficient interface between humans and machine. This means that AI is there to enhance and facilitate human decision making rather than replace it. AIRA achieves this by flagging quality concerns and by suggesting reviewers and allowing users to take the action required. Users are directed to the tasks that need their input and AIRA continues to learn with the feedback it receives when users interact with it. For example, the reviewer suggestions are not just provided as a list of names, but AIRA offers an interface where the handling editor can adjust the keywords used and apply filters to refine the recommendations. Similarly, for quality checks, visualising for example the patterns found in images that suggest manipulation aids the user to swiftly respond to a concern.<br/><br/>AIRA is a showcase for how publishing platforms are evolving to publish the ever-increasing output of research advances efficiently and rigorously. And although machine-learning algorithms are powerful tools, the value of this technology can only be realised when it works together with humans.

Machine and human together: How AI can empower editors and publishers to achieve quality at scale

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.