Dive Brief:
- It may soon be possible for robots to conduct peer review for academic papers, including the ability to detect plagiarism, finding instances of misused data and noting when statistical tests have been used incorrectly, according to Insider Higher Ed.
- StatReviewer President Chadwick DeVoss conducted a study regarding the possibility of AI use in peer reviewing, saying that systems that could verify an author’s identity and predict a paper’s potential impact were in design and could be possible at some point.
- However, DeVoss advised that institutions should be wary of making the peer review process fully automated. He noted that using AI to determine whether a paper should be published was “where a slippery slope gets extra slippery.”
Dive Insight:
The concern over encroaching automation into peer review is the inadvertent damage it could do to an academic community that is in consultation with each other on parallel pursuits. Peer review can engender dialogue and discussion among like-minded researchers spanning across multiple institutions, and the pursuit of knowledge in their field can benefit from that collaborative (and competitive) relationship. Could artificial intelligence truly accommodate the intangibles of a fellow researcher trying to poke holes in your work?
A recent study of postsecondary educators found a wide divergence between the benefits and drawbacks of in-person professors versus an online experience; there was an unknowable aspect of the in-person experience that could drastically change how a student responded, in contrast to an online class. The same may be true for peer review. Bringing automation into the peer review process could be a “slippery slope,” but what may be lost would be the intangible analysis a peer offers that would be hard to quantify; this may make the practice more difficult to defend in light of the savings AI would likely offer schools.