Basically not much in terms of time, but a lot in terms of expertise. Basically, when he accepts the revision of an article he must verify that our editor has not written nonsense. It is not a scientific review, which would be very expensive, it is more a generic verification of a task. You don’t have to correct it, ever, you just have to express your opinion to the editor and show him where he needs to pay more attention, where the article is lame. (If he corrected it, the training part would be lost.)
In reality, these are often articles taken from theses, therefore already verified by the supervisors, or from articles from reliable sources, or from books of a certain diffusion. If the expert then “uploads” his own articles or tells us where to find them, it is clear that the review becomes extremely easy. It should be kept in mind that by default, the names of the experts are confidential, they do not appear in the article on the APP, nor to the editors.
It is the expert, if he wishes, who communicates it to the editor or requests publication in the article. Basically the software keeps the names confidential. Obviously, if the editor derives his article from a writing by an expert, the revision will be assigned to that expert. In fact, if we manage to have a good number of experts, the review should not take up more than 1 to 2 hours per month. Nothing nagging.