Should AI have a role in assessing research quality?