In brief what will be the reply on the following response: "in human-centred processes, where should we draw the line when it comes to automation? A risk-tiered approach is a workable solution. Complete automation with monitoring is appropriate for low-stakes, reversible decisions (like prioritizing support cases) as long as users are aware of the process and can readily fix mistakes. Automation should function in decision-support mode for medium-stakes decisions (such as screening resumes to send applicants to recruiters): the system makes a suggestion, a qualified person discards it, and the company verifies the results for inconsistent outcomes. Automation should support, not decide, high-stakes decisions like hiring, promotion, termination, and the majority of loans. In these situations, organizations should: (a) maintain a human in the loop with the power and responsibility to override; (b) explain the primary factors influencing the recommendation in a way that is appropriate for the audience; (c) provide contestability and prompt review; and (d) continuously monitor for subgroup disparities and performance drift (Tabassi, 2023)