Ad maiorem Dei gloriam

Jesuit AI Ethics

A framework, a rubric, and a workshop · Catholic Social Teaching · the Rome Call · Ignatian discernment

Applications

The framework at work in higher education

What the framework looks like when applied to the four most common AI deployments on a campus. These are starting points for institutional review rather than exhaustive checklists, and the rubric's five questions apply to each.

i.Student-facing chatbots

Advising, admissions, financial aid

  • Human dignity. Accessible to all regardless of digital literacy. A student who struggles with the interface has the same claim on the institution as one who does not.
  • Subsidiarity. Clear, easy paths to human advisors. The bot is a bridge to a person, not a wall in front of one.
  • Transparency. The system identifies itself as AI and acknowledges its limitations rather than bluffing past them.
  • Cura personalis. The measure is the particular student the night before a deadline, not average response quality across sessions.

ii.Administrative AI

Enrollment prediction, retention risk scoring

  • Impartiality. Predictions are audited for bias. A risk score that encodes demographic disadvantage can compound the injustice it claims to detect.
  • Transparency. The institution can explain how scores are calculated, to itself and to the students they concern.
  • Human dignity. Students are not reducible to risk scores. Antiqua et Nova holds that “the uniqueness of the person” must not “be identified with a set of data” (¶94). A score can open a conversation. It should not close a door.

iii.Teaching tools

Writing assistants, tutoring systems

  • Reliability. Consistent performance across student populations. A tutor that works well only for well-prepared students can widen the gap it should close.
  • Inclusion. Accessible to students with disabilities, by design and by testing.
  • Human dignity. Tools support learning and human development. They do not replace the formation that education exists to provide.
  • Men and women for others. Arrupe's test asks whether the tool forms people for others or only makes individual output more efficient.

iv.Community-facing tools

Public services, outreach, neighborhood partnerships

  • Preferential option for the poor. Design for the underserved first and then generalize. Dilexi Te renews this priority.
  • Solidarity. Accountable to the community served, not only to the institution deploying it.
  • Security and privacy. Vulnerable users' data is protected with minimal collection and bounded purpose.