Our aim is to develop human judgment, not to produce finished work. Instructors decide
whether and how AI figures in their courses, according to the standards of their discipline
and learning outcomes; prohibiting it is as legitimate a choice as requiring it. Students
are entitled to know clearly what each course expects. Where AI use is required, it
should be integral to the learning rather than incidental. Requiring a tool also means
ensuring it is provided, or its cost made known before students enroll; where a student
still cannot access it, the outcome is met through support rather than a lower bar.
Instructors should not ask students to put their own or others' personal work into
tools that do not protect it. Where it is relevant to a course, AI literacy is worth
teaching, including a clear-eyed sense of what these tools can and cannot do, alongside
the forms of practice and productive struggle a field depends on.
AI is a legitimate research instrument, and using it well is part of good scholarship.
As with any instrument, human authors remain accountable for the validity and integrity
of their work. AI is not an author; authorship carries responsibilities only a person
can discharge. We treat AI-assisted results as findings to check and characterize,
not to trust, and we check what AI gives us before it enters the record: a reference
we have not verified is not a citation. For human subjects research, we disclose proposed
uses of AI in research to the Institutional Review Board. We disclose material AI
use as the norms of our field, venue, and funder require, and we document our methods,
so others can follow and, where the tools allow, reproduce them. We use AI to accelerate
the work while keeping the interpretive judgment at the heart of scholarship. And
we honor the terms under which data and others' unpublished work are entrusted to
us: confidential manuscripts, proposals under review, and protected data do not go
into tools that would expose them.
Where we use AI in service and administration, we use it to reduce burden, not to
offload responsibility. We do not enter confidential personnel or student data into
tools not approved for it, and we follow the university's data-protection requirements.
In decisions that materially affect a person's education, employment, or standing,
including admission, hiring, the promotion and tenure process, and aid, a responsible
person makes the decision and can be held accountable for it. If AI tools are used
to assist such decisions, its effect is examined for bias before deployment and reviewed
again as the tool or the use changes. We are transparent, in a manner appropriate
to the context, when AI materially shapes an official decision or communication. And
much of this work is carried out by staff, whose jobs these tools can change; their
perspective belongs in the decision to adopt them.
Some programs prepare students for professions (for example, teaching, nursing, social
work) that carry duties of care toward the people they will serve, including minors,
and standards set by external accreditors and licensure boards. In these programs,
using AI well includes learning the legal and ethical obligations of the field, protecting
the data of the people students serve, and modeling the practice students will carry
into it. Students on placement are also bound by the rules of the site that hosts
them, which may differ from Towson's; programs should prepare students to ask what
applies rather than assume.