PURPOSE
Generative artificial intelligence (genAI) has facilitated the rapid
creation of teaching materials for medicine as well as other disciplines.
Unfortunately, this boost to productivity has the side effect of lowering the
effort barrier to creating content. This has resulted in the creation of
misleading, low-quality, or ineffective educational content without due care
on the part of the creator. AI-generated material produced without care has
been termed “slop” and is becoming endemic on social media and sharing
platforms. Drawing on our recent analysis of slop in online educational
video, we propose a design and use framework for AI-generated teaching
materials so that diligent instructors can use AI to produce educational
content without the risk of spreading slop.
 
METHODS
The Care Lifecycle for Educational AI with No Slop (CLEANS) protocol spells
out distinct stages in the creation and use of AI-produced teaching materials
and specifically identifies checkpoints in the process at which creators can
demonstrate care (direct, expertise-based action and responsibility for the
outcome). It also establishes a prescribed schedule for evaluating the
effectiveness of the materials and revising or retiring them as needed. The
framework is divided into “development” and “curation” stages and spans
content creation from planning to ultimate revamping or retirement.
 
RESULTS
As proof of concept, we demonstrate the CLEANS framework using two
genAI-created study aids for first-year preclinical sciences courses, one
short video and one graphic diagram. The work is ongoing but early results
show that proper design and implementation of AI-created teaching materials
requires considerable effort.
 
CONCLUSION
The CLEANS framework provides a rigorous yet flexible organizational tool
for assisting instructors in creating AI-assisted teaching materials without
fear of creating and propagating slop.