AI evidence extraction

Implementing AI Course Assistants: A Rapid Design Case from Concept to Full Rollout at Los Angeles Pacific University

George Hanshaw, Mike Wilday · International Journal of Designs for Learning · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

5/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.14434/ijdl.v16i2.41926

Methodology & findings

Study design

Design case study with rapid implementation across an online institution; included pilot testing, iterative refinement based on faculty and student feedback

Main result

In early pilots, Spark "acted as a 'thinking partner,' supporting student engagement, critical thinking, and motivation." The study demonstrated that "the accelerated rollout required standardized prompts, curated course content, and scalable workflows, while faculty and student feedback informed iterative refinements."

Reports effect sizes.

Research paradigm

pragmatist/design-oriented

Author conclusions

Los Angeles Pacific University (LAPU), a fully online, student-centered institution, implemented Spark, an AI course assistant, across all courses within four weeks. Guided by Gabriel's Behavioral Engineering Model and an agile design mindset, the Digital Learning Solutions team made intentional design choices, including shaping Spark's interactions through the Socratic method, involving faculty as co-designers, and constraining knowledge access to maintain academic integrity.

Risk of bias

Selection bias: Early pilot participants may not be representative of all students; No control group for comparison; Lack of quantitative outcome measures reported; Potential Hawthorne effect from early adoption phase; No formal statistical analysis reported; Selection bias: participants were from a single institution (LAPU); Self-selection bias: early pilots likely included motivated faculty and students; Lack of control group or comparative analysis; No quantitative outcome measures reported; Potential confirmation bias in feedback collection; Selection bias in early pilot participants; Lack of control group for comparison; Self-selection of faculty as co-designers; Potential response bias in feedback collection

Data: not_statedCode: not_statedExtracted from: pdfAgreement 80%

Explore related topics

Responsible AI Use Human-AI Collaboration Case Study