Methodological Variation in Studying Staff and Student Perceptions of AI
How we measure AI sentiment is fundamentally flawed, new research shows.
Deep Dive
A new study analyzing staff and student perceptions of AI reveals that different research methodologies can produce contradictory results from the same data. Researchers compared sentiment analysis, thematic analysis, and word clouds from comments and focus groups. They found that a simple positive/negative sentiment score paints a completely different picture than a detailed content-based thematic analysis. This 29-page paper questions the reliability of countless previous studies on AI attitudes in institutional settings.
Why It Matters
This undermines the foundation of many policies and reports shaping how AI is integrated into education and workplaces.