Generative AI Critical Thinking Scale Developed at Sapporo Meeting
Researchers unveiled a preliminary scale for measuring critical thinking attitudes during generative AI use on the second day of the Japanese Society for Educational Technology conference in Sapporo in late September 2026, according to a report on note.com by Hideto Nachi. The presentation, titled “Preliminary Study on the Development of a Critical Thinking Attitude Scale in Generative AI Use” by Hideto Nachi, Ken Miyauchi, and Takashi Kusumi, drew attendees from diverse academic and professional backgrounds to discuss the psychological mechanics of interacting with artificial intelligence.
Cognitive Offloading and the Challenge of Intentional Thought
The research addresses a fundamental divergence in modern education: the gap between knowing how to use AI and choosing to apply critical evaluation. While training programs instruct users in prompt engineering and fact-checking, people frequently abandon these practices under time pressure or fatigue, much like drivers who stop checking blind spots on familiar routes. Because generative tools lower the cognitive cost of avoiding deep thought—a process known as cognitive offloading—researchers emphasize that possessing the technical capability to evaluate output differs entirely from having the disposition to actually do so.
Six Core Dimensions of AI-Related Critical Thinking
To capture this mental posture, the authors moved beyond abstract definitions by analyzing the actual phrasing used by 74 university students who regularly use generative AI. Through collaborative coding, the researchers categorized user reflections across four distinct stages of AI interaction into six specific dimensions:
- Information Source and Veracity Evaluation: Checking the origin of data and questioning its factual accuracy.
- Active Thinking Maintenance: Resisting over-reliance on automated answers and continuously applying personal critical judgment.
- Precision Prompt Engineering: Refining instructions with specific context and clear conditions.
- Premise Clarification and Goal Alignment: Defining the underlying purpose of AI use and ensuring the process stays on track.
- Multiple Perspective Exploration: Anticipating alternative viewpoints and seeking out competing answers.
- Responsibility Reflection: Assessing whether input data is appropriate to share and accepting accountability for the final output.
The authors note that these categories remain part of a preliminary framework and will likely evolve as the research progresses through further empirical testing.
Who Authored the Study and What Does It Measure?
Who authored the presentation in Sapporo?
The study was authored by Hideto Nachi, Ken Miyauchi, and Takashi Kusumi, and was presented at the Japanese Society for Educational Technology conference.
How many students participated in the preliminary study?
Seventy-four university students who regularly use generative AI provided written reflections detailing what they monitor during four distinct stages of interaction.
What does the newly developed scale measure?
The scale measures the critical thinking attitude or disposition of users when they interact with generative AI, distinguishing active mental engagement from passive reliance.
Worth a look