Critical Thinking – Center for Teaching and Learning /ctl Wed, 25 Mar 2026 20:25:40 +0000 en-US hourly 1 /ctl/wp-content/uploads/sites/88/2024/01/cropped-android-chrome-512x512-1-32x32.png Critical Thinking – Center for Teaching and Learning /ctl 32 32 Highly-Cited “AI Erodes Critical Thinking” Study Appears To Be AI Generated Slop /ctl/highly-cited-ai-erodes-critical-thinking-study-appears-to-be-ai-generated-slop/ Wed, 25 Feb 2026 22:10:54 +0000 /ctl/?p=5521 This week, I want to highlight a substack post critiquing a recent research article. The author critically examines a widely referenced paper claiming that increased AI use degrades critical thinking skills. Pookins argues that the study’s design and methodology are fundamentally flawed: the sample isn’t representative, the survey measures self-reported beliefs rather than actual critical thinking performance, and many items intended to measure different constructs are essentially redundant. Because of these flaws, he asserts that the paper does not provide reliable evidence that AI use causes a decline in critical thinking, meaning that its frequent citation in media and academic discussions may be misleading or premature. Moreover, he points to evidence that the paper itself may have been AI-generated.

]]> This week, I want to highlight a substack post critiquing a recent research article. The author critically examines a widely referenced paper claiming that increased AI use degrades critical thinking skills. Pookins argues that the study’s design and methodology are fundamentally flawed: the sample isn’t representative, the survey measures self-reported beliefs rather than actual critical thinking performance, and many items intended to measure different constructs are essentially redundant. Because of these flaws, he asserts that the paper does not provide reliable evidence that AI use causes a decline in critical thinking, meaning that its frequent citation in media and academic discussions may be misleading or premature. Moreover, he points to evidence that the paper itself may have been AI-generated.

Why is this such an important post to read? The key issue isn’t whether AI might affect cognition. That is a broader and ongoing research question supported by diverse studies on cognitive offloading and educational impacts. Instead, how we interpret and communicate evidence is what is critical here. The post highlights the importance of scrutinizing research methodology before adopting headlines about AI’s harms or benefits. In teaching and policy conversations, this means encouraging nuanced engagement with research on AI and critical thinking, distinguishing between correlation and causation, and integrating AI in ways that support, rather than inadvertently replace, deep learning and reasoning.

If you have found a quality research article on the impact of AI on learning, please share it with us by emailing it to umpi-ctl@maine.edu.

Read the full post here:

Pookins, N. (2026, February 15). Highly-Cited “AI Erodes Critical Thinking” study appears to be AI generated slop. Nebu’s Newsletter. Substack. 

]]> AI-Supported Civil Discourse /ctl/ai-supported-civil-discourse/ Wed, 11 Feb 2026 17:53:38 +0000 /ctl/?p=5096 Sway is a chat platform that connects students with differing perspectives into one-on-one conversations and facilitates better discussions between them. Developed with Heterodox Academy and inspired by John Stuart Mill's radical view that engaging with opposing perspectives is an essential tool for improving reasoning and solving complex problems, Sway aims to create online spaces where we can all learn to discuss controversial issues more openly and constructively.

]]> Sway is a chat platform that connects students with differing perspectives into one-on-one conversations and facilitates better discussions between them. Developed with Heterodox Academy and inspired by John Stuart Mill’s radical view that engaging with opposing perspectives is an essential tool for improving reasoning and solving complex problems, Sway aims to create online spaces where we can all learn to discuss controversial issues more openly and constructively.

Create a free account to try it out. When participants enter a question or response for their chat partner on a challenging topic, Sway will evaluate the message for inflammatory or unproductive content and provide a suggested way to rephrase their post. The Sway Guide will also interject comments and prompts throughout the discussion as a facilitator and provide factual information to learners related to the content when asked.

To see it in action and hear from the developers, register for the February 19 Beyond Bots session listed in the events below.

Check it out here: 

]]> Wrong answers, right learning: Using errors to deepen understanding /ctl/wrong-answers-right-learning-using-errors-to-deepen-understanding/ Wed, 12 Nov 2025 22:06:55 +0000 /ctl/?p=5359 This systematic review examines how instructional materials that embed errors (so-called “erroneous examples”) or juxtapose incorrect and correct solutions (“contrasting erroneous examples”) can influence student learning across a variety of domains (mathematics, medicine, science). The authors reviewed 40 studies and found that these approaches can enhance learning — especially by helping students grasp both what not to do (negative knowledge) and what to do (positive knowledge) — but the benefits depend strongly on how the errors are used, what scaffolding (prompts, feedback) is provided, how complex the task is, and how much prior knowledge the learner has.

]]> This systematic review examines how instructional materials that embed errors (so-called “erroneous examples”) or juxtapose incorrect and correct solutions (“contrasting erroneous examples”) can influence student learning across a variety of domains (mathematics, medicine, science). The authors reviewed 40 studies and found that these approaches can enhance learning — especially by helping students grasp both what not to do (negative knowledge) and what to do (positive knowledge) — but the benefits depend strongly on how the errors are used, what scaffolding (prompts, feedback) is provided, how complex the task is, and how much prior knowledge the learner has.

91ɫ this in your courses:

  • Use error-embedded examples when you want students to recognise common misconceptions or faulty procedures.
  • Pair them with strong prompts (e.g., “Why is this step wrong?” “How would you fix it?”) and/or feedback explaining why the error occured.
  • Be careful with complexity: ensure the task is appropriate for students’ level, and avoid overwhelming novices.
  • Consider alternating or fading into error-analysis as students gain expertise.
  • Monitor cognitive load.

Read the full article online:

Dieterich, S., Rumann, S., & Rodemer, M. (2025). Conditions for effective learning from erroneous examples: A systematic review. Educational Psychology Review, 37(4), 94. 

]]>