Month: June 2025

Majority of assessment must return to the classroom

In the age of AI tools like ChatGPT, a provocative argument has emerged: If it can be ChatGPTed, it shouldn’t be taught or tested. However, this perspective is flawed. While students can now Google or use ChatGPT to access much of the foundational knowledge we once memorized, it doesn’t mean they shouldn’t learn it. Foundational knowledge remains essential—it serves as the scaffolding that supports deeper understanding, critical thinking, and problem-solving.

In my Numerical Methods course, I take a deliberate approach. Students are only allowed to use the TI-30Xa calculator—a basic, non-programmable tool. This restriction is because they need to understand the fundamental algorithms behind the math. I encourage them to use the same calculator at home for homework and studying. Sure, they can find homework answers online, but homework is only 15% of the grade. The rest is structured to promote authentic learning: 10% comes from projects that are too specific to be easily ChatGPTed, and 75% is based on in-class assessments. To make this work, we need to increase classroom contact time and design more ChatGPT-less activities—those that require students to think, apply, and engage without relying on AI shortcuts.

In my Mechanics of Composites course, I use a similar grading structure. However, I allow any calculator approved for the FE exam. I’m no longer interested in testing whether students can integrate by hand. Instead, I focus on whether they understand the concepts behind composite materials. Again, 80% of the grade is based on in-class tests.

We don’t hand a first grader a calculator to add two numbers. Why? They need to understand what the concept of addition is before they can use a tool to perform it more efficiently. The same principle applies in college: Use the right tools at the right time, not the whole kitchen sink. AI is here to stay. But instead of fearing it, we need to rethink how we teach and assess. This rethinking begins by bringing assessment back into the classroom, where genuine learning takes place.

This post is brought to you by

Takeaways from the Paper – the The Memory Paradox: Why Our Brains Need Knowledge in an Age of AI

In a 50-page paper, “The Memory Paradox: Why Our Brains Need Knowledge in an Age of AI“, the authors examine the paradox related to human memory during the advancement of artificial intelligence. While AI grants significant access to information, it may also impact cognitive abilities essential for in-depth analysis and learning. The authors suggest that heavy dependence on digital tools can impact internal memory systems and lead to a surface-level understanding of knowledge. They propose a balanced model that merges technological tools with traditional educational methods to promote critical thinking and memory retention. This summary outlines the key points from the paper.

The role of human memory is examined in the context of artificial intelligence’s capacity to recall facts, solve problems, and generate content. The paper argues that while digital tools offer substantial information availability, they may also alter cognitive skills essential for reasoning and practical learning.

A key topic is cognitive offloading, that is, the use of external tools to reduce mental effort. While this strategy can be helpful for complex tasks or minor details, the authors note that over-reliance on such tools can influence the way individuals store and access information. Instead of forming comprehensive mental models, people may remember where information is located rather than its content, resulting in what is described as an “illusion of knowledge.” This phenomenon can lead to the perception of being informed without a thorough understanding of the underlying concepts.

Neuroscientific research is referenced to explain how prediction errors, discrepancies between expectations and outcomes, are essential for establishing memory traces and enhancing neural connections. The lack of initial internalization of knowledge hinders the ability to make predictions, which are crucial to learning, thereby impacting the development of comprehension and intuition.

The paper discusses educational trends, noting that some schools have shifted away from memorization towards teaching methods centered on critical thinking and discovery learning. The increased reliance on external resources coincides with observed declines in IQ scores in various regions, a reversal of the earlier Flynn Effect. The authors associate these patterns with educational practices that place less emphasis on memory and more on external aids.

With the growth of generative AI tools such as ChatGPT, these concerns are considered more pertinent. Studies cited in the paper indicate that students who depend heavily on such tools may spend less time reflecting, self-correcting, and retaining new information. This dependence has been referred to as “metacognitive laziness,” implying a tendency to avoid the mental effort required for lasting learning outcomes.

The suggested approach is to integrate technology thoughtfully, using AI to enhance, not replace, cognitive processes. The paper emphasizes the importance of retaining fundamental knowledge for critical thinking, error identification, and efficient acquisition of new information. A balanced strategy, including desirable difficulty, retrieval practice, and internalizing key knowledge, is recommended. It is noted that students benefit from challenges set at an appropriate level to stimulate engagement without causing undue frustration.

In conclusion, the memory paradox raises questions about learning approaches in the context of widespread digital technology. The paper concludes that although technology expands capabilities, it does not eliminate the need for mental effort in understanding material. Retaining knowledge internally remains crucial, as the ability to remember, reason, and reflect continues to play a vital role in navigating complex and information-rich environments.

This post is brought to you by