This meeting’s reading was a blog titled AI’s cognitive implications: the decline of our thinking skills?. Key discussion points:
How AI affects critical thinking
- Overuse of AI can diminish existing critical thinking skills as users may rely on quick answers without doing their own independent analysis. Many AI tools are not designed to encourage learning, and the subtle integration of GenAI into lots of tools can lead students to trust information without knowing what it’s based on. Current AI models can be dangerous ‘black boxes’ because we don’t know what data they’ve been trained on, and some don’t provide sources. We tend to believe people if they are confident, and the same may be true with AI.
- However, AI can enhance critical thinking by explaining things in a way you understand, therefore making topics more accessible and allowing you to explore them further. You can ask the AI to show different sides of arguments and provide sources for you to analyse. You can also use AI to enhance critical thinking skills by using the PAIR (Problem, AI, Interaction, Reflection) Framework: define a challenge/problem, ask AI to solve it, and then critically evaluate the AI solution.
Cognitive offloading: shifting memory and problem-solving tasks to technology
- AI is still in its ‘new’ phase where some students are testing its limits and trying to use it to do everything. However, if a student overuses it and offloads everything to AI, they are no longer doing the learning.
- We discussed positive and negative examples of cognitive offloading such as over-relying on a colleague to answer questions for you, autocorrect reducing the need for remembering/being accurate with spelling, and using calculators. Cognitive offloading can be beneficial by doing repetitive work for you and allowing you to focus on more complex and creative tasks, however, there is still a need to learn and understand what’s going on behind the technology e.g. although Excel formulas make a task much quicker, you still need to understand them to be able to sense-check the output.
- AI can help with learning, especially with initial knowledge acquisition, and it can act as a personalised tutor. We may see the time it takes to master something can be reduced by using AI.
- If AI is making everyone more efficient through cognitive offloading, people will be given more work, including students.
- AI tools such as Grammarly are especially useful for students whose first language is not English. We discussed what happens when there are in-person assessments without these tools. It should be made clear to students how much of what is being assessed is their use of English vs their argument. However, being able to clearly state your thoughts is important as this needs to be able to be picked up by the marker. Formative work should help students feel confident that they’ll be able to do this in their summative work.
- If a student uses GenAI, the output can appear to them to be well written. However, if a student does not have a good understanding of the topic, they may not recognise if it does not have substance.
Suggestions for learning design
- Design assessments so that AI can’t just complete tasks for students and so that there are opportunities to demonstrate critical thinking. Examples include in-person presentations, debates, and assessments based on experience.
- Develop AI agents which are embedded within units. You provide the AI with a knowledge base and how you want it to communicate with the learner. This can allow personalised learning and it’s beneficial for accessibility and inclusion.
- When students start at university, it’s important to provide onboarding about AI so that they know how to critically assess information and sources. AI should be used as a critical friend and not for coming up with ideas.
- We want to teach students to write effective AI prompts that show different sides of arguments and provide sources. It could be useful to teach this in communication with people too.
