Learning is Hard - and it's supposed to be - AI Literacy for Students

Learning is Hard - and it's supposed to be

Wouldn't it be fantastic if we could achieve mastery of our subject without effort - like Neo in the Matrix plugging the computer into his brain to instantly learn kung fu. The truth is, the science of learning shows that struggle is the engine of growth. Lasting learning requires an effortful, cognitive process to build the neural pathways in your brain to build deep understanding. Generative AI is a powerful tool that can support that process, but if you use it to bypass the "heavy lifting", you are undermining actual skill acquisition and knowledge retention.

Worth Watching

The inspiration for this page: What Everyone Gets Wrong About AI and Learning

Chapters

What's Next?

Science of effort ➔ forklifting vs weight-lifting ➔ guidelines for productive AI use.

The Science of "Hard"

Why Your Brain Needs the Strain to Get the Gain

You know how a workout hurts before it helps? Learning runs on the same principle. Each jolt of mental strain (remembering a fact, untangling a proof, re-testing yourself) signals your brain to lay fresh neural wiring. Skip the “burn” and no new circuits stick. Cognitive science shows that the very effort we try to avoid is the price we pay for automatic skill.

Think of your brain as a living switchboard. When one neuron keeps stimulating another, the connection between them strengthens in a process that can be summed up as “neurons that fire together, wire together.” (aka, neural plasticity). When learning, that wiring happens best under strain which psychologists call desirable difficulty. The strongest trio of desirable difficulties is:

Retrieval Practice (Testing Effect) - This is the heavyweight champion of the three. Each time you try to pull up a fact from your memory, solve a chemistry equation, or explain a theorem in your own words, you’re deepening the connections. Learn More
Spaced Repetition - Reviewing material at widening time-gaps trains the brain to hold onto it for the long haul. Learn More
Levels-of-Processing Principle: The deeper you think about meaning and connections, the stronger and longer-lasting the memory trace. Learn More

Forklifting vs Weightlifting

Would you let a forklift do your bench presses for you?

It would be absurd to drive a forklift up to the squat rack so that it can lift the weight for you. You'd leave the gym exactly as weak as you arrived. Now flip the scene: picture a worker carrying a pallet on his/her back across the loading bay instead of using a forklift. Both images are ridiculous. Similarly, it is ridiculous letting AI handle all your thinking (when the whole point of learning is to build your thinking muscle) it's also silly forcing yourself to grunt through thoughtless drudgery that a bot could do better.

Illustration contrasting forklifting for heavy lifting tasks vs. weightlifting for building strength.

Key Tips

Match the tool to the goal. If the task would build strength (analysis, drafting, problem-solving), grab the bar yourself.
Off-load grunt work. If the task is requires time, but no thought or effort (formatting citations, tidying a table, re-writing messy notes) let the AI forklift roll in.
Lift first, then automate: Even when AI use is permitted, put your own brain power to the task before you turn to the AI.
Documentation: Note every time AI assists with the lift so you and our instructor can see the dividing line.

On your next assignment, before you get down to work, identify every "weight-lift" step and every "forklift step" while making sure you follow the AI use policies for your class. If you catch a forklift in the squat rack, or catch yourself hauling pallets on your back, rethink your approach.

Human-Led, AI-Boosted

A Playbook for Productive Struggle

The table below outlines a general playbook for combining your own thinking with responsible AI assistance. Adapt the steps to suit your specific task - essay, lab report, slide deck, or problem set. Make sure you keep one principle constant: handle the core reasoning and final decision making yourself; let AI handle only the routine support that is permitted...and record where that boundary is.

Best-in-Class Approach to Balance Forklifting and Weightlifting

Stage	What you do	Why it matters	Where AI fits (never leads)
Figure out what to do	Read and understand assignment and (if supplied) the rubric	A clear target = less wandering & fewer restarts	After you've read and understood the assignment, ask AI to summarise the assignment; compare for gaps.
Understand the Policies	Read and understand course, department, and assignment AI policies. Write down the “no-go” items so you don't forget	Prevents accidental integrity violations.	None—policy comprehension is a human lift.
Weight-Lift / Fork-Lift Split	Make a two-column list. Weight-lifting (required learning) = idea generation, calculations, outlining, reasoning, debugging, drafting. Fork-lifting (support/enhance) = formatting, citation styling, converting raw data, quick background facts. Cross-check your list against the AI Policy; move anything forbidden back to the weightlift column.	Forces you to spot where true learning lives and keeps AI in the right lane.	After you create your initial list, upload the assignment and policy and prompt AI: “Suggest 3 routine support tasks for this assignment that would not violate policy.”
Do the Work (Iterate)	A. Start with a lift task you identified. B. When you hit a routine support step, pause and deploy AI (e.g., format citations, tidy a table). C. Return immediately to the next lift task. D. You'll generally want to rewrite or tweak AI output. At the very least read and check it.	Interleaving keeps momentum and shows where AI genuinely saves time without stealing reps.	Limited to “support” tasks you flagged, plus quick clarifying Q&A (“What’s the APA rule for…?”).
Accuracy & Voice Check	Verify every AI-touched sentence, number, or graphic against high-quality sources (course texts, peer-reviewed articles, instructor slides). Read aloud for tone drift; your voice should still sound like you.	Catches hallucinations and keeps things in your voice.	Optional: “Highlight any claim without a citation”; you confirm or delete.
Reflect & Log	At minimimum (your assignment may require more) log two quick bullets: Biggest lift insight (“I finally connected X to Y”). Most useful AI assist (“It reformatted 25 citations”).	Reflection converts experience into transferable skill.	Can format your log, but the insights come from you.

This might all seem really formulaic and probably doesn't fit every scenario. Use it as a guide to maximize your learning and minimize your chance of an academic integrity violation. It is helpful to be really deliberate about your approach at first and then as you get more comfortable balancing your own work with AI support you will be able to use your intuition more and more (oohhh, this is its own kind of learning).

You can find more subject specific examples HERE - (add link when section is complete)

Conclusion

Struggle isn't a flaw in the learning process; it's the mechanism that turns information into a durable skill. Each time you spell out an argument, solve a problem unaided, or rewrite a rough paragraph, you’re laying the neural routes that make future work faster and better. Generative AI can accelerate everything around that core effort, things like formatting tables, checking citations, generating quick background facts, but it can’t replace the reps your brain needs.

Use the playbook: understand the learning activity, follow the AI policy, separate human tasks from AI tasks, interleave your own thinking with carefully limited AI assists, verify the results, and log what you learned. Follow it and you’ll finish each assignment not just with a grade, but with stronger reasoning habits that no algorithm can provide.