1.4. Learning in Minds and Machines
Hou I (Esther) Lau
This chapter turns to the idea of learning itself—what it means for humans, what it means for machines, and why the comparison matters. By looking closely at both, we can better understand what makes human learning distinctive and where AI’s version of “learning” begins and ends.
When Learning Clicks
Think back to a moment when something finally clicked for you: grasping a difficult theory, connecting with a book in a profound way, or solving a challenging problem. What made that moment stick? Was it how someone explained it? A connection to your life? An emotional experience that pulled it all together? These moments remind us that learning is not just about absorbing information. It is transformative; it reshapes us.
Human Learning
Learning is not only cognitive. It is also relational, embodied, and emotional. To truly understand human learning, we need to ask: What causes us to change? What helps us adapt over time?
Three elements are essential:
- Behavioral change: learning shifts our capacity to act and respond.
- Endurance: real learning lasts; it is not fleeting.
- Experience: learning is grounded in practice and lived experience.
Learning is therefore not just something we do. It shapes who we are. The capacity to learn, adapt, and transform is an irreplaceable human trait, even in an age of advanced AI.
How Does Learning Occur?
At the cellular level, learning physically changes the brain. Neurons—the cells that communicate information—restructure their connections. Sometimes existing pathways are strengthened; other times new ones are built. This process, called neuroplasticity, increases the efficiency of communication in the brain. Think of it like carving a trail through a forest: the more you walk it, the clearer and faster the path becomes.
But learning does not end with neurons. We are not only wiring circuits; we are making meaning. Human learning integrates experiences into identity, action, and relationships. It is how we become who we are.
What Is AI Learning?
When we say an AI model is “learning,” we do not mean it is reflecting, growing, or making meaning. AI learning refers to how an algorithm improves performance on a task by finding statistical patterns in data. It is mathematics, not memory, culture, or emotion.
Three major approaches illustrate how AI systems “learn” from data:
- Supervised Learning: The AI is given input data paired with the correct output, like a giant answer key. Over time, it maps patterns between inputs and outputs so it can make predictions with new data. Example: spam filters trained on thousands of labeled emails. The system does not “understand” spam; it learns statistical correlations.
- Unsupervised Learning: The AI receives data without labels and looks for patterns, clusters, or associations. It is like sorting puzzle pieces without a picture of the completed puzzle. Analogy: walking into a family therapy session with no background information and gradually noticing patterns of interaction.
- Reinforcement Learning: The AI takes actions, receives rewards or penalties, and adjusts its behavior to maximize rewards. Analogy: teaching a dog tricks, or a video game character learning to avoid traps and collect coins. In human terms, it resembles a therapist-in-training experimenting with different interventions and adjusting based on feedback.
📖 Analogy: Maps and Journeys
📚 Weekly Reflection Journal
Think of a specific skill you learned—such as writing a paper, cooking a meal, or driving a car. How did you actually learn it, and how would a machine “learn” something comparable? Write down a few notes comparing the two processes.
Quick Self-Check
Test your understanding of the differences between human and machine learning:
Looking Ahead
Having compared learning in humans and machines, we next examine the core mechanisms of AI: how training, data, and optimization processes allow models to improve. These mechanics will clarify both what today’s systems can and cannot do.
