👾 Teaching in the Age of AI Personalization

The following is a guest post by Forward Future community member Mandy McLean, co-founder and CEO of ClassWaves.

I often think back to a quiet September morning in my second year of teaching. Notebook in hand, I slipped into a colleague’s high-school physics class, eager to learn from someone who made the hard parts look easy. At the front of the room, she’d lined up a dozen glass bottles, each filled to a different height with water. When the chatter faded, she blew a single note across one bottle and asked, “Which of these will have the highest pitch, and why?”

A student raised a hand, “The one with the least water because there’s more air to vibrate.”
Another shook her head, “I think it’s the one with the most water but I’m not really sure why.”
Someone across the room asked, “Is it like a flute? When more holes are covered, the sound is lower so would that mean that less air makes a higher pitch?” 

The teacher let the debate build, then challenged them to tune a perfect octave by pouring water from one bottle to another. Thirty or so minutes later, the room was alive with hand-sketched wave diagrams and arguments about why pipe organs boom while soda bottles whistle. 

Years later, while observing preservice science teachers for my PhD, I watched novices try to spark the same energy with props like paper clips and beakers. When they defaulted to explanation, eyes glazed over. But when they framed a puzzle, asked open questions, split students into groups, and let the silence breathe, even simple prompts unlocked talk about density, surface tension, and real scientific argument. The difference was never the curriculum, it was always the conversation. 

The Seduction of Perfect Fit

Large language models promise lessons tailored to every learner’s exact reading level, pace, and interests. In theory, this is a gift. In practice, hyper-personalized feeds can undercut the very habits we hope to grow. When an algorithm’s north star is minutes on task, it tends to deliver smaller, faster hits of content, like an educational cousin of infinite scroll. Micro-tasks keep the streak alive but leave little room for the messy give-and-take that deep understanding requires.

The stakes are even higher outside physics. In history, civics, or health, facts mingle with interpretation, and echo chambers can turn small misunderstandings into parallel realities. We’ve seen this happen on social media. Algorithms push what feels relevant to each person, but in doing so, they break apart shared understanding. Conversation is what keeps ideas honest and without it, even well-meaning AI can steer learners into narrow lanes of thought.

As AI makes facts cheap, it raises the value of judgment, empathy, and the ability to work across differences. Those skills do not grow in silence or emerge from auto-leveled worksheets, they take shape in the space between people—when a student pauses after hearing a new idea, wrestles with a challenge (often referred to as productive struggle), or revises thinking because someone saw it differently. Real learning is social before it’s personal. You can feel it in a classroom where students build on one another’s thoughts, circle back, and refine. Often the most important move a teacher makes is to stay quiet a moment longer.

Technology That Amplifies Voices

That conviction drives ClassWaves, the company I co-founded after a decade of studying how students learn. Instead of pushing each learner a private feed, ClassWaves listens to small-group conversations (no audio is stored) and surfaces patterns for the teacher: Who is engaged? Who is holding back? Are students moving from opinions to reasoning, or getting stuck? With those signals in real time, a teacher can step in to refocus a group, draw out a quiet voice, or prompt deeper thinking. The teacher stays in charge; the AI simply makes the invisible visible.

Measuring What Matters

Metrics shape behavior. If we reward streaks, platforms will optimize streaks. What if instead we tracked how often a student referenced a peer before disagreeing, or the number of reflections that began with “I used to think… now I think…”, or the true ratio of student-to-student talk? None of these metrics is perfect, but each one gets us closer to what real learning looks like: shared thinking made visible. 

A Call to Builders and Backers

1. Design with conversation at the center. Student talk reveals how ideas form, shift, and deepen, in ways that clickstreams can't capture.

2. Take a hard look at your engagement loops. Ask whether your prompts foster real thinking or just keep students busy.

3. Trust teachers with choices. Support professional judgment rather than nudging everyone down a single path.

4. Invest in the human layer. Curiosity, creativity, collaboration, and care cannot be automated.

The Story We Choose

Personalization supports learning, but without intention, it can also distort it. Similarly, AI can free teachers from one-size-fits-all instruction, or it can quietly fracture classrooms into a thousand silent lanes. The difference lies in our metrics, our design instincts, and our defense of the quiet pause that lets learners think aloud together. 

When those bottles finally rang a perfect octave, the class cheered. Not for the right answer, but for the shared thrill of figuring it out together. Our task now is to build AI that multiplies those moments, not mutes them.