Why artificial intelligence is only as good as the human judgment guiding it.
There is a classic story about an engineer who is called to a factory because a vital machine has broken down. It’s costing the company millions. The engineer walks in, looks at the machine for a few minutes, pulls out a hammer, and gently taps one specific nail.
The machine roars back to life.
Later, he sends an invoice for $3,000. The factory owner is furious. “You just tapped a nail! Send me an itemized bill.”
The engineer sends the new bill:
Tapping with a hammer: $1
Knowing which nail to hit: $2,999
In our latest conversation, we realized that this is the perfect metaphor for Artificial Intelligence in education.
AI can generate infinite content, solve equations instantly, and write essays in seconds. It has the hammer. But it doesn’t always know which nail to hit. That is where the Human in the Loop becomes non-negotiable.
The “Iceberg” of Expertise
We often think of teaching or tutoring as simply “delivering information.” But that is just the tip of the iceberg.
Under the water, there are years of context, nuance, and “taste.”
AI can generate a study guide based on a textbook.
A Human knows that the student struggled specifically with graphing systems of equations last week and needs to focus there, not on the definitions they already mastered.
AI is powerful, but it lacks context. It doesn’t know that a math problem using “8.5” instead of “8” adds a layer of cognitive load that might distract a specific student from the core concept. It doesn’t know when to push and when to pull back.
As we discussed in the episode:
“That kind of judgment call is always going to be important until AI is perfect... We underestimate the power of human judgment.”
The Danger of the “Quick Fix”
There is a growing concern—backed by studies from institutions like MIT—that unbridled use of AI can reduce critical thinking. Why? Because it eliminates the struggle.
If a student sees a hard math problem and immediately asks AI for the solution, they bypass the neural rewriting that happens during “Think Time.”
We are exploring a new paradigm for our students: AI as a Context Builder, not a Solver.
Imagine a tool that, when a student is stuck, refuses to give the answer or the next step. Instead, it asks: Do you understand what these symbols mean? Do you know what a quadratic is?
It forces the student to sit in the discomfort of not knowing. That discomfort is where learning happens. If we take that away, we do our students a massive disservice.
Why We Are “Someta”
This brings us to the core of our philosophy and the name of our app: Someta.
We want students to move beyond just “doing the math” to engaging in Metacognition—thinking about how they think.
When a student interacts with AI, they shouldn’t just be passive consumers of answers. They need to be the architects of the conversation. They need to step “outside the flow” (the meta perspective) and ask:
Is this answer making sense?
Did I ask the right question?
How can I articulate what I don’t know?
We recently worked with a student who, after weeks of “productive struggle,” could finally articulate exactly where he was stuck: “I can solve systems with elimination, but I can’t find the intersection when graphing.”
That level of self-awareness is worth more than any single correct answer. That is the goal.
A Note to Parents
Finally, we want to address the parents reading this.
We know that when your child struggles or fails a test, it feels like your failure. You feel that pain viscerally. We see it, and we want to tell you: You are doing a great job.
Your child’s academic struggle is not a reflection of your parenting failure; it is a stepping stone in their development. Our goal—whether through our content or our tutoring—is to take that weight off your shoulders and help your child build the confidence to hit the right nail, all on their own.
Key Takeaways
Human in the Loop: AI provides the data; humans provide the judgment and context.
The Iceberg: Real expertise looks simple on the surface but is supported by deep experience.
Productive Struggle: Students must be allowed to “wrestle” with problems before AI intervenes.
Metacognition: The ultimate skill is learning how to think about your own thinking.
See you in the next one!
