Like trying to ban phones or stop SparkNotes use, trying to prevent AI use is a losing game. DeepSeek’s breakthrough shows why – AI is becoming more accessible and harder to detect. But what if that’s not actually our biggest problem?

If you’ve been following the conversation around AI and student cheating, DeepSeek is about to make things even more complicated.

DeepSeek just upended AI. Instead of brute force computing like OpenAI’s GPT-4, it used a modular approach that made AI more efficient, adaptable, and, crucially, harder to detect. That last part is why teachers should be paying attention. 

AI is becoming impossible to block—but what if the real lesson from DeepSeek isn’t about stopping AI at all? What if it’s about rethinking how we build readers?

Wait—What Even Is DeepSeek?

Let’s back up for a second.

Most of us have heard of AI chatbots like ChatGPT—the tool students are either using to cheat or struggling to get good responses from. These chatbots run on massive AI models that generate human-like text based on vast amounts of training data.

DeepSeek is the newest player in the AI space, and what makes it different is how it thinks.

Instead of using one massive, resource-hungry model like ChatGPT, DeepSeek uses something called a Mixture of Experts—a system where different smaller AI models activate only when needed. It’s a modular, open-source approach that makes the AI faster, more adaptable, and harder to detect.

In other words, DeepSeek didn’t just make a better chatbot—it reengineered how knowledge gets processed. And that’s exactly the shift that reading instruction needs, too.

When DeepSeek’s breakthrough hit the news last week, I found myself in an unusual position. 

As both an English teacher and someone who’s spent the last few years experimenting with various large language models, I saw something most coverage missed: The parallels between how DeepSeek revolutionized AI development and how we might revolutionize reading instruction.

Not the surface-level “AI in education” discussions we’ve all heard. Something deeper – about how constraints drive innovation, how open-source development creates unexpected breakthroughs, and how rethinking basic architecture can transform what’s possible.

I encourage teachers to lean on their strengths when it comes to teaching and use them to build powerful systems. 

But what happens when what we’re doing isn’t working? What if the constraints we’re given is what makes teaching feel untenable? 

The Problem Isn’t Just AI—It’s That Reading Instruction is Buckling Under Pressure. 

Teachers are being asked to do the impossible:

• Keep students engaged when they barely want to pick up a book.
• Combat AI-assisted cheating without resorting to endless surveillance.
• Teach deep analysis when students skim everything at surface level.
• Bridge massive skill gaps in classes where some students read fluently and others struggle with basic comprehension.

And the go-to solutions? More test prep, more close reading exercises, more structured scaffolding. But what if the problem isn’t that students need more support—but that they need a different way of building understanding altogether?

The way we teach reading has barely changed in decades, even as the way students engage with information has shifted dramatically. If AI breakthroughs like DeepSeek are proving that constraints lead to innovation, maybe our biggest failure in reading instruction is clinging to outdated structures instead of reengineering how students build understanding.

By rethinking assumptions about how AI should be built (constrained resources, open source approach), DeepSeek achieved what billion-dollar companies couldn’t. What if English teaching isn’t an AI problem – it’s an engineering problem? How do we design systems that actually work for learning instead of just adapting existing ones?

What if English teachers took the same approach to building readers?

The Real Problem Isn’t Cheating—It’s That We’re Asking the Wrong Questions

Worrying whether AI wrote a student essay misses the bigger issue.

• AI is here.
• AI is getting better.
• And AI will keep doing the kinds of thinking that students used to be graded on.

So instead of spending energy on detection, we should ask:

• What kind of thinking still belongs to students?
• What skills will still matter when AI can generate perfectly structured five-paragraph essays on command?

The answer isn’t in catching AI use—it’s in raising the bar on what we ask students to do.

It’s a tall order, especially when students’ basic literacy is at an all time low.

But this is THE problem we need to solve.

The Big Shift: From “Teaching Books” to “Building Reading Minds”

The big misconception is that English teachers spend time teaching books—delivering content, guiding discussions, correcting misinterpretations.

But that’s not really it, is it? Or…that’s not just it.

When we really get to the crux of it, isn’t the real job is helping students think more strategically and intensely about texts?

• Creating mental architectures that allow students to navigate complexity.
• Designing systems that build schema rapidly and robustly.
• Training students not to know the answer—but to know how to think.

This is truly the work. 

And if DeepSeek just reinvented AI by making knowledge more modular, dynamic, and interconnected, maybe it’s time we do the same for reading instruction.

I know you’re picking up what I’m putting down.

DeepSeek and the New Floor of Learning: Why English Teachers Need to Raise the Bar

For years, we’ve taught students to recognize big themes in literature:

• What does this text say about power?
• How does this novel explore identity?
• What does this poem reveal about loss?

These questions used to be a mark of critical thinking. But now? AI can generate acceptable responses to them in seconds.

That’s what DeepSeek represents—not just a technical advancement, but a fundamental shift in what counts as “thinking.”

And if AI has changed the baseline for writing and analysis, English teachers need to rethink what learning actually looks like.

The DeepSeek Reading Problem: We’re Teaching Readers Like Outdated AI Models

Despite everything we know about how learning works, most reading instruction still follows a linear, closed-system model:

1. Read the book front to back.
2. Check comprehension at key points.
3. Identify key passages (usually pre-selected by the teacher).
4. Discuss teacher-generated questions.
5. Write an analysis using a structured template.

This is the reading version of brute-force AI—processing every word in sequence, following a rigid, one-size-fits-all approach, and treating knowledge as something students receive rather than actively shape.

But the best readers don’t work that way.

• They jump between ideas, connecting patterns across texts.
• They activate schema dynamically, pulling in knowledge from multiple sources.
• They adapt their strategies, shifting their approach based on purpose and genre.

If this is what is happening with reading, why don’t we teach around this reality?

AI Is Here, So We Need to Double Down on Intellectual Literacy

The real challenge isn’t AI replacing thinking—it’s people assuming that AI is thinking for them.

If students don’t understand how knowledge works—how ideas are formed, how arguments are built, how meaning shifts based on context—then AI will only make them think they know things.

That’s the real danger—not AI itself, but the erosion of intellectual self-awareness.

The New Role of the English Teacher

For years, English teachers have justified our work by saying, “Students need to learn how to think.”
That is still very, very true.

But now, AI can mimic a lot of what we used to count as “thinking.”

So the real question isn’t just what we teach—it’s why it still matters.

• What kinds of thinking are uniquely human?
• What is still worth teaching when AI can do so much?
• How do we ensure that our students’ learning goes beyond what a machine can generate?

DeepSeek isn’t just a new AI model. It’s a warning. 

It’s telling us the way we’ve been teaching English isn’t enough anymore.

But that’s not a crisis—it’s an opportunity.

Because if we teach students to analyze, differentiate, synthesize, and track complexity, we’re not just keeping education relevant.

We’re making it better.

AI isn’t making intellectual literacy irrelevant.
It’s making intellectual literacy the most important skill of the 21st century.

We need to teach them how to think critically, track complexity, and evaluate knowledge.

It’s not enough to just teach students how to read, write, and analyze. In an AI-driven world, students need to understand how knowledge itself is structured.

This means doubling down on:

• Helping students see how meaning is built—not just what the meaning is.
• Teaching students to track how an idea evolves across a text.
• Making students engage with how arguments are constructed, not just consume the final product.

Because if students don’t know how knowledge works, they won’t be able to tell the difference between real understanding and something AI-generated.

Looking Forward 

DeepSeek’s breakthrough isn’t just about AI. It’s a signal that old systems—whether in tech or education—are being disrupted faster than we ever imagined. 

English teachers can either fight the shift or embrace it. AI isn’t going away, but our best response isn’t to shut it out. It’s to design better systems for real learning—ones that are resilient, adaptable, and, like DeepSeek, built for the future.

We can’t out-teach AI with the same old methods. But we can design a better way to build reading minds. The question is: Are we willing to rethink the system—or are we waiting for something else to disrupt it first?