When AI Makes Learning Appear Easier, How Do We Argue for Struggle?

At the start of every academic year, millions of students walk into classrooms carrying more than just laptops, books and syllabi. They bring expectations about the life they hope education will help them achieve.

With AI tools, students can get an explanation, a solved problem, a drafted essay, in seconds. But access to knowledge doesn't automatically lead to understanding. Students still have to make meaning from what they encounter, connect new ideas to old ones, find the time to have a new experience or input, and sit with uncertainty long enough to let the process of learning take place.

We now have enough experience with AI in education to move past the novelty phase. Instructors and students have lived with these tools long enough that we're beginning to have data on what actually helps learning, what gets in its way, and which parts of the process cannot be skipped. We humans still need to embrace the productive struggle.

One could argue (and I often do) that in a world where work and life skills are changing at an accelerating pace, the ability to learn is one of the most important skills humans can acquire. We need to be curious, to be good at learning new things, to know when a productive intellectual challenge drives greater reward, and to develop our own metacognitive awareness.

The challenge now is how we teach all of this to students, when they can generate an answer in seconds or hand the work to an agent. How do we help a novice learner recognize when they're undermining their own learning? And how can we give them experiences where they practice confidence in their ability to learn?

I am often asked what I think of the latest AI tools, and whether I’m excited about what they will bring to education.  To be sure, the increased capabilities are genuinely impressive. I share the enthusiasm for personalized learning paths, and the possibility that any student could get the help they need, when they need it, in a way that will most likely be understandable to them.

But even more than the tech, I am excited to see the emerging wave of instructors leaning into a better enabled and more human kind of teaching.

Recent 2026 studies on cognitive offloading show that while AI/LLM tools improve task accuracy and speed, heavy reliance on them reduces neural connectivity, retention, and long-term skill formation when the tools are removed. The question then becomes how can we best use AI to serve one’s purpose? And how can we develop that skill in learners?

For those who engaged early with how we teach in the age of AI, some useful approaches emerged: helping students see past the simple answer, having them verify output, and creating activities that test critical thinking behind the prompting. Because we became aware of the risks of cognitive overreliance when AI tools are overused, we can now articulate the benefits of using AI as an intellectual sparring partner, or as a way of taking an idea further. But more and more we can see assignments and work habits that encourage taking a first creative pass on your own. Start with the messiness and uniqueness of creative human thought, the equivalent of the hand-written note on the back of the napkin, and then co-design with AI. Instead of outsourcing one’s thinking, it’s an enhancement. To me, that’s real progress. We continue to learn about learning.

Student attitudes are evolving too. While the initial pull toward the easy answer was clearly a motivator in using AI, a more experienced view is emerging. We can see in their comments, notes and feedback, that they want to learn how to use these powerful tools, but do not want their education short-circuited by them. They want clearer guidance on when using AI tools is acceptable, and they want clarity on the rationale behind different rules for different classes. They do not like the idea of an instructor outsourcing their work to AI, but they are happy to have more engaging projects, and practice using AI to truly assist in their learning experiences. The conversation is starting to be less about whether students should use AI, and more about what they gain or give up when they do.

The next era of education needs to be technologically ambitious, deeply practical, and grounded in how learning actually happens. We need tools that personalize support and learning environments that create connection, not isolation. We need faster feedback and better judgment, better insights into learning, and a deeper understanding of what it actually represents. This is where digital learning companies like Macmillan Learning have both a responsibility and an opportunity.

Building for the real modern classroom means listening to instructors who have too little time, students who are busy and need resources and support, and institutions navigating enormous pressure and change. It means being open to new learning pathways. It means studying what works to enhance the learning experience, being honest about what does not, and building solutions that reflect how learning actually happens -- not how we wish it did. Technology alone will not solve education’s hardest human problems. But I share an optimism that thoughtfully designed tools can help us reach students in more timely, personal and effective ways.

Some things should become easier. Students should be able to find help when they need it in a format that is optimized for them. Instructors should have better insight into where students are struggling and how to extend their expertise to more learners, in an amplified way. But thinking deeply, learning to relate to other humans, developing judgment and learning how to learn are not problems technology should solve for us.

At its best, education expands what people believe is possible for themselves and for the world around them. The tools, expectations, and classroom itself will keep evolving, and I would argue that they should. But the purpose of education will not. It is still about helping people move from uncertainty to understanding, from potential to progress, from “I don’t know” to “I can.” And that future will be built by people who understand not only what technology can do, but what learning requires.

 

Susan Winslow is CEO of Macmillan Learning, a Holtzbrinck company. She champions research-driven innovation at the intersection of pedagogy, technology, and AI, advocating for the thoughtful and responsible use of emerging technologies to advance teaching and learning success.