Learning Was Never Supposed to be Easy

By Devon Wible, Vice President, Teaching and Learning, Catapult Learning

Would I Want This for My Own Children? 

I think about Artificial Intelligence in education from several perspectives: as a 25-year educator; an educational services developer; a parent of three school-aged children; and the board chair of my local board of education. These roles make me hopeful and cautious about AI in education. 

AI offers real promise. It can help adults plan, summarize, organize, adapt, analyze, and communicate. It can also reduce paperwork and help educators identify patterns they might otherwise not see. When used well, it may allow teachers and tutors to spend more time on the work that matters most: working with their students. 

But I also worry that what makes AI powerful could also make it risky in education. 

AI is an effective shortcut. It can create answers, explanations, drafts, summaries, and solutions much faster than students can on their own. While this efficiency benefits many adult tasks, learning is different. Learning is not meant to be as simple as receiving a generated answer. 

Learning demands cognitive effort, including attention, memory, language, reasoning, persistence, background knowledge, schema, feedback, and time. Students must build understanding, not just arrive at an answer. They need to question information, discern facts from fiction, read for meaning, solve complex problems, communicate clearly, and collaborate. These are not extra benefits of education—they are its primary goals. 

As an educational services developer, I follow a simple rule: I will not design or support anything for other people’s children that I would not want my own children to use. 

When I think about my own children, I see how much they learn through curiosity, discovery, trial and error, confusion, revision, and mistakes. They need opportunities to wonder, test ideas, struggle productively, make errors, and understand why their answers work. That is not just parental instinct; it is consistent with what we know about the science of learning: durable learning depends on active processing, feedback, retrieval, practice, and gradual knowledge construction. 

AI often shortens this process, which can weaken the deep learning necessary for retention and transfer. Sometimes it gets students to an answer too quickly, too easily, or even the wrong answer with great confidence.  Children often lack the reasoning skills, background knowledge, or discernment to reliably challenge the “thinking” behind the technology. That is not a criticism of children. It is a reality of how their brains develop and how learning occurs. 

If we introduce AI to students without careful consideration, especially as a tutor or learning companion, we risk encouraging passive consumption of generated answers rather than active construction of understanding. This approach may teach students to accept any answer the technology provides rather than develop the skills necessary to question it. 

That is not the type of learning I want for my children, or for any children. 

This concern is even more critical for students who struggle. 

Too often, the debate centers on whether AI can deliver individualized instruction more efficiently than human tutors. But in learning, especially for students who struggle, efficiency is not always the main goal. What matters is deep learning, lasting understanding, confidence, the ability to explain, revise, persist, and transfer what they learned in new ways. 

The place to seek efficiency is not in the minutes a struggling learner spends with a skilled adult. We must stop seeing time as unnecessary overhead and instead see it as a crucial part of the learning process. 

Struggling learners need more than access to explanations or quick AI-generated answers. They deserve someone who can notice when they hesitate, get frustrated, avoid work, feel confident, have misunderstandings, face language barriers, and, most importantly, the moment a student starts to believe in themselves. They need feedback, encouragement, productive struggle, and real relationships. They need adults who can figure out what is behind a mistake and decide when to scaffold, when to push, when to reteach, and when to remind them, “You can do this.” 

AI can support educators in these efforts, but it should not replace them. 

Recent research reinforces that AI tools are not a panacea. The National Student Support Accelerator (NSSA) at Stanford has argued that “AI tutoring” is not one thing. There is a meaningful difference between human-led tutoring supported by AI, AI-led tutoring with some human oversight, and AI-only tutoring. NSSA’s 2026 brief concludes that current research supports using AI to enhance tutor effectiveness and educator capacity, rather than replacing high-impact tutoring with AI-led models.¹ 

In two randomized controlled trials involving 355 elementary students, those assigned to independent use of an AI literacy platform used it very little. Nearly half never used the platform, and students who did spent only two to five minutes per week engaged in instruction. Human support increased engagement, but usage remained low, and the intervention did not improve reading achievement (Robinson et. al, 2026).² 

The lesson is not that AI tools lack promise, but that access does not guarantee learning. Just because a student has access to a tool does not mean they use it in a way that helps them learn deeply. A platform generating feedback is not the same as students internalizing concepts. Scalability does not guarantee effectiveness. 

Recent reporting from The74 highlights similar concerns about engagement and usage. Several states are spending millions on AI tutoring tools, even though researchers caution that student uptake remains low and evidence at scale is still limited.³ This is especially important for education leaders to consider, as students most likely to receive automated tools are often those ones who most need support from real people. 

At the same time, AI can be a useful tool for educators. It can help tutors review assessment data, identify likely misconceptions, generate aligned practice, prepare multiple scaffolds, adapt materials for multilingual learners or students with disabilities, summarize skill growth, flag students needing extra help, and reduce documentation burdens. 

Operating in-person tutoring at scale is challenging, so meaningful efficiencies are crucial to program success. High-quality tutoring programs require scheduling, tutor training, consistency, progress monitoring, curriculum alignment, communication with schools and families, and continuous improvement. AI can help streamline some of these tasks. 

However, the purpose of efficiency must be clear: it should provide more time for instruction, better human preparation, and stronger relationships, not simply automated substitutes. 

We must also carefully define success. AI can make students appear more productive by helping them complete tasks faster and move through content quickly. However, increased speed or output does not always indicate genuine learning. 

A 2026 CEPR paper on generative AI use among Chinese secondary students found that AI adoption increased homework scores and reduced completion time, but exam performance on the same content declined.⁵ While the context differs from U.S. tutoring, the results raise a critical point: short-term productivity can mask weaker long-term learning. 

Learning is more than completing academic tasks. Reading, writing, mathematical reasoning, and problem solving require attention, memory, sequencing, language, persistence, and active sense-making. Some education writers describe the literacy crisis as a cognitive crisis, arguing that students need more opportunities for sustained mental effort, not fewer.⁶ For AI, the implication is clear: tools that make work faster and easier do not necessarily deepen learning. 

So yes, we should explore AI in tutoring. But we should be precise about where it belongs. 

AI should help tutors answer questions like: 

• What does this student understand now? 

• What misconception is most likely blocking progress? 

• What scaffold might help without lowering the rigor? 

• What practice should come next? 

• What pattern am I missing across students? 

• How can I spend less time preparing materials and more time teaching? 

AI should not lead us to ask: 

• How little human support can this student get? 

• How cheaply can we deliver intervention? 

• How quickly can students move through content? 

• How much adult judgment can we automate? 

The future of tutoring should not be less human, but more human where it matters most. AI can assist with planning, analysis, documentation, material creation, scheduling, and quality review. However, the core work of intervention must remain relational, diagnostic, and deeply human. 

The promise of AI in education is not to make learning easy for children, but to help adults protect the time, attention, and instructional depth that real learning requires. 

That is the line we must hold. 

Sources 

¹ National Student Support Accelerator, “AI Tutoring is Not a Monolith: What We Actually Know,” August 20, 2026. https://nssa.stanford.edu/briefs/ai-tutoring-not-monolith 

² Carly D. Robinson, David Gormley, Ana Trindade Ribeiro, and Susanna Loeb, “Access is Not Enough: Human Support Improves Engagement with AI Tutoring,” EdWorkingPaper No. 26-1451, June 2026. https://edworkingpapers.com/sites/default/files/ai26-1451.pdf 

³ Linda Jacobson, “AI Tutors Not Yet a Replacement for Humans, Research Says,” The 74, September 1, 2026. https://www.the74million.org/article/ai-tutors-not-yet-a-replacement-for-humans-research-says/ 

⁴ Natasha Singer, “How Big Tech Captured American Schools,” The New York Times, August 23, 2026. https://www.nytimes.com/2026/08/23/business/schools-big-tech-google-microsoft.html 

⁵ David Strömberg, Victor Lei, and Yanhui Wu, “The Generative AI Learning Penalty: Evidence from Chinese Secondary Education,” CEPR Discussion Paper No. 21577, June 2, 2026. https://cepr.org/publications/dp21577 

⁶ Tara Bonner, “The Literacy Crisis Is Actually a Cognitive Crisis,” Arrowsmith, June 9, 2026. https://www.arrowsmith.ca/blog/the-literacy-crisis-is-actually-a-cognitive-crisis

Ready to Learn More?

Catapult Learning is proud to offer an ESSA tier-2 validated tutoring program. Through our commitment to strong partnerships and seamless implementation, you can rest assured that your student’s will receive the support you need. It is why we see results such as 100% of struggling readers improving at least one grade level after our implementation.

Take the next step, and visit our High-Impact Tutoring page to see our model in action, read more success stories, and to contact us for a free consultation. Together, we can achieve brighter outcomes.

Does AI tutoring improve student reading achievement?

Not on its own. A 2026 randomized controlled trial of 355 elementary students found that independent use of an AI literacy platform did not improve reading achievement — nearly half of students never used it, and those who did averaged only two to five minutes of instructional engagement per week. Adding human support increased engagement, but usage still remained low. The takeaway for district leaders: giving students access to an AI tool is not the same as ensuring they learn from it.

What’s the difference between AI-led, human-led, and AI-only tutoring?

The National Student Support Accelerator (NSSA) at Stanford distinguishes three models: human-led tutoring supported by AI tools, AI-led tutoring with some human oversight, and AI-only tutoring with no human involvement. NSSA’s 2026 brief concludes that current research supports the first model — using AI to strengthen tutor effectiveness and educator capacity — rather than replacing high-impact human tutoring with AI-led approaches. For school and district leaders evaluating vendors, this distinction matters more than whether a product is simply labeled “AI tutoring.

Should school leaders replace human tutors with AI tutoring platforms to save costs?

Available research cautions against this. A 2026 CEPR study of Chinese secondary students found that generative AI use raised homework scores and cut completion time, but exam performance on the same material declined — suggesting AI can create an illusion of progress without deeper learning. Reporting from The 74 similarly notes that several states have invested heavily in AI tutoring tools even though student uptake remains low and large-scale evidence of effectiveness is still limited. The stronger use case for AI is supporting tutors — with data review, misconception-flagging, and materials prep — not substituting for the relational, diagnostic work of a human tutor.