AI is reshaping student learning in two directions at once: it can personalize instruction and remove barriers for struggling learners, but it can also weaken the critical thinking and writing skills students need most. The net effect depends almost entirely on how students and educators use these tools, and right now, many classrooms are still figuring that out.
Personalized Pacing and Instant Feedback
The clearest benefit of AI in education is its ability to adapt to individual students in real time. Traditional classrooms move at one speed for everyone. AI-powered platforms can identify where a student is struggling, adjust the difficulty of practice problems, and offer targeted explanations without waiting for a teacher to notice a raised hand. This kind of adaptive learning helps students who would otherwise fall behind or get bored stay engaged with material at the right level of challenge.
Khan Academy has been one of the most visible examples. Its AI tutoring system, Khanmigo, is designed to guide students through problems using a Socratic approach, asking questions rather than giving away answers. The organization measures success by tracking whether students show increased cognitive engagement, progress toward skill proficiency, and measurable learning gains on external assessments. As of early 2026, Khan Academy had completed more than 60 experiments to refine how its AI tutor interacts with students, focusing on improving accuracy and response speed while preventing the system from simply handing out solutions.
Automated grading is another practical gain. When AI handles routine assessment of quizzes, short answers, or even draft feedback on essays, teachers can spend more of their time on the parts of instruction that require a human: one-on-one conversations, classroom discussion, and identifying students who need extra support. For students, faster feedback means less time waiting to find out what they got wrong and more time correcting misunderstandings while the material is still fresh.
The Cost to Critical Thinking
The flip side is harder to measure but potentially more consequential. Research from Duke University’s Center for Teaching and Learning found that university students who used large language models to complete writing and research tasks experienced reduced mental effort but demonstrated poorer reasoning and argumentation skills compared to students who relied on traditional search methods. In other words, the work felt easier, but the learning was shallower.
A separate study found that students using LLMs focused on a narrower set of ideas, producing analyses that were more biased and superficial. This makes intuitive sense. When a chatbot generates a polished paragraph in seconds, students have less incentive to wrestle with conflicting sources, build an argument from scratch, or sit with the discomfort of not knowing the answer yet. Those uncomfortable moments are where deeper learning happens.
The risk is especially pronounced for writing. Composing an essay forces students to organize their thinking, identify gaps in their logic, and make choices about what matters most. When AI handles those steps, the student becomes an editor of machine-generated text rather than a thinker working through a problem. Over time, students who lean heavily on AI for writing may arrive at exams, job interviews, or workplace situations without the analytical muscles they assumed they were building.
How AI Helps Students With Disabilities
For students with learning disabilities, AI tools are solving problems that traditional accommodations never fully addressed. Students with ADHD, dyslexia, dyspraxia, and autism face specific barriers in academic writing, from difficulty organizing thoughts to challenges with spelling, grammar, and sustained focus. Generative AI tools, particularly chatbots like ChatGPT and rewriting applications, are helping these students overcome those barriers across a wide range of writing tasks.
A student with dyslexia, for example, might use AI to restructure a rough draft into clearer sentences, preserving their original ideas while removing the mechanical errors that would otherwise obscure their thinking. A student with ADHD might use a chatbot to break a large assignment into smaller steps or generate an outline to work from. Translation software helps multilingual students with processing differences work across languages more fluidly.
This matters because academic writing is often the gatekeeper for grades, degree completion, and professional opportunity. When AI removes the mechanical friction without replacing the student’s thinking, it can level a playing field that was never designed for neurodivergent learners in the first place.
Academic Integrity Under Pressure
Faculty are deeply worried about what AI means for honest work. College Board research found that 92% of faculty are concerned about plagiarism or dishonesty facilitated by AI. That concern is grounded in what they’re seeing: 74% of faculty report that students are using AI to write essays or papers, and 67% say students are using it to paraphrase or rewrite content. Almost half of faculty believe at least half of their students are turning to AI for writing-related tasks.
The challenge goes beyond catching cheaters. Nearly three-quarters of faculty say they face at least minor difficulties managing student AI use in their courses, and only 21% feel very confident guiding that use. The remaining 79% say they are either just beginning to explore what’s needed or still need guidance themselves. Faculty in English and other writing-intensive disciplines report the highest levels of classroom disruption.
Institutional support has been uneven. Almost half of faculty report having a formal classroom policy on AI use, but many say the guidance from their schools remains inconsistent. This creates a patchwork where expectations vary from one class to the next, leaving students confused about what’s acceptable and instructors unsure how to enforce boundaries they’re still defining.
When AI Helps and When It Hurts
The pattern that emerges from the evidence is fairly clear. AI tends to help learning when it acts as a tutor or scaffold: asking questions, adjusting difficulty, providing feedback, and removing barriers that prevent a student from engaging with the material. It tends to hurt learning when it acts as a substitute for the student’s own thinking: generating essays, summarizing readings the student never read, or producing answers the student never worked through.
The distinction often comes down to whether the student is doing the cognitive work. Using an AI tutor to get unstuck on a math problem, then solving the next three problems independently, builds skill. Pasting an essay prompt into a chatbot and submitting the output with light edits does not. For students with disabilities, AI can remove mechanical obstacles while preserving the intellectual challenge, which is perhaps the ideal use case.
Students who use AI strategically, as a study partner rather than a ghostwriter, are likely to benefit. Those who use it to avoid effort are likely to graduate with gaps in reasoning, writing, and problem-solving that will surface later. The tool itself is neutral. What matters is whether it’s amplifying the student’s thinking or replacing it.