Using AI for Student Success Without Losing Sight of Student Learning

In an earlier article, I argued that higher education is in a period of transformation, spurred by the effects of AI, that calls for institutional-level leadership and strategic thinking. The response to AI cannot be limited to the reactive and tactical. It cannot even be limited to forward-leaning experiments. Institutions need to bring together cross-functional collaborations that ground responses to AI in questions of mission.

In that article, I also argued that these institution-wide collaborations should be guided by a commonly understood “north star.” In this article, I want to suggest a particular north star, AI that works for student success for every learner, and to lay out some of the teaching and learning decisions that will keep institutions steering toward it.

Orienting questions around AI toward student success for every learner is a useful framing, because it aligns with the “explore, not ignore” posture Every Learner Everywhere encourages. Educators are right to be concerned about uses of AI that obscure whether learning has occurred or that benefit only some learners. Seeking uses of AI that work for student success for every learner helps distinguish those from those that expand access to active learning, data-informed learning, metacognition, and other evidence-based teaching practices.

3 factors for AI-enabled student success

1. Accessibility

AI can power captioning, text simplification, translation, and alternative formats that expand access for students with disabilities and for multilingual learners when designed with accessibility in mind. Work from Every Learner and Teach Access focuses on how AI can generate alternative formats, adapt content, and support inclusive design at scale.

2. Feedback and learning support

Generative AI can give low-stakes, iterative feedback, help students practice explanations, and support writing and problem solving, especially when guided by clear prompts and faculty framing. Faculty can position AI as a “thinking partner” that students must critique, correct, and compare with human sources, building critical AI literacies. (For an inspiring example, read how April Crenshaw, now Every Learner’s Instructional Design Implementation Manager, built an AI math tutor in her previous faculty position.) This can be especially powerful in large-enrollment gateway courses.

3. Faculty capacity and course redesign

Used well, AI can streamline routine prep work so instructors can spend more time on direct student support. AI can also help instructors prototype activities, generate examples, and align materials with course objectives. Every Learner’s growing library of resources (such as the Faculty Development and GenAI Playbook) emphasizes pairing tools with structured reflection and redesign, not copy-paste lesson plans.

Careful exploration of AI’s potential

As I emphasized in my earlier article, Every Learner encourages open-minded exploration with AI but not uncritical adoption. Faculty are right to consider the many serious questions this unproven technology raises.

Some of those questions are around assessment, authorship, and integrity. Overreliance on AI for writing or problem solving without redesigning assessment can blur what learning is actually happening and can widen student success gaps if expectations are unclear. If we don’t rethink assessment, we risk grading large language models, not learners. Faculty are asking for clear institutional guidance on acceptable AI use in assignments and how to design assessments that foreground process, reflection, and in-class performance. The answer is not an arms race with detection tools; it’s redesigning assessment around what AI cannot do well without the student’s active thinking.

To be fair, students also are asking for clear, consistent guidance on acceptable use of AI. Research on student use conducted by Every Learner’s student interns shows that they also worry about their own agency and growing dependence on AI. If students are not taught to use AI as a thinking partner rather than an answer engine, they may offload too much cognitive work and weaken critical thinking, their disciplinary identity, and their career prospects in professional roles that increasingly require AI literacies. Our job is to help students use AI to see their thinking more clearly, not to hide it.

Another reason to move carefully with AI is how it can change faculty workload and expectations. AI integration can easily become “one more thing” if institutions expect faculty to individually experiment and rebuild courses with no time, training, or support. Emerging practice from faculty-led AI groups shows that communities of practice, templates, and shared language (e.g., recommended syllabus statements) reduce burden and inconsistency. If institutions want responsible AI use in teaching, they must give faculty and staff the resources to learn, test, and iterate together.

AI that supports evidence-based teaching

Higher education should accelerate use of AI where it clearly advances student success for every learner, but colleges and universities should move deliberately where AI touches judgment about students or the roles faculty and other staff play.

Every Learner’s latest initiative is supporting implementation of a new courseware in calculus courses. The Lighthouse institutions in this initiative have selected courses with high DFW rates, and departments will pilot an AI-supported tutor that lives inside the courseware and is trained on the exact terminology and thought process as the lessons, homework, and quizzes. Students can ask questions in their own words, get step-by-step explanations, and request simpler language or another example when they’re stuck at 11 p.m.

Critically, what they cannot get from the AI tutor is the answer, which is trained to keep prompting with hints and learning strategies without doing the work. This kind of AI use allows the faculty teaching these calculus courses to redesign assessments. Instead of relying on auto-gradable problem sets, they add in-class concept checks, group work, and oral explanations.

That initiative illustrates the sweet spot for the work Every Learner Everywhere does: Use AI to improve student success, not efficiency, and to widen access to explanation, practice, and other teaching methods that experienced faculty know work.