Students’ Guide to Using GAI for Learning

Effective Practices and Cognitive Pitfalls

©Learning Strategies Center, Cornell University

February 2026

What do Cornell students think about using GAI in learning?

“Your goals should inform if using GAI is helpful or not. If your goal is to take the brainpower out, that’s helpful. But if you want to really learn a concept, using GAI may not be meeting your goal.”- Cornell Info Sci student

“Students need to learn to use GAI in a way that helps them think critically. Instead of teachers saying they cannot use it.” –Cornell Physics student

“GAI can save time, especially with repetitive tasks or analyzing lots of information.” –Cornell Global & Public Health Sci student

“GAI can be addictive, enabling you to not be ok with discomfort and imperfection. It can hinder you from having the patience to struggle, which is an important part of being human.” –Cornell English student (pre-law)

“It may not be accurate.” –Cornell Biological Sciences student

“Students need to be good prompt engineers to use GAI well.”-Cornell Human Bio, Health, & Society student 

CAN I use AI for my class?

Every class and every instructor has different expectations and policies about if, when, and how it’s OK for students to use AI for coursework. As a student, you are responsible for learning and adhering to your instructors’ policies and expectations, which should be in the course syllabus. If you don’t know whether a certain use of AI is allowed, or if you’re unclear about the policies in general, ask your instructor in class, stop by their office hours or send them an email.

It is students’ responsibility to follow instructors’ expectations. If you are not sure if a specific GAI use is allowed in your class, you need to ask. It’s on you to know.

Cornell’s recommended GAI platform is Co-Pilot: When you log in to Copilot with your NetID, your prompts, answers, and viewed content are not used to train the underlying large language models.

Asking GAI the right questions is key! Here is some Cornell guidance on that.

SHOULD I use AI for my learning?

In learning, as in most applications, there is a big gap between what AI can do and what it should be used for. On one hand, AI tools are being used by professionals, executives, and scholars, so figuring out how to incorporate them productively into your work can be beneficial down the line. On the other hand, using AI carelessly can rob people of the chance to hone the complex cognitive skills that are essential to our professional, social, and intellectual lives.

Mastering a body of knowledge, developing marketable skills, and improving the capacity for critical thinking are core values of a college education. Achieving these things takes time, hard work, and repeated practice. Cavalier use of AI might appear to make this work faster or easier, but be cautious. When students use AI to outsource their thinking it can cause more harm than good. Emerging research is shedding light on the cognitive cost of using AI like a crutch instead of as an appropriate helper or partner.

AI is very good at rote tasks and quickly synthesizing (mostly accurate) information, but it is not able to create new knowledge. As a human, accessing that level of creativity requires deep knowledge, and you already know that genuine proficiency takes time and effort. When students use AI to take shortcuts in their learning, they rob their future selves of the ability to truly innovate. (Ironically, the ability to create and innovate sets humans’ skills above those of AI!) Our goal for you is that you will learn to use AI as a learning partner, without offloading work that sacrifices your cognitive development.

Whether or not you should use AI in your learning is not always clear-cut. Depending on your goals and the context, there are advantages and drawbacks. Ultimately, it’s up to you to decide for yourself: what tasks are you OK outsourcing to AI (assuming your instructor allows it!), and is the long-term cognitive cost worth the short-term time savings?

Use this chart to help you decide when and whether you want to use AI for your learning!

You MUST keep in mind that different instructors have different expectations and classroom policies on if, when, and how it is okay for students to use AI. It’s on you to know and follow your instructors’ expectations. Using AI as a study tool that we haven’t listed? Please let us know!

 

Specific uses of GAI in learning (in absolutely random order)	Why students find this helpful:	Potential pitfalls to look out for:
Make practice tests, summarize lecture notes, make flashcards	Saves time You can get AI to cut out extraneous or repeated information to pare down the material you need to work with.	“If you’re using chat to summarize lecture notes etc., it skips the step of trying to understand it.” Part of the benefit of flash cards, summarizing, etc., comes from the work of making them-not just reviewing them. Producing these materials engages the cognitive process of “elaboration,” which has an important role in the retention of new information.
Quiz you	“Like a virtual study buddy!”	You need to know enough to know when it’s wrong. “AI can hinder interpersonal skills.”
Simplify complex readings or concepts, “explain it to me like I’m in kindergarten.” Or, explain something multiple ways.	“Asking AI to rephrase or to explain a concept in a differently allows me to understand the content much more easily. I’m able to get paragraph explanations, links to different references, and bullet points.”	“This could be dangerous because as we know, AI isn’t always correct. I am sometimes able to catch mistakes, but sometimes I miss them, leading to confusion down the road.” Grappling with complex texts and concepts builds critical thinking and higher-order cognitive skills like analysis and synthesis. Longer and more complex texts require you to slow down connecting new material with your existing knowledge. This cognitive process of “elaboration” is important for retaining new information. Taking time to work with complex ideas without AI is slower, but you learn the information better.
Editing and writing (words to use in papers, labs, emails, or code) and reviewing your work (“is this code right, am I hitting the right tone in this email, did I answer all the prompts in this essay?”)	“It’s good at generating boiler plate.”   AI can be useful at various stages of the writing process – check out the “Human-AI Writing Loop” in the Student Guide to AI for specific ideas.	You might not want to sound like boiler plate.   AI may not be faster than you at producing good work. A recent study found that coders who used AI were 19% slower at developing a satisfactory product.
Create a plan for studying or completing a project	Faced with a big, complex project, it can be difficult to know where to begin – and the sheer size of the task might prevent you from wanting to even start it. GAI tools can help you make a plan by breaking down a project into smaller pieces.	Plans developed by GAI might appear shiny on the surface, but look closely and you’ll find they may be generic and not well-suited for your specific needs. Sometimes AI just gives bad advice. You might spend more time revising what a chatbot suggested than you would just making a plan yourself. Being able to plan out a big task is a useful skill that can only be developed through repeated practice. Leaning on GAI for planning could be harmful to your ability to “adult” in the long run.
Problem-solving, the whole range from getting help AFTER you did it yourself so you can understand where you went wrong (good!), to asking for hints when you’re stuck (neutral), to having it solve the problem for you (bad!).	Can be really helpful if you’re stuck.   Getting stuck can be frustrating.	“If you’re used to knowing or not knowing and have Chat fill in the gaps, you may be frustrated on the exam because you haven’t had experience with getting yourself unstuck.” “There are things Chat just can’t solve, you need to have the patience to sit down and do it. Whether it’s for a test or just something too complex for Chat.” “If you use AI to explain problems, it can mislead you. You can trick yourself into thinking you know how to do the problem. But if you haven’t done the practice, you may not really be able to do it on your own.“ Being stuck and frustrated is part of learning.
Develop mnemonics and memory devices	Makes memorization more efficient “As a biology major, there is a lot to memorize, with so many cycles, pathways, etc. I’ve found that mnemonics created by AI helped me remember lots these concepts. It’s helpful for a class where memorization is the main challenge.”	One of the reasons mnemonics are helpful is that coming up with them on your own means engaging in “elaboration” – the psychological process where you think about how a new piece of information fits into your existing knowledge. When you outsource this to AI, you shortchange the learning process.
Research	AI can be a good place to start understanding basic facts and ideas, which can give you ideas for more in-depth research with human-made sources.	“Chat is trained to give you some facts. The tricks of the trade, the info that you need to be able to use yourself, is completely lost when you just take the summary instead of doing the research yourself.” “For classes where you need to apply concepts to new situations, you can get the false idea that you can do application.”
Turn lecture notes into a podcast	It can be great to listen to your lecture notes in a more conversational format.	Might not hit all of the important points. Overreliance can lead to gaps in knowledge and/or not knowing what your professor thinks is most important.
Time management	Outsourcing basic stuff can save time.	“With skills like time management, you can get Chat to help. But like with everything else, you are not gaining the skill of managing your own time well. That goes away when I offload it on chat.”
Answer questions instead of going to office hours or tutoring	24/7 availability, from the comfort of anywhere you are	“Mind the gap between what the professor is teaching/wants students to know and what they are learning from Chat.” “With AI, you might only get the conclusion but not how it was reached.” “GAI makes students less patient when they come to tutoring, they don’t want to explain their thought process, they just want the answer.”
Brainstorming ideas	GAI can help get started. If you have a vague idea for a project or essay, AI can help generate more specific, concrete suggestions – including  possibilities you may not have thought of on your own.	“A very grayscale world can be made from the mass of people asking AI for inspiration and brainstorming, rather than having them think on their own or collaborate with their peers.” Sometimes, the creative thinking required to get started is the whole point.

 

 Want to learn more?

Research from the MIT Media Lab about the cognitive effects of using LLMs for writing tasks https://www.brainonllm.com

A student’s guide to AI – some evidence-based ideas for how to (and how not to) use GAI for learning https://studentguidetoai.org

The METR research group evaluates various dimensions of GAI tools https://metr.org

Some of METR’s findings are surprising, including a study that showed using GAI slowed down users engaged in coding tasks https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

“Stochastic Parrots” is a term that describes how LLMs produce language. This article points toward the environmental concerns of LLMs as well as implications for our information ecosystem. https://dl.acm.org/doi/10.1145/3442188.3445922

More on LLMs’ persistent tendency to “hallucinate”, to “lie”, or, “BS” https://www2.csudh.edu/ccauthen/576f12/frankfurt__harry_-_on_bullshit.pdf

https://link.springer.com/article/10.1007/s10676-024-09775-5

In this small study, GAI use was correlated with students performing about the same on homework, but far worse on exams. https://research-ebsco-com.proxy.library.cornell.edu/linkprocessor/plink?id=faa23ed8-b599-36b6-ac04-f43da359ffef

Op-ed about teaching writing with AI, concerns about students developing the capacity for critical thinking. https://www.chronicle.com/article/im-an-ai-power-user-it-has-no-place-in-the-classroom