Artificial Intelligence in Education
How LLMs and Generative AI are changing the ways we work, teach, and learn.
At the end of 2022, generative AI and large language models (LLMs) entered the cultural zeitgeist as people began to take notice of OpenAI’s release of ChatGPT. The release was public, and anyone with a computer or mobile device with access to the internet could use it. Seemingly overnight, it opened possibilities for the ways we access and author information. Myriad generative AI tools are available for use today and versions update frequently, indicating the development of AI tools is on a certain forward trajectory into the future.
The availability and ease of use of these powerful AI tools have sparked both curiosity and concern for how they impact education from the perspectives of students, educators, and administration. Where do we draw the line on academic honesty with the use of AI? How will AI change our roles as instructors? How do we prepare students for a future AI-rich work environment? The use cases for ChatGPT and other generative AI tools in the classroom are nuanced and rich with possibility.
What Can It Do?
It is no secret that generative AI tools can be used to complete assignments, solve equations, and write essays. This might be comparable to the development of broad access to internet search engines, which deliver domain-specific information with a keystroke. The power comes in AI’s ability to edit, refine, collaborate, and assist in activities that involve higher-order thinking, not just simple tasks that we may have once relegated to rote memorization. That being said, how do we develop our skills to prompt these tools effectively? How do we turn them into teaching and learning assistants so that we might become better teachers and learners? How do we model these skills for our students?
We have more immediate access to information today than ever before in human history. This has been the trajectory of technological advancement for some time now and will continue for the foreseeable future. How do we foster a culture of curiosity in a world where information is everywhere?
The biggest concern is plagiarism. Artificial intelligence trained on large language models (LLMs) can return writing in the style of well-known authors. It can regurgitate passages of published writing. It can easily write in narrative, expository, and persuasive writing styles. Knowing this, anyone might be excused for entertaining the thought of using an AI tool as a writing assistant. Explain to students what constitutes plagiarism, and how that impacts the academic integrity of their work. Using AI as a writing assistant may become as commonplace as using your phone calculator to do basic arithmetic. None of us think of this as cheating in our day to day lives, and many of our assessments allow the use of calculators provided you show your work. Use this example to explain the difference between plagiarizing with AI and using it as an effective assistant. This is an important discussion to have early in the semester with students. Clarify your expectations for the use of generative AI up front in your syllabus. Whatever your course policy is, instruct students on how they should cite ChatGPT and other generative AI tools in their work, and clarify the protocol you will follow should a student deliberately misuse the tool for classwork. See the UAF Office of Rights, Compliance and Accountability for information on our institution’s Academic Misconduct Policy.
Generative AI has a number of known shortcomings that users need to be aware of.
- The companies behind generative AI tools train them on large language models (LLMs) that contain some content that infringes on copyright and is used without consent.
- Generative AIs replicate the stereotypes and biases contained in the large language modules (LLMs) they are trained on.
- Generative AIs do not always comply with FERPA and HIPPA policies. Furthermore, some systems do not enable users the option to opt-out of having their own work used to train the model. Honorlock is one of these systems.
Students can benefit from learning to use generative AI in many aspects of their lives, and knowing the limitations of these tools is critical. Discuss how generative AI can return incorrect information and can even “hallucinate” false information. Help them develop their critical thinking skills to enable them to more easily identify gaps in how generative AI authors information. Knowing the pitfalls of these tools and seeing examples of where they can fail demonstrates the importance of fact-checking generated results. Do students consider their use of chatGPT an acknowledgment of their endorsement of the generated result? This is a great conversation starter.
A few ideas on how students might use generative AI in class include but are not limited to:
- polishing language
- generating ideas for article titles
- manipulating spreadsheets and data
- solving equations
- identifying blindspots in their thinking
- discovering opposing points of view
Revisiting the learning outcomes in your course may help you determine if and how generative AI is appropriate for students in your class. What are your expectations for student writing in your course? Do you want very polished work or do you get excited when you encounter typos and a unique voice? Communicate your expectations with your students. In some classes, writing is intrinsically linked to the learning outcomes of the course. In others, writing may be just one way students can demonstrate their mastery of concepts and skills. In these classes, using generative AI for proofreading and editing allows students to focus on the content and less on the mechanics of writing. Doing this also makes the act of providing feedback to them easier for faculty, for the same reasons. Additionally, generative AI can also be considered an assistive technology that helps reduce the anxiety of students for which English is a second language, and for some neurodivergent learners as well.
The UAF library has a resource that introduces generative AI students to get them started and thinking about what it is and what it can do.
If you are teaching, generative AI can also be a tool in your arsenal. Use it as an assistant to
- Iterate on a syllabus – To explore this process, provide ChatGPT outcomes for your course. Give it your course description. Provide some constraints such as the length of the semester, target language, and any industry standards the course might also follow. Edit ChatGPT’s recommendations based on gaps you identify as a subject matter expert, and adjust them to your teaching style.
- Rethink assignment design – Do your students need to master the mechanics of writing to achieve the stated outcomes of your course? Are there opportunities allowing them to demonstrate evidence of their learning allowing generative AI in as an editor or collaborator?
- Capture student imagination – Give them voice and choice by involving them in the decision to use chatGPT for particular assignments.
- Create an example of an assignment artifact – If you don’t yet have an example of a completed assignment to show your students, ask ChatGPT to create one.
- Offload the task of troubleshooting – Do you spend more time fielding questions related to code syntax or formatting than you can afford? Show students how to leverage generative AI as a teaching assistant.
- Summarize the main points of a lesson – ChatGPT is very good at rephrasing, reframing, and reducing information into different formats. Students may benefit from a CliffsNotes style summary of content that you have already provided, where the summary can be tailored to meet their own remaining questions about the material. This functionality is already being built into some learning platforms, including on edX, which provides automated unit summaries using ChatGPT for learners.
These are all ideas and questions meant to inspire conversation and spark debate about how we all might use generative AI to facilitate learning. We want to be part of them too. Join us for upcoming workshops to explore and share ideas. As educators, we not only prepare students to become part of the workforce. We engage in the act of developing critical thinkers and lifelong learners. Our aim is to leverage these tools to improve teaching practices, enhance student learning experiences, and foster the development of AI-literacy skills. By doing so, we equip our students to become future leaders in a society driven by technology.
Our Latest Articles on AI
Join us, from 10:00am until noon, 11 January, 2024, for a semester-start discussion on generative artificial intelligence. This session will offer an overview of current efforts at UAF to integrate AI with teaching and learning, as well as share resources available to faculty and your students. We will look at example assignments that help introduce…
Join us 30 November 2023 from 1000-1200. In this workshop, participants will explore what generative AI can and cannot do when it comes to the writing process, emphasizing the human skill of revision while incorporating use of AI to help with stuck-points. This workshop will focus on research writing and is a collaboration between the…
Generative AI is disrupting normal practices in education, including one of the most often used forms of assessment, the essay. Approaches offered here include using other forms of assessment or embracing AI by modeling good behavior for your students. This tip builds on other ideas published in July.
This workshop was delivered Tuesday, 19 September 2023, from 1-3pm. Generative AI (ex: ChatGPT) can produce writing in multiple forms/genres and on almost any topic you might ask for. In this workshop, participants will explore what generative AI can and cannot do when it comes to the writing process, emphasizing the human skill of revision…
This workshop was delivered 7-8 September 2023, and is a repeat of the initial workshop titled the same. How do we design courses and speak with students about academic dishonesty in the era of artificial intelligence? What kind of work might AI’s be helpful in completing? These discussion-based workshops expect participants to have experience working…
These are workshop materials for the session. “Teaching in the Era of Artificial Intelligence”, delivered 17-18 August 2023.
Large Language Model Artificial Intelligence technology like ChatGPT is disrupting traditional methods of assessment, especially as it relates to grading via the essay. This teaching tip offers possible solutions, including the incorporation of a revision cycle with reflection.
Fed by massive amounts of data scanned from the Internet, the ChatGPT AI does a remarkable job at producing conversational text and even structured code in response to prompts. The amazing thing is that you can “ask” chatGPT questions in a conversational tone (with text), and it will respond in kind. The answers are not…