by Faith Lovell
In 2022, OpenAI released ChatGPT to the public. This platform made the use of AI significantly more accessible to the general public and forced many education systems to rapidly adapt their policies to discourage cheating with this technology. Since its release a little over a year ago, there have been many debates on the effect this has had on students in the classroom and their ability to use the tool academically… but what about its effect on the educator side of the classroom?
Although there are many guidelines on students’ use of AI tools such as ChatGPT, there are next to no standards or regulations for educators and their use of such tools. Despite the hesitation to allow students the ability to use AI technology, it has been embraced in educational tools, student feedback, and even grading.
Educators and educational technology developers appear to be praising this new technology, but there are significant flaws in their starry-eyed support of AI in education. The argument in favor of using AI widely in education, particularly in previously personalized aspects such as assignment feedback, could damage the personal connection between educators and their students.
Why Use AI at All?
There are many positives to integrating AI into some aspects of education, and it’s not hard to see why there has been support for the use of these tools. Companies such as Feedback Fruits, a peer review tool, use AI to analyze students’ peer reviews and give them feedback on the quality of their responses. Even here at UD, the Computer & Information Sciences department is experimenting with using AI to auto-grade students’ worksheet submissions. These solutions provide near-immediate feedback to students that they may not have access to if a human had provided the feedback.
Quick and Easy
The speed and convenience that this can be done, especially using OpenAI’s accessible (and in many cases, reasonably cheap) GPT models, provides a unique opportunity to quickly provide students with suggestions for improvement. If the AI feedback model suggests improvement, students have the potential to learn and correct their mistakes on their own, allowing them to improve on their own time without having to wait for a human grade or time to inquire about feedback from their educator.
In some cases, the quality of feedback may even be better through AI. Larger classrooms limit the amount of feedback educators are able to provide for students, as it takes a substantial amount of time and can be fatiguing. A paragraph of suggestions or critiques on every assignment for hundreds of students is near-impossible in most educational settings and a pitfall that AI is able to remediate.
If students are able to quickly receive feedback, this also saves educators time. For larger classes, where it is difficult to provide extensive feedback without hours of extra work, educators can focus more on other important aspects of their classroom without focusing on grades or critique. The education field is often overworked and understaffed, and using AI as a supplement to instructor duties could lighten the workload.
Leveling-Up Analytics
AI has not just been used to “reinvent” long-existing practices such as feedback, as researchers begin to get creative with the use of AI and student data.
Often, educators struggle to keep up with the individual progress and needs of each student. Although they may have an idea, and know their students well, it is difficult to keep a mental note of what students are struggling with and how much- or little- students are improving in the classroom.
Using assignment and grade data, there has been research on using AI to present educators with summaries and analyses of the performance of their students. Having statistics and a “snapshot” of student performance aided by AI could be revolutionary for instructors to better personalize and evaluate the needs of their students and the classroom.
What’s Wrong With Using AI, Then?
Although the benefits of AI in the classroom are promising, a lot of these tools and systems are being viewed with rose-colored glasses. Accessible AI technology is still extremely experimental, and missing a critical pillar of education: human connection.
Out of Touch
A student-teacher connection is arguably one of the most critical aspects of the classroom. Building a relationship and level of trust with students helps them feel more connected to the classroom, more interested in learning, and more comfortable asking questions. Almost everyone has a teacher who made a significant impact in their lives; can you recall yours?
Maybe it was a high school teacher who intervened with a student at the right time before they got too off track with their grades and their education. It could have been a Biology teacher who taught so well that a student was inspired to pursue Biology for their career. Many students are mentored by their collegiate professors who open doors for future opportunities. Personal connections are bound to make a profound difference in the education of students.
No matter how much data we provide, or how intelligent AI becomes, no machine can replicate these connections.
Although AI can be supplemental, the teacher-student relationship is deemed crucial to student success in the classroom. If students are not receiving feedback from their instructors, and instead receive this feedback from AI, they are losing an important avenue to discuss their performance and work in the classroom. Instructors, on the other hand, are missing out on an opportunity to analyze their students’ work and gain a deeper understanding of what their students are working on.
Suppose AI is held on a pedestal to supplement education, and we continue to see it as an all-encompassing and revolutionary avenue for feedback. In that case, there is real potential for personal connections in the classroom to be limited. As cool as this new technology is, there is no way to circumvent the impact that human connection has on students.
“Saving Time”
Many educational AI tools are still in highly experimental stages, and while the immediate benefit of saving time is enticing, how much time are educators really saving?
In an analysis of modern educational systems using AI by MIT, there are concerns that the difficulty of implementing and testing AI systems may be more time-consuming than its conventional predecessors.
One concern is that customizing AI feedback per student and per assignment would be very laborious not only for developers but for educators as well. If students struggle to understand their feedback, educators have to take the time to learn both the concerns of the students and interpret the feedback from AI tools before being able to assist them.
The biggest, and most severe concern, is the unreliability of experimental AI. We are still at a point in AI education tools where there are bound to be many errors or inaccuracies in any AI implementation; these errors will be at the expense of quality education.
AI feedback and analysis, in its current form, is significantly inaccurate and inconsistent across students. In any scenario where there is an error or inconsistency in AI feedback, an instructor will have to intervene and likely have to review most, if not all, of students’ submissions to discover these errors. Time saved will likely become time wasted attempting to resolve these issues.
Students, instead of reading their feedback to assist them with their assignments, may have to read the feedback and question its accuracy. It may even cause confusion between students if feedback is inconsistent, something that could be extremely problematic for educators and classroom management.
Not Quite There Yet
There is no argument that using AI in education could be revolutionary, but it is clear that we are not at that level of technology yet. Not only does the technology itself have a long way to go, but educators’ approaches to using AI without sacrificing important aspects of education do as well.
Integrating AI technology that we have access to now into the current education system is like walking a very thin and fragile tightrope; any missteps are at the expense of students, so it is essential that educators are cautious when approaching what as seen as a “magical” tool.
The topic of AI in education is very touchy. While I do recognize its merit in idea generation and saving time from mundane tasks, I think the risks outweigh the benefits. Cheating is just too easy with it, and with the extreme advancements made in such a short time, it has become almost undetectable when AI has been used. I'm not sure how to stop that, and perhaps it's too late, but I think it's important for educators to be aware of all of its capabilities so that assignments can be adjusted to better assess understanding of material.
-Clara Kelly
You represented both sides of the AI education agruments very well. I agree that it can be an extremely useful tool, but I feel like it takes away the authenticity and personability of learning in a class room with a teachers own made lesson plan.
Very interesting read! I agree that AI can be a useful tool for students, but it can also be a detriment if they develop too much of a reliance on it. I also believe that the technology isn't developed enough yet to be able to match human quality (like in creative writing).
-Daniel Rubin
This is a very relevant topic! You present some good points for both sides of the argument. The "snapshot" of student's strengths and weaknesses seems especially beneficial. However, I agree with your point about how important the human element of teaching is. Especially because today's AI resources are so new: as you mentioned, there is a potential for glitches. As we learn more about AI and its effects within society, we can assess how to best implement it.
Olivia Wayson