Artificial Intelligence, Learning and Social Engagement among Undergraduate Students: A PRISMA-Guided Systematic Review with Narrative Thematic Synthesis
This PRISMA-guided systematic review of 29 studies synthesizes evidence showing that while AI tools like intelligent tutoring systems and GenAI chatbots can enhance undergraduate learning and social engagement, their effectiveness is highly contingent on pedagogical design, learner self-regulation, and institutional context, and is accompanied by significant risks to critical thinking and equity.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the classroom of the future not as a room with rows of desks, but as a bustling digital marketplace where students are constantly juggling their homework, their friendships, and their own motivation. In this marketplace, a new kind of helper has arrived: Artificial Intelligence (AI). Think of AI as a super-smart, always-available tutor that can explain a tricky math problem, chat with you about your day, or help you draft an essay. But here's the big question that scientists are asking: Is this robot helper actually making students smarter and happier, or is it just a fancy toy that might accidentally make them lazy or lonely?
To understand the answer, we need to look at two main ideas. First, there's "engagement." Imagine engagement as the fuel in a car. It's not just about sitting in the driver's seat (behavioral); it's about feeling excited to drive (emotional) and actually knowing how to steer and navigate (cognitive). If a student has high engagement, they are fully in the game. Second, there's "social presence," which is the feeling that you aren't alone in the room. Even if you are studying from your bedroom, social presence is that warm feeling that your teacher and classmates are right there with you, listening and caring. The paper we are about to explore dives deep into whether AI tools are the perfect fuel for this car and the best way to keep everyone feeling connected, or if they might actually sputter the engine and leave students feeling isolated.
This paper is a massive detective story called a "systematic review." Instead of running a single experiment, the authors, Menyene-Abasi Andem Udo and their team, acted like literary detectives. They hunted down 29 different studies from around the world—ranging from smart tutoring programs to the new, trendy chatbots like ChatGPT—to see what the whole picture looks like. They didn't just count numbers; they looked for patterns in how these tools changed the way students learn and interact.
Here is what they found, and it's a bit more complicated than a simple "good" or "bad" story.
First, regarding learning, the paper draws a clear line between two types of AI. The older, more established type is the "Intelligent Tutoring System" (ITS). Think of these as specialized robots built just to teach one subject, like a math coach that never gets tired. The evidence here is rock-solid: three different major studies found that these tutors actually boost test scores significantly, moving a typical student from the middle of the pack to the top quarter. It's like having a personal trainer who knows exactly which muscles you need to strengthen.
However, the newer, flashier tools—Generative AI (GenAI) like ChatGPT—are a different beast. The paper suggests that these tools are a bit of a mixed bag. They can definitely help students get things done faster and feel more motivated, but the results depend entirely on how they are used. If a student just asks the AI to write their essay and hits "send," they might get a good grade but lose their ability to think critically. It's like using a calculator for simple math; it's efficient, but if you rely on it for everything, you forget how to do the math yourself. The paper explicitly rules out the idea that GenAI is automatically a magic wand for learning; instead, it suggests the benefits are "contingent," meaning they only happen if the teacher sets up the right rules and the student stays in control.
Second, the paper tackles the social side. Can a robot be a friend? The authors found that AI chatbots can act like a "digital buddy" that helps reduce the feeling of isolation, especially for students taking classes online. They can offer a quick "hello" or help you set goals, which is better than nothing. But the paper is very clear on one point: a chatbot cannot replace a real human friend or teacher. If a student starts talking to a bot instead of their classmates, they might actually feel more lonely. The AI is a supplement, like a side dish, not the main course of human connection.
Finally, the paper introduces a new way of thinking called the "AI-Mediated Engagement Contingency Model." Imagine this as a three-legged stool. For AI to work well, three legs must be strong:
- The Task: The teacher must design the lesson so the AI is used as a tool for thinking, not a shortcut for doing the work.
- The Student: The student must have the self-control to use the tool without letting it do all the thinking for them.
- The School: The school must have the internet, the rules, and the training to support the technology safely.
The paper argues that if even one of these legs is weak, the whole stool falls over. For example, even if a school has great internet (a strong leg), if the students are just using the AI to cheat or if the teacher hasn't designed a good lesson, the AI won't help them learn. This is especially important for schools in places with fewer resources, like parts of Africa and Latin America, where the "legs" of the stool might be wobbly due to poor internet or a lack of training.
In short, the paper suggests that AI is a powerful tool, but it is not a miracle cure. It works best when it is part of a well-planned lesson, used by students who are in charge of their own learning, and supported by schools that are ready for it. If we treat AI as a magic wand that fixes everything on its own, we might end up with students who are faster but less thoughtful, and more connected to robots but less connected to each other. The future of learning isn't about replacing humans with machines; it's about finding the perfect balance where machines help humans shine.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.