Effects of Generative Artificial Intelligence on K-12 and Higher Education Students’ Learning Outcomes: A Meta-Analysis

Jul 6, 2026

Upload by Logic

Generative artificial intelligence (GenAI) offers educational innovation, yet its influence on students’ learning outcomes remains debated. This study evaluates GenAI’s effects on K-12 and higher education students and identifies moderators through a meta-analysis of 49 articles. Mean effect sizes for learning achievement (0.857) and learning motivation (0.803) are positive overall, while impact varies by education level, subject classification, interface, development, interaction approach, and experimentation time. GenAI benefits higher education more, works better through text interaction than mixed media, and motivational gains diminish over time.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages43

Reading Time1 h 48 min

Abstract

Introduction

Background and research gap

Meta-analysis approach and moderators

Literature Review

GenAI in education

Article Effects of Generative Artiﬁcial Intelligence on K-12 and Higher Education Students’Learning Outcomes: A Meta-Analysis Journal of Educational Computing Research 2025, Vol. 63(5) 1249–1291 © The Author(s) 2025 Article reuse guidelines: [sagepub.com/journals-permissions](sagepub.com/journals-permissions) [DOI: 10.1177/07356331251329185](DOI: 10.1177/07356331251329185)[ ](DOI: 10.1177/07356331251329185)[journals.sagepub.com/home/jec](journals.sagepub.com/home/jec) Xiaohong Liu 1 􀀁, Baoxin Guo1, Wei He 1, and Xiaoyong Hu 1 Abstract Generative artiﬁcial intelligence (GenAI) has signiﬁcant potential for educational innovation, although its impact on students’ learning outcomes remains controversial. This study aimed to examine the impact of GenAI on the learning outcomes of K-12 and higher education students, and explore the moderating factors inﬂuencing this impact. A meta-analysis of 49 articles showed that the mean effect sizes of GenAIon students’ learning achievement and learning motivation were 0.857 and 0.803, respectively, indicating a positive impact of GenAI on education. However, this effect varied according to moderators, including education level, subject classiﬁcation, GenAI interface, GenAI development, interaction approaches, and experimentation time, which enhanced the impact of GenAI on education. Speciﬁcally, GenAI had a greater impact on the academic performance of higher education students, and students interacted more effectively with GenAI using text than with mixed media, such as images or audio. Although GenAI has a novel effect on students’ learning motivation, the effect size decreases over time. These ﬁndings provide empirical support for the beneﬁcial effects of GenAI on education and offer insights for optimizing its use in teaching practices. 1 South China Normal University, China Corresponding Author: Xiaoyong Hu, Institute of Artiﬁcial Intelligence in Education, South China Normal University, 55 Zhongshan Avenue West, Tianhe District, Guangzhou 510631, China. Email: [huxiaoy@m.scnu.edu.cn](huxiaoy@m.scnu.edu.cn) Keywords generative AI, learning outcomes, learning achievement, learning motivation, metaanalysis Introduction In recent years, there have been rapid advancements in the development of artiﬁcial intelligence (AI) . Notably, the launch of ChatGPT on November 30, 2022, marked asigniﬁcant milestone in the history of AI development (Alier et al., 2024) . ChatGPT’s adaptability, with its ability to learn from both structured and unstructured data, makes it a highly versatile conversational AI tool, and has thus been attracting increasing attention for its educational applications (Jauhiainen & Guerra, 2023) . With the widespread use of generative artiﬁcial intelligence (GenAI) products in education, tools such as ChatGPT show great potential in teaching practice, while also fostering in-depth exploration of educational models and methodologies within the academic community. GenAI has implications for both teaching and assessment (Bower et al., 2024), prompting researchers to consider teaching strategies that integrate GenAI (Kong & Yang, 2024) . Researchers are increasingly examining the impact of GenAI applications in speciﬁc domains and environments. For instance, Baha et al. (2023) found that chatbots can signiﬁcantly enhance the learning experience for middle school students. However, Qureshi (2023) observed that sophomore students ’ performance was hindered by inaccuracies and inconsistencies in code submissions created with ChatGPT in programming contexts. Kosar et al. (2024) reported that ChatGPT use did not affect the programming performance of ﬁrst-year undergraduate students. In other words, there is no consensus on the effectiveness of generative AI in education. Meta-analyses, which aggregate data from multiple studies, enable researchers to measure overall effect sizes, providing more robust and persuasive insights into outcomes. Sun and Zho

FAQs

What overall impact does GenAI have on students’ learning outcomes?

The meta-analysis finds a positive overall effect, with mean effect sizes of 0.857 for learning achievement and 0.803 for learning motivation.

Which factors moderate the effect of GenAI in education?

The study reports moderation by education level, subject classification, GenAI interface, GenAI development, interaction approach, and experimentation time.

How does GenAI’s effect differ between higher education and K-12, and what interaction style works best?

GenAI has a greater impact on higher education students’ academic performance. Students interact more effectively with GenAI using text than with mixed media such as images or audio.

Jul 6, 2026

Upload by Logic