AI Does Not Automatically Make Students More Critical Thinkers
The emergence of GenAI has transformed how students search for information, develop ideas, write, and complete academic assignments. Technologies such as ChatGPT can provide rapid explanations, offer alternative answers, provide feedback, and support more personalized learning.
However, this convenience also creates risks. Students may accept AI-generated answers without checking sources, verifying information, or evaluating the arguments presented. Therefore, AI use does not automatically improve or reduce critical thinking skills.
The researchers distinguish between two patterns of GenAI use. First, AI can function as a cognitive scaffold, serving as a tool that supports students’ thinking processes through comparing perspectives, examining arguments, identifying weaknesses in answers, and providing feedback.
Second, AI can encourage cognitive offloading, in which students transfer processes such as remembering, analyzing, evaluating, or making decisions to the technology. Under these conditions, AI may replace thinking processes that students should perform themselves.
Inconsistencies in previous research also highlight the importance of this study. Some studies have found a positive relationship between AI use and critical thinking, while others indicate that dependence on instant answers can reduce independent thinking. Previous studies in Indonesia have also frequently relied on one-time measurements, making it difficult to capture changes in students’ abilities throughout a learning period.
Following 186 Students for One Semester
Agustina and her colleagues employed a quantitative approach with a longitudinal correlational design. Rather than receiving a specific experimental treatment, the students were observed over one semester to examine changes in critical thinking skills and their relationship with patterns of GenAI use.
The study population consisted of 1,238 active students from three higher education institutions in East Java during the 2025/2026 academic year. Of the 210 students who participated in the initial measurement, 186 completed all stages of the study and constituted the final sample. They were students from semesters 2 through 8 who had used GenAI for academic purposes.
Data were collected over 16 weeks. At the beginning of the semester, students completed a GenAI-use questionnaire and a critical-thinking test. Measurements were repeated during weeks 15 and 16.
The GenAI-use questionnaire assessed three aspects: usage intensity, reflective and evaluative use, and passive use and dependence. Meanwhile, critical thinking was assessed through case-based tests measuring interpretation, analysis, evaluation, inference, and explanation.
The data were analyzed using descriptive statistics, a paired t-test, Pearson correlation, and linear regression. Because the study used a correlational design, the findings demonstrate relationships between variables and cannot directly establish causal relationships.
Critical Thinking Scores Increased by 6.5 Points
The study found an improvement in students’ critical thinking skills over the course of one semester. The average score increased from 68.4 at the beginning of the semester to 74.9 at the end, representing a 6.5-point increase.
The largest improvement occurred in evaluation skills, which increased by 7.6 points, followed by analysis skills, which increased by 7.1 points. A total of 137 students, or 73.7 percent, experienced an increase in their critical-thinking scores. Twenty-eight students showed relatively stable changes, while 21 students experienced a decline.
Statistical testing indicated that the increase was statistically significant. The p-value was below 0.001, with a Cohen’s d effect size of 0.79. Nevertheless, the researchers did not conclude that the improvement was entirely caused by GenAI use because the study did not include a control group.
How Students Use AI Matters More Than How Often They Use It
The most important finding emerged when the researchers examined patterns of GenAI use.
Overall, GenAI use had a positive relationship with the development of students’ critical thinking, with a correlation of r = 0.34. However, the relationship was stronger when students used AI reflectively and evaluatively, with r = 0.46; p < 0.001.
In contrast, passive and dependent AI use showed a negative relationship with the development of critical thinking, at r = −0.30; p < 0.001. Usage intensity alone had only a weak positive relationship, at r = 0.19.
Regression analysis reinforced this pattern. Reflective and evaluative use was the strongest positive predictor, with β = 0.38 and p < 0.001. Conversely, passive and dependent use was a negative predictor, with β = −0.22 and p = 0.001. After usage patterns were taken into account, GenAI usage intensity no longer made a significant contribution.
In other words, students who use AI more frequently do not necessarily experience better development in critical thinking skills. What matters more is what they do with AI-generated answers.
Students who verify sources, compare different arguments, evaluate AI outputs, identify errors, and subsequently revise their answers showed better development in critical thinking.
Agustina and her colleagues describe GenAI as a tool that can support thinking when students retain intellectual control over the process. Conversely, when students simply accept or copy AI-generated answers, analytical and evaluative processes may be transferred to the technology.
Implications for Universities and Lecturers
The findings provide an important message for higher education institutions. The issue is not simply whether students should be allowed to use AI, but how AI should be integrated into the learning process.
The researchers recommend that lecturers design assignments requiring students to verify sources, identify AI errors, compare answers, explain their revision processes, and take responsibility for their work.
Assessment models can also place greater emphasis on process rather than only on the final product. Case-based assignments, source critiques, presentations, prompt documentation, and reflective journals can help ensure that students remain actively engaged in the thinking process.
For students, the message from the study is equally simple: use AI as a thinking partner, not as a replacement for thinking. AI-generated answers still need to be checked because they may contain errors, biases, or information that does not correspond with reliable sources.
The Study Has Several Limitations
The researchers emphasize that the findings should be interpreted in light of several limitations. The study used a correlational design without a control group, meaning that improvements in critical thinking cannot be attributed entirely to GenAI use.
In addition, data concerning AI-use patterns were collected through questionnaires and may therefore be subject to perception bias. Measurements were conducted only twice, and respondents came from three higher education institutions in East Java.
Future research is recommended to use experimental or quasi-experimental designs, longer observation periods, more measurement points, and data from AI prompt histories and outputs. The researchers also recommend examining factors such as AI literacy, metacognition, motivation, trust in AI, and cognitive offloading.
Author Profiles
Their article focuses on the relationship between generative artificial intelligence use and the development of critical thinking skills among university students in higher education.
Research Source
Agustina, Dwi Otik Kurniawati, Boy Lumoindong, and Irwan Suanto. “Exploring the Relationship Between Generative Artificial Intelligence Usage and the Development of University Students’ Critical Thinking in Higher Education.” Formosa Journal of Science and Technology, Vol. 5, No. 8, 2026, pp. 2247–2264. DOI: 10.55927/fjst.v5i8.145.
0 Komentar