Category: Education · Psychology · Artificial Intelligence
Generative artificial intelligence (AI) is transforming how students learn, research, and solve problems. Tools such as ChatGPT can explain difficult concepts, summarize research papers, generate examples, and provide personalized learning support. But an important question remains: Does using generative AI actually improve students’ critical thinking, or does it encourage them to rely on ready-made answers?
A study published on October 24, 2026, in a journal associated with Dove Medical Press examined 431 health informatics students in Saudi Arabia to investigate the relationship between generative AI use and higher-order cognitive performance. Its cross-sectional research design provides a snapshot of the relationship between AI use and student cognition, rather than proof that AI directly causes improvements in critical thinking.
The findings are relevant to educators, students, researchers, and policymakers seeking to understand how AI can support meaningful learning without weakening independent reasoning.
What Did the Study Discover?
The study, titled Generative Artificial Intelligence Use and Applied Higher-Order Cognitive Performance Among Health Informatics Students: A Multicenter Cross-Sectional Case-Based Assessment Study, was published on October 24, 2026, in Advances in Medical Education and Practice. It involved 431 undergraduate students from seven universities in Saudi Arabia.
The researchers examined whether students’ patterns of generative AI use were associated with their ability to think critically, solve problems, and make decisions.
Key findings at a glance
431 students studied
Participants came from seven Saudi universities.
No independent association with routine use
Frequency, duration, and weekly hours of AI use were not independently associated with assessment performance after adjustment for university.
Academic background mattered
Performance was associated with factors such as university, academic year, and GPA in some analyses.
More research is needed
The observational design cannot establish whether AI improves or harms critical thinking.
Source: Alhur, Al-Kahtani and Mahgoub, 2026.
An important distinction is that the researchers assessed students using case-based questions designed to measure higher-order thinking, rather than relying only on students’ opinions about whether AI helped them learn.
Can Generative AI Improve Critical Thinking?
Generative AI can support critical thinking when students use it as a learning partner rather than an answer generator.
For example, a student studying a complex medical informatics problem might ask AI to explain several possible solutions, identify weaknesses in each approach, and suggest evidence that could support a conclusion. The student can then verify the information, compare alternatives, and make an independent decision.
This process encourages analysis and evaluation. However, if the student simply copies the generated answer without checking its accuracy, the opportunity to develop those skills may be reduced.
The technology itself does not guarantee better reasoning. Its educational value depends partly on the task, the student’s approach, the quality of the AI output, and the learning environment. The published study is consistent with this distinction, although it did not directly establish which specific learning behaviors improve critical thinking.
The Risk of Depending Too Much on AI
One concern surrounding AI-assisted education is cognitive offloading—the practice of delegating mental tasks to external tools. This can be useful because it frees time and attention for more demanding activities. However, excessive reliance may reduce opportunities to practise independent reasoning.
Students who regularly ask AI to summarize textbooks, write assignments, or solve problems may complete tasks more quickly. Yet task completion and genuine learning are not necessarily the same thing.
There are several potential risks:
- Reduced independent problem-solving: Students may become less accustomed to working through difficult questions themselves.
- Uncritical acceptance of information: AI-generated answers can contain errors, unsupported claims, or fabricated references.
- Superficial learning: Reading a clear explanation can create a feeling of understanding without demonstrating the ability to apply the concept independently.
- Academic integrity concerns: Submitting generated work without following institutional rules can undermine assessment and learning.
These are plausible educational risks, not proof that every student who uses AI will experience weaker critical thinking.
How Students Can Use AI to Strengthen Their Thinking
Attempt the problem first
Try to solve the question independently before asking AI for help. This gives you a chance to identify gaps in your understanding.
Challenge the AI answer
Ask for alternative explanations, possible counterarguments, assumptions, and limitations. Do not assume that a confident answer is necessarily correct.
Verify important claims
Check facts against textbooks, peer-reviewed research, official documentation, and other reliable sources appropriate to the subject.
Practise without AI
Regularly complete practice questions, explain concepts from memory, and test yourself without assistance to check whether you can apply what you have learned.
What Should Teachers and Universities Do?
Educational institutions need to move beyond the simple question of whether students should be allowed to use AI. They should also consider how assignments and assessments can encourage meaningful learning.
Teachers can ask students to explain the reasoning behind their answers, compare competing solutions, identify weaknesses in AI-generated content, and document how they verified important claims. Oral examinations, supervised case studies, practical exercises, and reflective assignments can also help educators evaluate students’ independent understanding.
AI literacy should include more than learning how to write effective prompts. Students need to understand the limitations of AI systems, recognize unreliable information, protect sensitive data, acknowledge AI assistance where required, and take responsibility for their own conclusions.
These recommendations are educational implications rather than interventions tested by the Saudi study itself.
What Are the Limitations of the Research?
The findings must be interpreted carefully. The research was cross-sectional, meaning that it measured AI-use patterns and assessment performance during a particular period rather than following students over time.
The sample was also based on convenience sampling, and AI-use information was self-reported. Although students were instructed to complete the assessment independently, researchers could not verify that all responses were completed without outside assistance. In addition, the assessment instrument was newly developed and requires further external validation.
Consequently, the study cannot establish that AI use causes either improvement or deterioration in critical thinking. It also cannot establish that formal AI training harms performance simply because the trained students scored lower in the adjusted analysis. Other unmeasured differences may help explain that association.
Future research could follow students across several semesters, compare structured AI-supported teaching with conventional teaching, and use supervised assessments to measure changes in independent reasoning.
Conclusion
Generative AI offers valuable opportunities to make education more interactive, accessible, and personalized. Nevertheless, the latest Saudi study suggests that simply using AI more frequently is not independently associated with better performance on a case-based assessment of critical thinking, problem-solving, and decision-making.
The most useful question is therefore not merely whether students use AI, but whether they continue to question, verify, analyse, and understand the information it provides. Students who combine AI assistance with independent practice and critical evaluation may be better positioned to use these technologies responsibly.
More rigorous longitudinal and experimental research is needed before firm conclusions can be drawn about AI’s long-term effects on student cognition.
