AI evidence extraction

Portuguese secondary school students’ perceptions regarding the use of ChatGPT

Ana Callado, Isabel Saúde, José Luís Araújo · Contemporary Educational Technology · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

6/10

Relevance

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.30935/cedtech/17516

Methodology & findings

Study design

Cross-sectional survey study with 114 secondary school students (aged 16-17) using a 15-item questionnaire containing closed-ended and open-ended items covering five dimensions: knowledge of AI and ChatGPT, use of ChatGPT, perceived accuracy of ChatGPT-generated outputs, potential of ChatGPT for teaching, and ethics and academic integrity.

Sample

N = 114, 2 groups

Primary method

Descriptive analysis for closed-ended items; content analysis for open-ended responses

Main result

The study found that "although most students are familiar with AI and ChatGPT, their demonstrated knowledge is largely utilitarian and superficial." Additionally, "Students predominantly use ChatGPT for schoolwork and attribute high reliability to its outputs, often without the necessary critical evaluation," and "weaknesses in their ethical understanding, particularly regarding academic integrity and plagiarism" were identified.

Reports effect sizes.

Research paradigm

Mixed methods (quantitative descriptive and qualitative content analysis)

Author conclusions

The authors conclude that "Education for the ethical and responsible use of these technologies proves essential to prepare young people for future challenges," and they "argue for critical AI literacy and teacher professional development to support pedagogically grounded and ethical integration of GenAI."

Risk of bias

Selection bias: Single school sample may not represent broader Portuguese secondary school population; Sample size: 114 students is a relatively small sample for generalization; Self-report bias: Questionnaire-based responses may not reflect actual behavior; Cross-sectional design: No temporal assessment of changes in perceptions; Convenience sampling: Non-random recruitment may introduce selection bias; Limited sample size (n=114) reduces statistical power for subgroup analyses; Potential self-selection bias in survey participation; Cross-sectional design limits causal inference

Limitations

The paper states it surveyed "114 students from one school in 2024," which represents a limited sample from a single institution, restricting generalizability
The authors argue for "critical AI literacy and teacher professional development" as necessary interventions, implying limitations in current educational approaches to addressing these gaps.

Open questions raised

The authors identify a need for critical AI literacy and teacher professional development to support pedagogically grounded and ethical integration of generative AI in education.

Data: not_statedCode: not_statedExtracted from: pdf

Explore related topics

AI Literacy Training Doctoral Training Survey