In the age of artificial intelligence (AI), the rapid development of generative AI tools such as ChatGPT presents both unprecedented opportunities and challenges in education. These tools have the potential to fundamentally improve learning experiences and outcomes. However, their responsible use requires a robust set of critical thinking skills. This paper addresses the urgent need for a structured approach to analysing and assessing these essential skills in the context of using generative AI tools.

We propose a comprehensive framework designed to assess and enhance the critical thinking skills necessary for the responsible use of ChatGPT. This framework builds on established educational theories and practices, integrating Paul-Elder's model of critical thinking (2001) and results from other recent studies with a focus on generative AI (e.g. Hsiao et al., 2023; Essel et al., 2024).

The proposed framework consists of 4 dimensions. Each dimension is further broken down into specific competencies that are critical for responsible engagement with generative AI:
1) Analytical skills: focuses on the ability to interpret and understand the outputs of ChatGPT, recognising patterns, biases, and underlying assumptions,
2) Evaluative skills: assessing the credibility, relevance, and reliability of the information generated by the AI, evaluation of Pro- and Contra Arguments of AI-generated (written) Outputs, Interpretation and Comparison of AI-generated (graphical) Outputs,
3) Ethical reasoning skills: enable individuals to process information, draw logical conclusions, and make informed decisions. These skills involve a variety of mental processes including deductive reasoning and pattern recognition, inductive reasoning: formulation of hypotheses and conclusion drawing, justifying decisions in ethical dilemmas,
4) Metacognition: encompasses self-awareness and regulation of one's own cognitive processes while interacting with AI tools, reflection on personal biases, reflection on ChatGPT-usage.

To validate this framework, we conducted a mixed methods study involving students, educators and AI experts. Quantitative data were collected through task-based assessments designed as an experiment. The sample consists of 40 students at bachelor level in higher education (20 EG, 20 CG). After task instruction, the experimental group has to solve typical academic tasks focused on the 4 dimensions using ChatGPT. The control group could use internet resources, but ChatGPT was not allowed. After each task, each student in the EG has to reflect briefly on their task (division of labour: estimated ratio between AI and own contribution, comments). Qualitative data will be obtained from student interviews, providing in-depth insights into the experiences and perceptions of users interacting with ChatGPT. The results of the developed tasks are evaluated and scored by three reviewers.

The proposed framework could serve not just as a guide for developing critical thinking skills, but also as a benchmark for assessing an individual's readiness to engage responsibly with generative AI technologies. Developing these skills can help integrate AI into education to support intellectual growth, ethical behaviour and social well-being. Future research should focus on refining this framework and exploring its application in different educational contexts.

Keywords: Artificial Intelligence, Generative AI, critical thinking skills.