Question-answering systems facilitate adaptive learning and respond to student queries, making education more responsive. Despite that, challenges such as natural language understanding and context management complicate their widespread adoption, where Large Language Models (LLMs) offer a promising solution. However, existing research is predominantly focused on English, proprietary models, and often limited to a single question type, subject, or skill, leaving a gap in understanding LLMs' performance in languages like Brazilian Portuguese and across questions of various characteristics. This study investigates how LLMs could be integrated in an educational question-answering system efficiently to answer different question types (multiple-choice, cloze, open-ended), subjects (mathematics and Portuguese language), and skills (summation/subtraction, multiplication, interpretation, and grammar), evaluating answers by GPT-4 - the main LLM at the time of writing - and Sabiá - the open-source Brazilian Portuguese LLM - based on grades assigned by two experienced teachers. Overall, both LLMs demonstrated strong overall performance, with mean scores close to 9.8 out of 10. However, specific challenges emerged, with distinct strengths and weaknesses observed for each model, such as GPT-4's error in a multiple-choice subtraction question and Sabiá's misinterpretation of a cloze question.

Cognitive testing is concerned with quantitative and qualitative evaluation of an individual's intellectual functioning in its broad sense. Tests for evaluating cognitive components are based on submitting the subject to a given task and then assessing performance according to an established set of reference parameters. In order to analyze subject's behavior and test results, a software tool was developed based upon digital ink technology, which permits the digitalization of the assessment procedure, from the undertaking of the test to the production of the assessment results. By doing so, the test procedure is computerized and its data is saved in InkML format and processed to analyze tasks previously defined by the evaluator. We have used this tool to develop a Computational Thinking test defined within the Cattell-Horn-Carroll CHC framework of intelligence. Further investigating the relationship between fluid intelligence and computational thinking allows a better understanding of the main set of cognitive skills which need to be developed by students and professionals that aim to work in this domain. Moreover, it is suggested that assessments, independent of their context, may benefit from using the InkML tool, mainly due to the richer set of information that can be collected.