SciELO - Scientific Electronic Library Online

 
vol.65 número3Another translation of the Bible in Afrikaans: What makes it different?The Old Testament in the 2020 Afrikaans translation: Some contentious issues índice de autoresíndice de assuntospesquisa de artigos
Home Pagelista alfabética de periódicos  

Serviços Personalizados

Journal

Artigo

Indicadores

    Links relacionados

    • Em processo de indexaçãoCitado por Google
    • Em processo de indexaçãoSimilares em Google

    Compartilhar


    Tydskrif vir Geesteswetenskappe

    versão On-line ISSN 2224-7912versão impressa ISSN 0041-4751

    Resumo

    DU PLESSIS, Elize. The potential impact of the generative pre-trained transformer (ChatGPT) and similar language models on online assessment. Tydskr. geesteswet. [online]. 2025, vol.65, n.3, pp.822-840. ISSN 2224-7912.  https://doi.org/10.17159/2224-7912/2025/v65n3a4.

    The rapid pace at which natural language processing (NLP) and artificial intelligence (AI) are developing provides educators with new and creative tools. The Generative Pre-Trained Transformer (ChatGPT), a language model that simulates human speech and conducts interactive dialogues, is one of them. As online assessment becomes more common, ChatGPT and related language models present not only exciting opportunities but also significant challenges. Online assessment is about to undergo a revolution owing to the integration of artificial intelligence (AI) and language models such as ChatGPT. Applications driven by AI make efficiency improvements possible, including personalised feedback and automatic grading. However, the use of AI-powered technologies, such as ChatGPT, in online learning raises questions about the validity, fairness and originality of the assessed information, especially in higher education. For example, ChatGPT is based on the OpenAI Chat-3 model, which can autofill photos, translate, build computer programmes and create narratives. GPT-4 is even more intelligent than GPT-3. It is not only capable of handling longer prompts and conversations but also less prone to factual errors. Nevertheless, GPT-3 responds quicker and is, unlike GPT-4, not constrained by hourly prompts. Since COVID-19, online assessment techniques have grown in popularity because they are a quick and easy way to evaluate students' knowledge, skills and competences. For this reason, online assessment could undergo a substantial revolution with the introduction of sophisticated language models such as OpenAI's ChatGPT. Despite constant debate about ChatGPT, real-world examples of its application in online tests are scant. Clarifying the evaluation process is essential since many students prioritise passing tests over embracing the year-long learning process, which creates a gap between the process and its result. This paper examines the potential of integrating ChatGPT into online assessments, its potential benefits for educators and educational institutions, as well as its potential drawbacks and provides practical examples to use. The purpose of this study necessitated document analysis as a research methodology. The possible effects of generative technology, such as ChatGPT, on teacher education were examined by means of an exploratory research methodology. The intention was to suggest teacher preparation programmes to improve assessment procedures by using internet resources. Furthermore, an interpretative paradigm emphasising the understanding of human perspectives was adopted in this research. To ensure the validity and legitimacy of the information, the researcher collected official data from ChatGPT's website and published research papers. The transferability of the study depends on the extent to which the conclusions can be applied to other groups or environments. Although she was fully aware of the possibility of transferability and the subjective and contextual aspects of reality, the researcher did not claim that the same findings would be achieved under all circumstances. Ethical clearance was obtained from the institution. This research suggests ways to overcome online obstacles using ChatGPT. It is feasible to turn the learning process into something more significant than merely rating student performance by utilising ChatGPT. ChatGPT has the potential to enhance the learning process immensely. Opportunities to create creative and inclusive teaching, learning, and evaluation are among the strategies of online assessments. Use your past knowledge and abilities to take on new tasks and encourage group learning. These advanced language models can improve assessment experiences for both educators and students and transform adaptive examinations into automated evaluations and feedback. Practical suggestions for the use of ChatGPT in online assessment, addressing gaps in the literature, include the following: Use ChatGPT to provide immediate, formative feedback on students' writing, coding, or problem-solving. Integrate ChatGPT as a tutor for academic writing tasks - idea generation, planning, and refinement - which is ideal for students with language barriers or limited access to writing support. Students can compare their answers with those generated by ChatGPT and critically evaluate them. However, prejudice, dependability, security, and privacy need to be addressed with great care. By working together, academics, educators, and legislators can maximise the advantages of language models while maintaining moral principles and guaranteeing the validity and fairness of online tests. ChatGPT has given higher education another chance to offer and assess inclusive, creative teaching and learning. In fact, it is universally recommended that educators, scholars and academic institutions use generative AI technology. They ought to investigate it, learn about its drawbacks, think about its possible applications in their fields of study or instruction, and discuss it with their students, who are probably already using it.

    Palavras-chave : Natural Language Processing (NLP); Artificial intelligence (AI); the Generative Pre-Trained Transformer; ChatGPT; online assessment; personalised feedback; automatic grading; validity; fairness; learning process.

            · resumo em Africaner     · texto em Africaner     · Africaner ( pdf )