This chapter explores the impact of the development and implementation of the ChatGPT language model on the traditional framework of Bloom's taxonomy of learning stages. Through examination of data and case studies, the study argues that the advanced natural language processing capabilities of ChatGPT have led to a shift away from the linear, hierarchical model of Bloom's taxonomy, towards a more dynamic and fluid approach to knowledge acquisition and application. The results of the study suggest that the incorporation of ChatGPT and similar language models into education and training programs may lead to more effective and efficient learning outcomes.
TopIntroduction
ChatGPT is a large-scale language model developed by OpenAI. It uses a transformer architecture, which was introduced by Vaswani et al. (2017). The model is trained on a massive dataset of internet text, allowing it to generate human-like text with a high degree of coherence and fluency. One of the key features of ChatGPT is its ability to perform various language understanding tasks such as question answering, machine translation, and text completion. This is achieved through fine-tuning the pre-trained model on a smaller task-specific dataset. This approach is similar to that used in BERT, a transformer-based language model developed by Google, which is trained on a massive dataset of internet text and then fine-tuned for various natural language understanding tasks (Devlin et al., 2018).
ChatGPT has been used in a wide range of applications, including language translation (Liu et al., 2020), dialogue systems (Hosseini-Asl et al., 2020), and text generation (Zellers et al., 2019). It has also been used in the field of education, where it has been employed as a tool for automated essay scoring (Yang et al., 2019) and for generating feedback on student writing (Wang & Hovy, 2020).
Language generation models such as ChatGPT have the potential to revolution-ize the field of education and training by introducing the capabilities of Artificial Intelligence (AI) for 21st century education (Wogu et al., 2019). These models can assist educators in a wide range of tasks, from generating feedback on student writing to creating personalized learning materials (Fernando et al., 2023; Meyer et al., 2023). One of the key advantages of language gen-eration models is their ability to generate high-quality, human-like text (Hsu & Ching, 2023). This can be useful in tasks such as automated essay scoring, where the model can provide detailed and accurate feedback on student writing. For example, a study by Yang et al. (2019) (Yang et al., 2019) used BERT, a language generation model similar to ChatGPT, to evaluate the quality of automated essay scoring. The study found that the model performed comparably to human graders in terms of both accuracy and consistency.
In addition to automated essay scoring, language generation models can also be used to generate personalized learning materials. For example, a study by Wang and Hovy (2020) used a pre-trained language model to generate personalized feedback on student writing. The study found that the feedback generated by the model was effective in helping students improve their writing skills. Language generation models can also be used in conversational agents, such as chatbots, for creating interactive and engaging learning experiences. A study by Hosseini-Asl et al. (2020) (Hosseini-Asl et al., 2020) proposed a few-shot dialogue generation model using pre-trained language models such as GPT-2 and GPT-3 to fine-tune the model for a specific task. The study found that the fine-tuned model generated high-quality and coherent responses in a dialogue setting.