As we navigate through a significant technological paradigm shift, particularly in the education and healthcare sectors, my new book (Krumsvik, 2023) delves into the intricacies of artificial intelligence (AI), spotlighting the role of AI technologies like ChatGPT and GPT-4. It addresses two key themes: first, the potential impact of AI – is it positive or negative? And second, the challenges and opportunities it brings.

In Norway, AI has been somewhat overlooked in schools and higher education, necessitating a more thorough examination of this field. Recently, the conversation has evolved from everyday discussions about AI to a more profound analysis of large language models such as ChatGPT and GPT-4. This shift is essential as these models represent a significant advancement in AI capabilities. My book aims to enhance current knowledge about AI by showcasing international research findings on AI, ChatGPT and GPT-4. Given the retrospective nature of the existing knowledge base, the book includes new literature reviews about these advanced language models, focusing on their applications and implications.

A significant concern is whether GPT-4 perpetuates biases and misinformation, as earlier versions exhibited gender and class stereotypes. An extensive testing of GPT-4 over six months has been carried out, examining its performance in various contexts, including Norwegian medical exams. This was crucial to understand GPT-4’s proficiency in Norwegian and its application in complex academic and research settings. I also revisit themes from an international trial study using an early version of GPT-4 (Bubeck et al., 2023), to see if the model has evolved over eight months.

‘A significant concern is whether GPT-4 perpetuates biases and misinformation, as earlier versions exhibited gender and class stereotypes.’

The findings suggest GPT-4 has made notable progress since Bubeck et al.’s (2023) testing from late 2022. Its performance in medical exams, both internationally and in Norway, is particularly impressive, handling one of the most challenging exams in the sixth year of medical studies with high competence. It is important to dig deeper than everyday discourse about AI to provide a perspective that comprehensivey explores the possible reasons for GPT-4’s success, analysing its foundation in machine learning, intelligent tutoring systems and learning analytics.

From a practical standpoint, my book illustrates how GPT-4’s capabilities are utilised in real-life scenarios. For example, medical student Vegard Slettvoll at the University of Bergen, Norway, uses ChatGPT as a versatile tool and ‘teacher assistant’ in his studies. He leverages it for simulating patient cases and preparing for practice periods and exams, benefiting from its round-the-clock availability and multimodal capabilities. Slettvoll’s experience indicates how AI is poised to transform higher education, enhancing student-active learning and potentially reshaping foundational educational principles. However, this technological shift also calls for rethinking traditional assessment methods, where formative and summative assessments need to be redesigned, stand in a dialectic relationship (as in Vegard’s example) and complement each other.

Digital Competence in the AI Society asserts that AI and large language models like ChatGPT and GPT-4 are not just technological tools but reflectors of our societal and ethical values. On one hand, it is impressive how GPT-4 adeptly handles complex language tasks, including Wittgenstein’s (1997) language games. However, the ethical landscape surrounding AI is vast and complex. The book raises questions about the ethical use of AI, particularly in sensitive areas like patient data handling. The EU’s AI regulation is positioned as a potential solution, yet the extent to which AI can utilise patient and registry data for training remains a topic of debate. Furthermore, the intersection of AI with major publishers raises concerns about the use of copyrighted materials for AI training. The book questions the implications of AI’s training on educational sector registry data, and the potential perpetuation of stereotypical biases through the texts it learns from.

In conclusion, AI challenges us to reconsider how we envision the future of education and healthcare. The journey ahead with AI demands digital competence, critical thinking, and a solid foundational knowledge base to navigate the complexities and capitalise on the opportunities presented by these advanced technologies.

References

Bubeck, S., Chandrasekaran, V., Eldan, R., … Zhang, Y.  (2023). Sparks of artificial general intelligence: Early experiments with GPT-4. ArXiv:2303.12712v5. https://doi.org/10.48550/arXiv.2303.12712

Krumsvik, R. J. (2023). Digital competence in the AI society. A look at how artificial intelligence shapes our lives (in Norwegian). Cappelen Damm Akademisk.

Wittgenstein L. (1997). Philosophical investigations. Pax forlag.