Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.
ChatGPT sitting for FRCS Urology examination: Will artificial intelligence get certified?
2
Zitationen
5
Autoren
2024
Jahr
Abstract
Introduction: The use of artificial intelligence (AI), including neural networks and deep learning, has rapidly changed many industries. One recent application of AI that has gained the attention of physicians and the general public alike is ChatGPT, a chatbot developed and released by OpenAI Inc. in November2022. We challenged ChatGPT to sit the Fellowship of the Royal College of Surgeons (FRCS) Urology examination, both parts, to assess its performance and determine its eligibility for the title. Methods: We prepared a set of 100 Multiple-Choice Questions (MCQs) for Part 1 and 8 viva scenarios for Part 2. We choose to use ChatGPT 3.5 that is the free version available online, and responses were documented during the simulation. Results: In Part 1 MCQs, Chat Generative Pre-trained Transformer (ChatGPT) managed to answer some questions correctly with an overall score of 35%. However, in Part 2 Viva, its performance improved to an extent, demonstrating proficiency by offering detailed explanations and reasoning, achieving average score in some stations. ChatGPT’s performance raised some concerns on answering questions. Conclusion: AI is definitely an asset that can be used in our daily practice, yet it is crucial to take into consideration the concerns about the accuracy of data provided, including AI hallucination. ChatGPT failed to get the FRCS Urology title this time. Still in the early stages of development, further research is necessary to comprehensively understand their limitations and capabilities. Level of evidence: Not applicable
Ähnliche Arbeiten
Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI
2019 · 8.400 Zit.
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead
2019 · 8.261 Zit.
High-performance medicine: the convergence of human and artificial intelligence
2018 · 7.695 Zit.
Proceedings of the 19th International Joint Conference on Artificial Intelligence
2005 · 5.781 Zit.
Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)
2018 · 5.506 Zit.