Vetnews | Oktober 2024 8 « BACK TO CONTENTS Moreover, recent research has investigated ChatGPT’s proficiency in human clinical challenges. One study found that GPT-4 could accurately diagnose 57% of complex medical cases, a success rate that outperformed 72% of human readers of medical journals in answering multiple-choice questions (24). Additionally, GPT-4’s top diagnosis concurred with the final diagnosis in 39% of cases and included the final diagnosis within the top differential diagnoses in 64% of cases (50). In veterinary medicine, a notable case is a man on social media platform X (previously known as Twitter), who reported that ChatGPT saved his dog’s life by identifying immunemediated hemolytic anaemia—a diagnosis his veterinarian had missed (51). Veterinarians should recognize that pet owners may consult ChatGPT or similar AI chatbots for advice due to their accessibility (26). While the proliferation of veterinary information online can enhance general knowledge among clients, it also risks spreading misinformation (52). Customizing ChatGPT could address these challenges (refer to ‘Using ChatGPT in Veterinary Education’ below). In a human medicine study, GPT-4 can interpret ECGs and outperformed other LLM tools in correctly interpreting 63% of ECG images (53). A similar study has yet to be found in veterinary medicine. A veterinary example is provided in the Supplementary material, showing that GPT-4 did not identify an atypical atrial flutter with intermittent Ashman phenomenon in a 9-year-old Pug despite the addition of asterisks in the ECG to indicate the wide and tall aberrant QRS complexes (35). This example emphasizes that while ChatGPT is a powerful tool, it cannot replace specialized AI Leading Article Using ChatGPT in academic writing • Most journal publishers agree that ChatGPT cannot be listed as a co-author. • Several veterinary journals request authors to declare the use of ChatGPT in methods, acknowledgement, or designated sections in the manuscript. See Supplementary material. • Reviewers could mistakenly classify human writings as AI-generated content, while ML tools built based on specific language features could achieve 99% accuracy in identifying AI-authored texts. • The official ‘ChatGPT detectors’ are currently underdeveloped by OpenAI. ChatGPT’s limitations and ethical issues • Most veterinary professionals are familiar with AI and its application in veterinary medicine, while some remain sceptical about its reliability and accuracy, data security and privacy, and a lack of training. Hallucination and inaccuracy • Hallucination, or artificial hallucination, refers to the generation of implausible but confident responses by ChatGPT, which poses a significant issue. See Supplementary material. • Inaccuracy is not an uncommon finding when using ChatGPT. These unexpected errors can potentially harm patients. Intellectual property, cybersecurity, and privacy • ChatGPT is trained using undisclosed but purportedly accessible online data, and user-generated content is consistently gathered by OpenAI. • When analyzing clinical data, uploading de-identified datasets is recommended. • Alternatively, considering local installations of open-source, free-for-research-use LLMs, like Llama 3 or Gemma, for enhanced security. U.S. FDA regulation • Most FDA-approved AI and ML-enabled human medical devices are in the field of radiology, followed by cardiovascular and neurology. • FDA has not set premarket requirements for AI tools in veterinary medicine. • The AI- and ML-enabled veterinary products include dictation and notetaking apps, management and communication software, and radiology services, which may or may not have scientific validation. • The continual learning and updating of LLM pose a special regulatory challenge for FDA. Practical learning resources • Resources for learning about ChatGPT and generative AI are abundant, including OpenAI’s documentation, online courses from Vanderbilt University via Coursera, Harvard University’s tutorial for generative AI, and the University of Michigan’s guides on using generative AI for scientific research. Links are provided in Supplementary material. • Readers are encouraged to ask ChatGPT for learning resources: https://chat.openai.com
RkJQdWJsaXNoZXIy OTc5MDU=