The latest research conducted by OpenAI has shown groundbreaking results in the field of medical assessments. OpenAI’s advanced AI model, GPT-4, demonstrated remarkable proficiency when tested on a set of 87 multiple-choice questions typically used to evaluate eye assessment skills in medical professionals.
In the study, GPT-4 successfully answered 60 out of the 87 questions, outperforming both junior doctors and trainee ophthalmologists. This achievement is particularly impressive considering that the AI model had not been specifically trained on these questions prior to the test.
While GPT-4’s performance was commendable, a group of expert ophthalmologists still outperformed the AI model, with an average score of 66.4 correct answers. This indicates that human medical professionals continue to possess a nuanced understanding and expertise in the field of eye assessments that AI models have yet to fully replicate.
Other learning language models (LLMs) such as Google’s PaLM 2 and GPT-3.5 also participated in the assessment, with varying degrees of success. While GPT-4 emerged as a frontrunner, the study highlighted the ongoing need for a collaborative approach that leverages the strengths of both AI models and medical professionals in providing comprehensive and accurate medical assessments.
The results of this study showcase the potential of AI models like GPT-4 to enhance medical assessments and support healthcare professionals in their decision-making processes. As technology continues to advance, the integration of AI in medical practices is poised to revolutionize the field of healthcare, offering new possibilities for improved patient care and outcomes.
