In the realm of medicine, generative artificial intelligence (AI) is making remarkable progress. Google's Med-PaLM 2 achieved an impressive 86.5% on the U.S. medical license exam, a 20-point leap from its previous best. Similarly, ChatGPT's clinical note writing is now indistinguishable from human-written notes.
These advancements in generative AI are making skeptics reevaluate their doubts about the technology's capabilities in fields, such as writing, education, and healthcare.
People Are Increasingly Recognizing ChatGPT's Capabilities
A significant number of doubters who criticized AI's limitations earlier this year, contending that extensive language models could not effectively replicate the roles of writers, educators, or doctors, have now shifted their perspective on the potential of this technology.
With the continuous advancement and expansion of generative AI systems' capabilities, such as ChatGPT, it's becoming evident that they can rival human cognitive capacities by drawing from a growing pool of information. The remaining uncertainty revolves around whether there are any aspects of fundamental human nature that generative AI cannot imitate and potentially enhance.
Healthcare professionals argue that chatbots will not reach the human levels of compassion, empathy, and reliability crucial for doctor-patient relationships, a view shared by patients valuing these connections. Nevertheless, a recent study indicates rapid advancements in machines in these areas.
Is AI Developing EQ Comparable to Humans?
Behavioral therapy for alcohol abuse at the University of Texas, Austin, proved ineffective. To enhance patient engagement, the chair of internal medicine tasked a team to develop a compassionate script, but with no progress, ChatGPT was eventually employed and completed the task within a week.
The well-crafted letter not only demonstrated sincerity and thoughtfulness but also managed to avoid the use of complex medical language that often hinders patients' understanding of treatment instructions.
Social workers further utilized the generative AI to adapt the communication for a fifth-grade reading level and translate it into Spanish, resulting in improved clarity and relevance in both languages. Clinicians who employed chatbots to craft empathetic patient remarks have been consistently amazed by the outcomes.
In a recent evaluation, a doctor expressed that their experience with ChatGPT was astonishing. Additional medical professionals remarked that they regret not having access to such technology during their training and emphasized the value of giving it a chance.
ChatGPT Is Incapable of Diagnosing Diseases
Although ChatGPT's performance in the study might appear impressive, it is important to note that it does not replace a medical professional. The chatbot merely provided suggestions and potential diagnoses for health-related queries on Reddit, without conducting physical examinations or clinical follow-ups to validate its accuracy or effectiveness in suggested solutions.
Nevertheless, these results indicate the potential for AI assistants like ChatGPT to enhance doctors' productivity by handling simpler tasks, potentially leading to fewer clinical visits and improved healthcare access for underserved populations facing barriers.
Chatbots hold the potential for improved transparency in considering and explaining various possible diagnoses for patient symptoms, potentially reducing misdiagnoses and enhancing patient education.
Dr. Teva Brender from the University of California San Diego suggested that AI chatbots could assist doctors in tasks like test result notifications, discharge instructions, and medical charting, ultimately saving time.
Dr. Isaac Kohane from Harvard highlighted that GPT-4, the latest iteration of language models like ChatGPT, could aid doctors in effectively communicating with patients. A physician's tweet emphasized the importance of patient care and self-improvement in light of the study's findings.
In conclusion, the potential of chatbots lies in the increased transparency in diagnosing symptoms, potentially lowering misdiagnosis rates and improving patient education.
RELATED ARTICLE: Doctors Warn Against Using ChatGPT for Medical Advice; AI Chatbot Unreliable, Makes up Health Data
Check out more news and information on Artificial Intelligence in Science Times.
