Artificial intelligence will likely replace some of your healthcare services in the future. How do you feel about that?
Science fiction writers have long presaged a world where human services are provided by robot and software providers. There is no novelty to the idea of computers responding to human needs in ways that suggest an innate intelligence of some kind. Rogue computers have long been the baddies of a cinematic genre. Can they be healthcare heroes too?
In November 2022, Open AI released Chat Generative Pre-trained transformer (ChatGPT). While you have been talking with Alexa and Siri for years, ChatGPT is something altogether different. ChatGPT is chat software, an algorithm, that has been trained on enormous data sets. This so-called large language modelling (LLM) gives the algorithm training, access, and knowledge to literally a world of knowledge. More than can be known by any one human.
It makes good sense that someone would sit ChatGPT down to take licensure exams for medical or legal providers. The American Bar Association (ABA) reported in March that ChatGPT, passed the Uniform Bar Examination, in the 90th percentile. In February, a study published in PLOS Digital Health reported ChatGPT was able to achieve a passing score on the United States Medical Licensing Exam (USMLE).
In the research, study authors presented ChatGPT with standard testing required for medical licensure and commonly taken by medical students. The three-step exam covers medical management, bioethics, and clinical reasoning. ChatGPT was provided with 350 of the 375 publicly available questions from the USMLE exam of June 2022. On the three sections, ChatGPT scored roughly between 52 and 75 percent across the exams. The study showed ChatGPT scored around 60 percent, theoretically passing the assessment.
Interestingly, ChatGPT outperformed PubMedGPT, a LLM chatbot trained only on biomedical information. Chatbots do not use real-time internet review to respond or behave. Instead, chatbots respond according to their innate language modelling to create word relationships—much as the human brain recognizes and generates associations based on known patterns.
As to the question of how you might feel about being treated by an algorithm? It was a trick question. Chances are good you already have received some diagnostic or monitoring software services enhanced with AI. Algorithms are potentially less likely to be affected by the bias and fatigue which can lead to missed and misdiagnosis.
In Radiology Today, Dr. Bibb Allen of the Data Science Institute notes, “We believe that AI is poised to significantly increase the value radiology professionals are able to provide their patients. Adding information acquired from AI algorithms to our reporting and workflow can significantly improve patient care.”
AI will likely not replace human physicians altogether. While there will be problems and errors, AI undeniably has promise and vast potential for medical technology and the provision of healthcare services in the future.
Experienced legal representation when you are injured through medical malpractice in Baltimore
Schochor, Staton, Goldberg and Cardea, P.A. provides aggressive legal service on behalf of our clients seriously injured through medical error. Contact us today or call 410-234-1000 to schedule a free consultation to discuss your case. We have offices in Maryland and Washington, DC.