The Role of Generative AI in Advancing Medical Research
Generative AI has become a driver of productivity throughout the country as a quick reference tool, but perhaps its capabilities of research can be utilized in a professional environment. Since the dawn of the information era and the introduction of the internet, research has become ever more accessible. Notably though, this access and connectivity has created new issues.
While fifty years ago it was difficult to find applicable research because it was not available, today it is difficult to find research in an objective flood of information. With upwards of two million research articles published each year, it is a wonder academia is able to collect the information crucial to the continued and collaborative building of any field of study. Take, for example, the field of medicine. Research is fundamental to the existence of medical care. Study and advancement of the field is the driving force behind most all medical care institutions in the world. You may think that in the United States, a country with largely private healthcare coverage, that most hospitals are similarly for profit. In fact, the vast majority of hospitals are either government run or non-profit organizations. These hospitals conduct an incredible amount of research, driving the miracles of technology we witness in modern day: robotic surgery, corrective laser eye procedures, cutting edge drug invention. Yes, these products are developed privately, but without the infrastructure of research hospitals and academic institutions, there would be nowhere to deploy these new technologies.
In this way, the field of medicine is unique. Unlike law or environmental research, medicine is perpetually revolutionizing its own field. Similar to the renowned Moore’s law, it appears that medicine continues to exponentially expand its capabilities and perhaps it must. Discovery, it appears, is the lifeblood of modern medicine, and in an era of overwhelming information, perhaps generative AI is the key to continuing the centuries long progress of the medical field.
AI Research Aid
It doesn’t take much imagination to see that generative artificial intelligence will do wonders for medical research on a day to day basis. A significant proportion of research in the field, whether academic or practical, is quite literally research. Whether browsing the internet, reading books, or referencing recent medical journals, significant amounts of time are spent combing through data. Humans are not good at this. While we have in fact invented this task of searching, we have also invented technology that can do it significantly better than us. A search for a topic in a mountain of textbooks that would take hours for a human would only take a second for a computer. In a similar leap, with a properly trained generative AI, there is no need for search or even reading through a dense passage of jargon. AI models are able to instantly recall, contextualize, and summarize information from training data and even live input such as the internet.
The ability for a computer algorithm to develop a human recognizable, useful form of context is groundbreaking. Only in the last few years was this even thought possible, and now it is at the fingertips of every medical researcher across the globe. Perhaps in the past we have valued memorization and the process of searching for relevant information, but even the smartest person in the world cannot possibly contextualize the 2 million new research articles published each year let alone read them. For generative AI, this is easy.
The possibility of generative AI as a research aid is something that must be taken advantage of. Instead of sitting down at a computer to complete background research for a new study or even for a doctor to familiarize themself with a new viral outbreak, a well prompted and specifically trained generative AI model can complete the work for them. With billions of lines of research milliseconds away, summarizing the important information and connecting important context is a trivial task for these algorithms. Indeed, an argument could be made that doctors should and must take advantage of this technology for the access and reach it provides. Instead of trying to keep up with an endless flood of new information, generative AI can allow medical professionals to focus on patient care and academics can enjoy access to more information than they could possible read.
As a research aid, AI can certainly not be ignored. It may seem like a scary concept to consider, trusting a nondeterministic algorithm with our medical research, but the tradeoff is worth it. This technology is offering access to an abundance of information that would otherwise be unattainable in many lifetimes. As long as it is used responsibly, the strides to follow in medical research are close at hand. This responsibility will come in the form of an understanding that AI may not be accurate all the time and the information we extract from these algorithms must be approached with care. If we can do that, generative AI will help us revolutionize medical research.
Of course, AI is not limited to gathering knowledge, it can also collect information. Consider the medical subfield of clinical psychology. Experimentation and tests are often performed via surveys, interviews, observations. This, like any text in a book, is data. Data which can be contextualized and summarized by generative AI. Indeed, a specially prompted chatbot not dissimilar to chatGPT could be positioned to conduct live research and report findings. For example, a chatbot could be instructed to engage in empathetic conversations with patients, listening to concerns and answering questions. This is an invaluable resource to research which cannot be conducted in person due to distance or simply as a less invasive option. Additionally, given specific prompts, chatbots can screen for certain conditions, able to collect virtually endless data points with only one computer system.
In a field of research that struggles with funding just as much as any other, generative AI is positioned to make the gathering of data cheaper and more efficient as well as potentially more accurate. Intuitively, algorithms are never tired, can be trained to minimize bias, and can offer precisely the same treatment to every human subject, helping to eliminate variables in an experiment. Further, with one codebase, millions can interact with an AI model to produce data. In the absence of more researchers, generative AI chatbots can be invaluable stand-ins, contributing to the work of experimentation. However valuable the data or research, no one is excited to contribute to a lengthy survey, especially about potentially sensitive medical or mental health details. Generative AI offers the possibility of adding a human touch to the scenario.
Models such as GPT-3 (utilized in chatGPT) show remarkable abilities of empathy and understanding, capabilities imperative to making human subjects feel comfortable enough to aid in important research. Further, generative AI has been shown to be able to effectively anonymize data which maintains the same statistical weight. In other words, given a large input, such as many AI conducted interviews, these models can output all the data slightly skewed here and there so as to anonymize the data while retaining the statistical properties. As long as we remember the importance of considerations such as this, generative AI is well positioned to aid in the collection of much needed research data for the medical field.
Yet another valuable integration of generative AI into the field of research may be in the form of workplace study. Whether at a general practitioner’s office or in the emergency room, generative AI can be positioned to take input and record data. Much of the crucial research being conducted in the world of medical practice is recorded by written word. Surgeons testing new methodologies will write reports and doctors conducting treatment studies will record their observations, but what if this is not the best we can do. If this data is provided directly to a generative AI as input, perhaps meta-analysis can be performed concurrently.
In a larger perspective, what if we use these algorithms to research the field itself. With the proper considerations, generative AI could be integrated into the field of medicine such that it is able to access every corner of the building, so to speak. Suppose we view a whole hospital or medical practice as a large experiment. Where are the places we can improve, how are results in surgery connected to long term patient health. These are connections that a generative AI can do trivially when given the proper data. Improvements in medicine we haven’t even considered are perhaps simply waiting to be uncovered via generative AI analysis.
The Path to AI Medical Research
Generative AI has immense potential for medical research by serving as a powerful tool for information retrieval, data collection, and workplace integration. With the overwhelming volume of research publications available today, generative AI can quickly and efficiently gather and contextualize information, enabling researchers and medical professionals to access a vast pool of knowledge. By facilitating data collection through empathetic interactions with patients this technology will also provide valuable insights for clinical psychology and other research areas. Furthermore, integrating generative AI into medical practices and hospitals opens up avenues for meta-analysis and uncovering hidden connections within the field.
Responsible use and consideration of accuracy and ethics are crucial, but with care, generative AI holds the key to revolutionizing medical research and advancing the progress of medicine in ways we have yet to fully comprehend.