Shortly after the original iPhone was launched by Apple, my father, an unlikely early adopter, purchased one with the intention of keeping it in the trunk for emergencies, as he shared with me.
At that time, he did not anticipate that this gadget would eventually take over the functions of maps, radar detectors, AM stereo traffic updates, CD players, and even coin-operated parking meters. This unforeseen evolution went on to revolutionize the entire automotive industry.
Similar to many individuals, my father struggled to envision how a mere phone could completely transform our personal and professional lives. His response was typical when encountering cutting-edge technology. Often, we try to assimilate new advancements within our existing understanding by relating them to what we already know.
The trajectory of Generative AI mirrors this pattern. Despite significant investments by companies in this technology to streamline processes, improve operations, and expedite research and development, even experts in the field are amazed by the rapid advancements of AI. I find myself among those astounded by its progress.
Half a year ago, I embarked on writing a book titled “ChatGPT, MD,” exploring the potentials and challenges of Generative AI in the medical field. Initially, I was apprehensive that my enthusiasm for the technology might be overly optimistic.
With its projected exponential growth, indicating a doubling of capabilities annually, I envisioned it becoming a central hub of medical knowledge for both physicians and patients within the next five years (when it would be 32 times more powerful).
As my book nears publication next week, I am beginning to realize that my initial timeline may have been too conservative.
In March, at the 2024 GTC AI event hosted by Nvidia, a series of groundbreaking announcements were made, leaving the technology and healthcare sectors in awe. The California-based semiconductor giant, which dominates 80% of the global semiconductor market, introduced a range of innovative collaborations and advancements that have the potential to fundamentally reshape various aspects of healthcare, from diagnosis and treatment delivery to drug discovery and healthcare economics.
One of the most significant revelations emerged from the partnership between Hippocratic AI and Nvidia, focusing on cognitive AI agents that are purportedly more efficient and capable than human nurses.
According to data released by the companies, these AI systems outperform midwives by 16% in assessing the effects of medications on laboratory parameters, exhibit 24% higher accuracy in detecting unsafe dosages of over-the-counter drugs, and demonstrate a 43% improvement in identifying condition-specific adverse interactions from OTC medications. All of this is achieved at a cost of \(9 per hour, significantly lower than the average hourly wage of \)39 for nurses in the United States. These AI nurse-bots are designed to offer new diagnoses, manage chronic illnesses, and provide patients with comprehensive yet concise explanations of medical guidance.
These swift advancements signify that the world stands on the brink of a technological revolution that could surpass the impact of the iPhone. For both patients and healthcare professionals, three key implications arise:
1. Progress in GenenAI Healthcare Is Outpacing Expectations.
While the human brain can comprehend linear and geometric growth rates quite well, understanding the implications of constant exponential growth poses a challenge. This is precisely what we are witnessing with cognitive AI.
Imagine a pond with a single lily pad. If the number of lilies doubled each night, the entire pond could be covered in just 50 days. However, by day 43, only 1% of the pond’s surface might be covered. It might seem implausible that the lily pads would fill the entire pond in just seven days.
Experts predict that the computational advancements in AI are likely to double annually, if not accelerate even further. Despite conventional estimates, tools like ChatGPT are projected to be 32 times more potent in five years and over 1,000 times more powerful in a decade. This rapid progress poses challenges for both healthcare providers and patients to grasp, but it also signifies that now is the opportune moment to prepare.
2. Distinctive Features of GenAI Compared to Earlier AI Models
When evaluating the transformative potential of cognitive AI in healthcare, it is crucial not to let past failures, such as IBM’s Watson, cloud our expectations. IBM envisioned Watson to enhance care by assisting in diagnoses, care planning, and interpreting complex health data for cancer patients.
At that time, I held reservations, not due to the technology itself, but because Watson relied on data from electronic medical records, lacking the precision necessary for accurate “narrow AI” diagnoses and recommendations.
In contrast, cognitive AI integrates a broader spectrum of information sources for more practical applications. It not only draws from published medical literature and textbooks but also incorporates real-time data from global health databases, ongoing clinical trials, and medical conferences. In the near future, it will encompass clinical feedback and continuous loops from actual patient outcomes. This comprehensive data integration will enable cognitive AI to stay at the forefront of medical knowledge, setting it apart from its predecessors.
However, cognitive AI still requires several iterations before it can be widely implemented without direct supervision from doctors. Nonetheless, Nvidia’s bold venture into the healthcare sector demonstrates a much-needed willingness among tech companies to navigate the legal and regulatory complexities of the field. Once an AI clinician chatbot is introduced, numerous other companies are likely to follow suit promptly.
3. Ubiquitous Presence of GenAI in Healthcare Settings (Hospital, Office, and Home)
Many individuals find it challenging to envisage the transformative influence cognitive AI will exert on healthcare, akin to how my father never envisioned that his iPhone, initially stowed in the trunk, would become an indispensable tool for daily life.
The concept of accessing medical advice and expertise continuously—affordably, reliably, and conveniently around the clock—represents such a departure from current healthcare models that our minds tend to dismiss it as implausible. Nevertheless, it is becoming increasingly apparent that these capabilities are not only feasible but probable.
Feedback from both cognitive AI practitioners and patients frequently highlights that the responses, especially when appropriately prompted, closely align with clinician recommendations. This underscores the improving accuracy and reliability of cognitive AI in healthcare settings, signaling a fundamental shift in the delivery of medical care.
In a decade, we will likely view today’s skepticism similarly to how I regarded my father’s initial underestimation of his iPhone. We stand on the cusp of a significant transformation that will render cognitive AI as indispensable to everyday life as smartphones.
