Artificial intelligence is here. While the hype is new, it has existed longer than most think. With a history rooted in the mid-20th century, technologies like natural language processing (NLP) and machine learning (ML) spent decades in an academic chrysalis before emerging to revolutionize our daily lives. Both were introduced in the 1950s and 1960s but have only taken off in the last decade.
The latest of these is LLM, which stands for large language model, one of the evolutions of NLP, which, with OpenAI’s release of ChatGPT in Nov. 2022, made this technology accessible to all. Within five days only, OpenAI reached 1 million users. In contrast, Facebook, for example, took ten months to get that many subscribers.
The accelerated adoption is startling. Yet, in the world of health care, AI’s promise remains largely unfulfilled. This piece explores the potential of AI, specifically large language models (LLMs), to humanize the interface between humans and machines, significantly reshaping health care communication and practice.
The health care industry’s conservative adoption of technology stands in stark contrast to the rapid embrace we’ve seen in other fields. From regulatory hurdles at every corner to interoperability issues between disparate health care systems and the ongoing resistance to change among health care practitioners, especially regarding patient privacy and safety, the challenges persist,
So what is different this time?
It is fair to say that health care providers prioritize the human body and mind, which are complex interfaces by themselves. That is where their expertise lies and it is precisely why the proper adoption of technology is a struggle.
Understanding one system very well makes the bandwidth to introduce or adopt another challenging. So when, it takes approximately 11 to 15 years or more to become a fully trained and licensed doctor, that is quite a long time focused on one system.
Thus, even if all other systems promise to deliver efficiency or improve the quality of care, their influence will dwindle, and adoption will lag, which is what we see.
Technology is yet another complicated system that promises significant improvement, yet it still sits outside the key focus area of health care professionals.
And while that has been true thus far, the scales are just about to tip, and an opposite trend will rise in the next decade. Believe it or not, it is because of artificial intelligence, but it has nothing to do with the intelligence part at all.
While the hype often focuses on the content of what new technology brings, in this case, the newly found creations from imagery and text, very few pay attention to how we connect to it. Thus far, we have interfaced with technology through a keyboard, a mouse, or a touch screen.
All are not natural, nor do they conform to how we truly connect or interact with each other as humans. The connection between technology and health care has been broken for so long because you had to learn a new interface to leverage the technological innovations developed.
Not anymore, though, because of AI. While it is not 100 percent here, one of the key advantages of AI is its ability to communicate closer to how we do, which means that we can now interface with the machines more naturally. The same way we connect to each other, whether verbally or visually.
For the first time, we get to ask a question, and instead of having a search result, we get an answer. And instead of typing a command, we can have a conversation with a machine. For the first time, making a request against these inanimate systems, which so far we needed buttons to engage with, feels more human.
As such, AI will humanize the interface of machines and systems that sit behind them. This innovation will propel the adoption because no one understands the human interface better than health care professionals. And now they can leverage all their experience without learning yet another interface.
In a health care ecosystem, many languages are spoken, and I am not talking about English vs. Arabic but one syntax vs. the other, whether between nurses and doctors, specialists and patients, and so on. Most medical errors result from bad communication, lack of standardization, documentation errors, or simply a lack of the correct information at the right time to administer the right intervention.
The opportunities to bridge the communication divide between these stakeholders in this ecosystem of saving lives are immense. If done properly, that could be one of the most advanced optimizations. Once we start looking at the breadth of complications that can arise from the many communication interfaces in a health care ecosystem, we realize the magnitude of the points of failure this system is primed to face.
And that is where AI could help the most. It is not about its capability of creatively answering but simply translating from one system to another and presenting the data to the relevant stakeholders in the correct format at the right time, which allows for the proper intervention to save a life.
And yes, AI might get it wrong, but at some point, that can also be optimized. A checks and balances layer can be added, and with the advancement of AI frameworks such as retrieval augmented generation (RAG) for retrieving accurate data from a defined knowledge base to ground large language models, the risk of errors is even further reduced.
Once we optimize the interface, we allow for more flexibility and adaptability to move at the pace of the innovation and not lag. AI Technology is still in its infancy, yet imagine the day when technology speaks our language rather than us speaking it. What opportunities may arise from the seamless integration of those two worlds with reduced friction from one system to another is a future worth looking forward to.
Carlo Mahfouz is a technology executive and author of Reality Check: In Pursuit of the Right Questions.