Moving Forward: Integrating Bots and Generative AI into Healthcare Settings

By Maika G. Mitchell, PhD, Senior Director- Researching COVID to Enhance Recovery (RECOVER) Administration & Operations, NYU Langone Health, and Carmen Mitchell, Hunter College, CUNY

Healthcare organizations that operate on a not-for-profit basis struggle with putting artificial intelligence to work in patient care. But this doesn’t mean they shouldn’t try, and it certainly doesn’t mean they can avoid the fundamental question of how bot and AI use will redefine what it means to deliver healthcare to their patients. On the contrary, such use promises to majorly affect the pay-for-performance equation that underlies current healthcare delivery and the 3Ps (price, precision, pay-off) that affix their imprimatur on a diagnosis and the carry-through to treatment.  Integrating artificial intelligence into healthcare is not a straightforward process either. It is a complex, multifaceted issue with no clear pathway. And when the matter is further examined across the convoluted, almost mysteriously disorganized body of healthcare regulations, the problem looks even more dismal. Strangely, the temptation that accompanies this line of thought is to figure that inserting a shiny new AI technology must inevitably define simple rules that clarify valid pathways in healthcare regulations.

Any organization can be tough nuts to crack when they attempt to make a technological change. This is especially true at the intersection of healthcare and technology. As in any change process, understanding the good reasons for altering the way work is done—and even more fundamentally, for altering the division of labor in the provision of healthcare—is certainly important. Indeed, in order for workers to “get” what AI is and what it does, they have to first understand why AI and bots are necessary. Workers also have to be involved in the AI implementation process, to whatever extent feasible, in order to “own” those good reasons themselves and not just be told about them by someone on high.

Nevertheless, these not-for-profit entities possess a single, potent, disruptive tool that could help make a consummatory difference as they strive to mobilize and harness the potential of the big technologies.

An adoption team was established with members from all corners of the organization. This group is essential for carrying the message about the new AI technologies. They also serve as a first line of contact for anyone who is worried about how these technologies will affect them. And they are there to smooth the way, as necessary, when retooling job responsibilities and workflows to integrate the AI systems.

Using artificial intelligence (AI) in not-for-profit healthcare is stalled by the high costs of these technologies. Medical care also typically has very tight budgets, and the interface with new technology can be expensive. AI and “generative adversarial network” (GAN) technologies are not much safer than human health workers and do not “pay for themselves” in nonprofit settings. This is not to say that some leadership teams and industrial partners don’t try to get sell-in to their new improvements by saying their “innovations” will save money. Still, the overall financial impact on the system is just as hazy as the near-future impact on personnel.

To introduce and effectively implement bots and generative AI in healthcare, a lot has to shift in the knowledge and capabilities of the people in the industry. This will require a swelling up of understanding that has to be led first and foremost in the industry changes companies involved will have in mindset. They will need to educate and train their staff to know what they are really getting into with these technology ventures so that the first step in potentially life-changing work for a patient is not misunderstood and, consequently, pitched in a way that misrepresents the genuine changes these two technologies can make in a patient’s life and that of a healthcare worker.

To make sure medical personnel can use the AI-driven diagnostic assistant well and trust its recommendations, there has to be a series of different training sessions. The AI provider should supply training materials and support. All staff members will need to be educated on the basics of the AI system and then commence “training” them on the system. However, getting the system to this point will not be a smooth or instantaneous process. Fine-tuning the devices in tandem while the staff and machines are learning from their collaborative mistakes can be arduous.

Yet the not-for-profit healthcare industry can derive considerable advantages from incorporating bots and generative AI. These two technologies can help promote a healthcare industry that operates more efficiently and accurately. Suppose healthcare is the not-for-profit industry that tends to lacquer most a human labor force that accomplishes quite human labor-like any number of motions with a hand interacting with a large amount of work and a patient’s appearance of well-being. In that case, the industry has underutilized human resources. Using bots for finance (reverse transactions, invoicing) can save man hours and allow the human staff time to troubleshoot and QA.  To make bots and generative AI work in not-for-profit healthcare institutions, we are surrounded by unavoidably complex and expensive realities. (Financial, Educational, Regulatory, and cultural realities create formidable obstacles.) These come at us with force, an overwhelming number of voices seemingly against what we are trying to do, and they keep me up at night.

Nevertheless, these not-for-profit entities possess a single, potent, disruptive tool that could help make a consummatory difference as they strive to mobilize and harness the potential of the big technologies. And what, I ask, is this singular, mighty tool? The massive workforce of healthcare personnel who are not just committed to doing their jobs effectively but also bound together in meaningful ways with the patients and communities they serve.