Despite the constant barrage of stories about AI in the media, the truth is that AI is still in its infancy. Comparing current AI capabilities to future capabilities is sort of like comparing Orville and Wilbur’s Kitty Hawk Flyer, which flew a few hundred feet, to a modern jetliner that can fly nonstop from New York to Tokyo. They are both flying machines but that is where the similarity ends.
So too with AI. Here is an example of how AI will be used in the near future. Let us say that the manager of a manufacturing facility regularly experiences delays in shipments of component parts from component suppliers. Currently, this manager might use Generative AI to write a letter to the tardy component suppliers, complaining about production delays. While the Generative AI program is helpful and no doubt saves time for the manager, it will not necessarily produce actual, on-time delivery of the required component parts.
Future supply chain management will be much different. An AI software program (referred to as an “agent,” hence the name “agentic”), will closely monitor a manufacturer’s inventory. When the agent finds that inventory is low, it will send an electronic message to the purchasing department where another agent will send a purchase order to the component supplier. An agent system at the supplier will receive the purchase order and, in turn, issue an invoice for payment. The supplier’s agent will automatically order the removal of the necessary components from shelves in the warehouse, for delivery to the loading dock. Quite possibly a robotic forklift will move the palette of component parts. A different agent system will electronically order a truck to pick up the component parts and deliver them to the product manufacturer. It will also send a message to the manufacturer’s purchasing department
indicating when the truck will arrive. The truck, by the way, will probably be a driverless vehicle and will be unloaded by the product manufacturer’s own robotic forklifts.
Little, if any, human interaction will be required to move these component parts from one place to another, on time, just as little, if any, human interaction is required for the monitoring of a modern jet flying on autopilot.
Mark Purdy, in a recent Harvard Business Review article, defines Agentic AI as follows:
“It refers to AI systems in models that can act autonomously to achieve goals without the need for constant human guidance. The Agentic AI system understands what the goal or vision of the user is and the context of the problem they are trying to solve.”
(Purdy, Mark: What Is Agentic AI, and How Will It Change Work? Harvard Business Review, December 12, 2024.)
If you think Agentic AI is a part of a utopian (or dystopian) distant future, it is not. Startup companies and big tech are both rolling out Agentic AI systems right now. Yes, those systems may be new and untried but like all technology, it will get better and better at amazing speed.
The future use of Agentic AI in health care promises the most important and exciting possibilities. Knowledgeable physicians and administrators believe that AI will very soon review a patient’s chart and diagnose diseases and conditions faster and more accurately than human doctors. And that is just the beginning. In addition to reviewing a patient’s electronic chart, AI software will “listen” (much as Alexa now does) as a physician speaks with and examines a patient and then incorporate the oral information in rendering a diagnosis. The hope is that AI will relieve physicians of the mountainous paper work currently required and allow them more opportunities to use empathy and interpersonal skills to reassure their patients by explaining diagnoses and potential outcomes.
