The first essays in this series explored one contribution of human–machine synergy: making the invisible visible. Machines can preserve and relate information across timescales and volumes that exceed human memory and attention. People can bring experience, curiosity, and judgment to what becomes noticeable. Together, they can ask questions that neither might have asked alone.
The previous essay introduced the challenge of complexity. A human being is not a collection of independent problems. Health unfolds through interactions among conditions, treatments, behavior, relationships, and circumstances. Understanding more of those interactions matters. But care ultimately requires something more than understanding: it requires the ability to act.
More than a division of labor
Human–machine cooperation is often described as a division of tasks. The machine observes or analyzes; the human interprets or decides. That can be useful, but it leaves out something essential. The machine's contribution may change how a person acts, and the person's response may change what the machine does next. Their interaction can create possibilities that did not exist in the original arrangement.
Consider someone managing several chronic conditions at home. A family caregiver offers reminders and advice, but those suggestions are not always welcome. The difficulty may have little to do with the quality of the advice. Relationships carry histories, expectations, and sensitivities. The same suggestion, offered in a different way, may lead to a different response.
A digital companion might help the person organize daily tasks, notice changes, or ask for assistance on their own terms. The caregiver could then contribute knowledge and encouragement without having to direct every action. The technology has not replaced the relationship or simply automated the reminders. It has altered the interactions, making a different form of cooperation possible.
Feedback changes what is possible
This is a cybernetic idea. In cybernetics, feedback connects action with its consequences: what happens in response to an action can influence what happens next. Norbert Wiener helped establish this way of thinking about communication and control in animals and machines.
In care, feedback is not confined to a machine adjusting a numerical setting. A person responds to a prompt. A caregiver notices hesitation. A clinician changes an instruction. The machine records what happened and helps make subsequent changes visible. Each response can alter the next action available to someone else. The resulting possibilities arise from the interaction, not from any participant acting alone.
Of course, feedback does not guarantee a good outcome. A reminder may irritate rather than help; an alert may create unnecessary anxiety. What matters is whether the participants can recognize the consequences and adjust. A system that keeps repeating an ineffective response is connected, but it is not learning to care more effectively.
Human–machine synergy is not simply the combination of capabilities. It is the possibility that feedback between them changes what the combined system can do.
Learning how to act together
Over time, the significance of feedback can deepen. The combined system may learn not only which action helps, but who or what should act, when, and how. A machine-generated prompt may work well for one person and poorly for another. A conversation with a trusted caregiver may be more useful at one moment; a clinician's assessment may be essential at another. Sometimes the most effective response is to give the person more room to act independently.
This learning need not mean that software autonomously reallocates clinical responsibility. People can interpret feedback, revise a care plan, change how support is offered, and decide when clinical intervention is warranted. The machine can maintain the history and make the effects of those changes easier to recognize. What evolves is the organization of the combined human–machine response.
That brings us to a second contribution of synergy. The first is making the invisible visible: expanding what can be noticed and understood. The second is making new actions possible: expanding the ways people and machines can respond together. The two reinforce one another. New observations can change action; action generates new observations; experience can change how the participants work together.
From more information to more effective care
The ambition for AI in healthcare, then, need not be to build a machine that knows everything about a person or makes every decision. It can be to create relationships between human and machine capabilities that become more effective through experience. In a complex world, the important question is not only what each participant knows or can do independently, but what their continuing interaction makes possible.
That raises the next question: how much variety in possible responses does a care system need to meet the variety of challenges a complex human being presents? The cybernetician W. Ross Ashby gave us a way to think about that question.