The patient wants a human. When the agent changes the game

"I want to talk to a person." It is probably the most repeated sentence in digital customer service worldwide. PwC's Future of Customer Experience survey sizes the feeling: 59% of consumers say companies have lost touch with the human element of the experience, and 82% of American consumers (74% elsewhere) want more human interaction in the future, not less.

The preference for humans is not romanticism. It is self-defense. Customers learned the hard way that talking to the machine meant menu loops, tickets without resolution and the feeling that the company had built a wall between them and anyone who can decide. Bot became a synonym for hiding.

But the real preference is for the outcome, not the species. In 2019, Logg, Minson and Moore published six experiments on what they called algorithm appreciation: in objective tasks, such as estimates and forecasts, people follow identical advice more when they believe it comes from an algorithm than from another person. Trust migrates toward whoever gets it right, not toward whoever has a face.

The most cited turning-point case is Klarna. In February 2024, one month after launch, the fintech's AI assistant was already handling two-thirds of customer service chats: 2.3 million conversations, work equivalent to 700 full-time agents. Average resolution time fell from 11 minutes to under 2, repeat inquiries dropped 25%, and customer satisfaction stayed on par with human agents. When the machine resolves faster and better, preference flips.

In healthcare, the division of labor becomes even clearer. The patient wants a human for the diagnosis, for fear, for the decision that changes treatment. But they do not need a human for "your exam is scheduled", "it is time for your medication" or "is this normal after my appointment?". For those questions, they want an immediate answer, at 10 p.m., without judgment and without a queue.

The agent does not replace the human. It replaces the silence.

AI changes the game under three conditions. First: it answers in seconds, not in a queue. Second: it remembers the journey, and does not ask again what the patient already said. Third: it knows when to hand off to a human, and does so with context, without making the patient start over. Under these conditions, the comparison stops being agent versus human and becomes agent versus no one: the call that was never returned, the follow-up visit no one confirmed, the check-in that never happened.

This is the design we follow at Lisa: the agent takes care of the continuous part of the journey, the human takes care of the decisive part, and the boundary between the two is drawn by clinical protocol, not by operational convenience. The patient does not prefer the machine over the person. They prefer presence over abandonment.

The question for whoever designs health journeys is not "human or AI". It is: at which moment of the journey does each one build more trust? Whoever gets this division right discovers that "I want to talk to a person" was never a request for humans. It was a request for attention.

PwC - Experience is everything Logg, Minson and Moore - Algorithm appreciation, 2019 Klarna - AI assistant handles two-thirds of customer service chats, February 2024