A reader sent me this question: “Yesterday, after my MCAT class, two biomedical engineering students and I talked about this article and the future of medicine. We debated whether such robots could reduce the need for doctors by 80%.”

When I read such predictions I chuckle at the naivety of those who make such pronouncements. The computer advocates do not really understand medical care and diagnosis.

What do we do that computers/robots will have great difficulty replacing? The most important thing that we do involves understanding our patients and what they are really saying. We understand how to ask questions and how to interpret those questions with carefully couched follow-up questions.

We understand how to approach each patient, what vocabulary to use, and how to read body language.

Let’s consider possible uses for computers.

Computers do not think, rather the provide results of calculations. They cannot choose which data to include.

Computers can only help with diagnosis when we know that we have the wrong diagnosis. Computers will make the wrong diagnosis if we do not know what information to put into the computer.

Remember the classic adage, junk in, junk out. We have to understand the patient before we can take the history, do the physical and order appropriate tests.

Many patient diagnoses are delayed because we have premature closure. We stop collecting data and stop thinking.

Computers cannot provide the healing touch. Computers cannot comfort.

Maybe one day we will have androids who appear human. Until then patients need us, because the human relationship really matters to most patients.

Robert Centor is an internal medicine physician who blogs at DB’s Medical Rants.