Bernard Nordlinger: “Artificial intelligence will not take over humans”

Artificial intelligence (AI) challenges medicine by profoundly modifying practices. In the history of scientific research, when did a test of this magnitude take place?

The comparison goes beyond the scientific world. Take the advent of the printing press, of the steam engine: society had no choice but to adapt in order to move forward. AI falls into this category of major change. And like any major change, its acceptability takes time. We are only at the beginning of its application in all areas of society, particularly in the world of medicine.

It must be said that the fantasies and the fear that AI arouses are still very present, in the minds of doctors as in those of others. The term “artificial intelligence”, which is not very suitable, does not help to appease the spirits. The media, which sometimes prefer the sensational to objective information, contribute to this feeling.

We wanted to combat this conception of AI with our book, which calls on indisputable experts in their field of action (doctors, researchers, teachers, data specialists, sociologists, Editor’s note) and is based on facts.

Oncology is a medical discipline where AI is breaking through. Can we envisage, tomorrow, the eradication of certain cancers?

It’s too early. AI allows doctors to avoid some tedious work, such as analyzing genes in large quantities. It helps in the examination of tumors and the careful evaluation of cancerous cells. It can classify breast cancer tumors according to genetic changes, by family, which helps the doctor to more accurately choose one treatment over another.

This is what is called – again with a bit of awkwardness – precision medicine. Step by step, oncology is moving, with the help of AI, towards a specialty where cancers are no longer treated according to the diseased organ, but according to the genomic modifications of the cells of the sick person.
But there is no method that can eradicate cancer through AI.

Conversely, are there medical specialties where AI has no place?

I cannot answer this question since the place of AI in medicine and surgery is constantly evolving. AI is very present in image specialties: imaging of course, radiology, anatomopathology (cell analysis, editor’s note), ophthalmology and even cardiology.

One might think that psychiatry is too “human” a discipline to which AI would bring nothing more. Think again ! It helps in diagnosis and even in treatment.
Laurence Devillers (university professor in AI, editor’s note) talks about it in a chapter devoted to the use of chatbot (or conversational agent) in health. Psychiatrists have realized that autistic children, with difficult communication, will confide more in a small robot “animated” with an AI of their size than in a doctor in a white coat!

Same observation at the other end of the spectrum of life, in nursing homes. Some establishments no longer hesitate to pay nearly 7000 euros for the purchase of the Paro seal soft toy. This is a small robot that is able to analyze the manifestations of emotions. With a bit of cynicism, we could say that we are giving our elderly a robot instead of human caregivers. Yet, the therapeutic benefits of Paro have been demonstrated. He is not a person but he has a function.

Exactly, a stuffed animal instead of a nursing assistant, isn’t that an admission of weakness? In the book, you also mention the use of an AI in a maternity hospital to predict deliveries and avoid bed saturation. To what extent can an AI replace a human? Where should the limit of what is ethically acceptable be placed?

We have to keep control by what is called the human guarantee. It’s essential. A patient will always need a person by his side, not an algorithm. I ran a digestive oncology service in Paris. I have seen patients who arrive for consultation having done research on the Internet beforehand. They had general statistics in mind without knowing where they were, without knowing where they were, without knowing to what extent they were affected.

It’s the same for AI. An algorithm can confirm a tumour, define a treatment, and tell a patient that their cancer survival rate is 30% over five years. He can’t do anything more. This is why AI will never replace the doctor. Only he can empathically explain the diagnosis. Only he can do a clinical examination. There are limits that should not be crossed. This is one.

You cite the example of the Parisian hospital Saint-Joseph. I’m not shocked by his use of AI. It’s not directly for care, it’s for logistics, of course. But this is not a shortcoming. Isn’t that better than doing nothing? We know that births follow cycles. At certain times of the year, the maternities are more congested than at others. It seems relevant to me to organize staff management using AI with this in mind. You might as well use caregivers wisely at the right time. In the end, everyone wins, especially the patients.

An AI knows neither fatigue nor mood swings and never takes a vacation. Geographical distances do not scare him. To what extent can this technology fill the medical deserts?

In France, we are just starting to use telemedicine, which boomed at the height of the pandemic. But without a doubt, digital and AI are two valuable tools to improve access to care in medical deserts.

In my opinion, one of the most interesting approaches concerns gynecology-obstetrics. Olivier de Fresnoye, who is co-director of this book, is developing portable ultrasound probes that can be connected to a smartphone (for future single mothers, editor’s note). A nurse can send the images remotely to a center, where they are analyzed by an algorithm that clears the ground or directly by a doctor.

Tomorrow, the doctor will not be automatic, it will be increased. In hospitals, care professions, which concern agents in contact with patients, will not disappear, unlike purely administrative ones.

The technology is there, or almost. But are the French ready to “consume” AI? To consult a prevention algorithm rather than a doctor when they are sick?

The sociologist Gérald Bronner (with Laurent Cordonier, doctor of social sciences, editor’s note) explains it very well in the book: confidence weakens when one has the impression of being dispossessed of control of events. If we want the French not to be afraid of it, we have to explain, to demonstrate pedagogy. This book, which is not intended only for specialists, is part of this popularization approach. I repeat: be careful not to make people believe that AI can do the doctor’s job. I don’t believe it for a second. She helps him without replacing him. Tomorrow, the doctor will not be automatic, it will be increased. At the hospital, the care professions, which concern the agents in contact with the patients, will not disappear, unlike those purely administrative.

So the myth of Frankenstein’s creature taking power over its creator has lived…

Absolutely. AI is an advisor, an aid, an extraordinary tool in the hand of man. She will not take power over humans. She is perfect. Take the Watson software for oncology. The IBM firm is at the origin of this diagnostic aid software. Ten years ago, Watson was considered a small revolution. But today we are back. IBM designed its algorithm based on data from an American hospital, collected from a geographical area that was too small to be representative. AI is not invincible and without empathy, awareness, emotion and above all common sense, it will not replace humans. We must stop thinking like this if we want to move forward in the direction of progress.

We wish to give thanks to the writer of this short article for this amazing content

Bernard Nordlinger: “Artificial intelligence will not take over humans”

Take a look at our social media profiles and other related pages