The digital revolution is disrupting the established order, provoking discussion, giving rise to concerns and hopes, and dividing society into enthusiasts and sceptics. Let’s discuss the future of artificial intelligence (AI) in healthcare.
YES for AI
It is already present in our daily lived. It suggests books that we may enjoy, controls production processes and autonomous cars, writes newspaper articles, and composes music. It wins chess tournaments, but it can also diagnose patients and select individual therapies. The best known example is the IBM Watson supercomputer, used in over 150 hospitals around the world. This is merely the beginnings of a vast change that healthcare has to face. The list of potential AI applications in medicine continues to expand. Chatbots like Ada or Woebot serve as virtual physicians, diagnosing illnesses based on a series of questions (and lab test results, if necessary). All it takes us is to launch an app – and this new generation of health assistants based on AI is available 24/7, anywhere in the world with access to the Internet. AI systems absorb huge amounts of information from worldwide medical literature resources, far beyond anything that a human is able to memorise, improving themselves with every diagnosis. AI experts have no doubt that, in the not so distant future, machines will exceed humans in their ability to think and solve complex problems. Even now, their processing power and memorisation abilities are overwhelmingly in their favour. Medicine is a highly complex science. Uncomplicated illnesses, such as sore throats or flu, can be detected by asking the patient a few simple questions. Rare diseases require the analysis of hundreds of parameters, not just health-related, and a thorough familiarity with the patient’s medical record, not infrequently covering their entire life. What takes doctors several days, weeks or years, an artificial intelligence system can do in a few seconds or minutes. Another reason that mankind needs AI in medicine is the insufficient number of doctors, difficulties in accessing healthcare services for some groups of patients, the rapidly growing amount of data that must be analysed, and increasing expectations from patients used to digitalisation in other sectors.
NO for AI
The level of “intelligence” in artificial intelligence systems depends on the level of algorithm refinement, as constructed by humans. And these may be flawed, so that an erroneous diagnosis process could be repeated thousands or millions of times by the system. This is not the limit of the reasons for scepticism towards AI. The patient is more than a mere set of data to be analysed like a mathematical formula, but is above all a human being in need of empathy, psychological support, and conversation with the doctor. A diagnosis requires a comprehensive analysis of the patient’s condition, and in addition to access to laboratory test results, the physician observes the patient, their behaviour, movements, and appearance. In this way they can tell a lot about the patient’s emotional and mental condition. Such a holistic attitude is likely to remain beyond the grasp of AI for a long time, creating the risk of the approach to the patient becoming mechanical. Another matter is the certification of such decision-making systems, which may decide the course of a person’s life or their death. We need to answer difficult ethical questions related to liability in the event of medical errors. One can also imagine the threats posed by computer virus attacks on AI systems, with algorithm tinkering set to become a new field of cybercrime. What will be the role of medical workers in the future if everyone has their own all-knowing doctor of every possible speciality in their smartphone? Will human doctors be reduced to validating the decisions of machines? Is artificial intelligence an adequate response to medical personnel understaffing? Would it not be better to seek solutions in the healthcare system, instead of technology? While for minor illnesses, AI can take the burden off the doctors’ shoulders, communicating such grievous diagnoses as cancer or mental illness requires compassion and emotions that remain in the exclusive domain of humanity. These cannot be programmed even in the smartest of systems, as they are characteristic only of members of the same species – humans.
And what do you think? Feel free to leave your comment!