The Medical Futurist | April 30, 2019
In spite of common fears and perceived dangers, artificial intelligence won’t take the jobs of physicians and won’t monopolize medicine. On the contrary: as A.I. will help automate administrative tasks and take over monotonous day-to-day assignments, it will free up time for medical professionals to finally fulfill the mission they signed up for: to help people on their journeys to health with compassion, creativity, and care. Moreover, rather than wiping it out, A.I. will, in fact, facilitate the art of medicine. Let us show it through the example of the AlphaGo documentary.
Artificial intelligence and the talent of cats
Experts have already achieved marvelous results with artificial intelligence – sometimes even nearing the predicting skills of the finest cats. The New York Times, in its account of whether computers could sense death, recounted the story of a 2-year-old black-and-white cat named Oscar reigning over one floor of the Steere House nursing home in Rhode Island, who was apparently better than most doctors at predicting when a terminally ill patient was about to die. Since then, Researchers at Stanford University trained an A.I. system to increase the number of inpatients who receive end-of-life-care exactly when needed – meaning the smart algorithm is able to predict when very seriously ill patients are nearing the end of their lives.
However, that’s just one sign pointing towards the phenomenon that A.I. will completely move the world of medicine. Hugh Harvey, radiologist, and clinical academic estimates that in 10 years using A.I. will be routine in NHS practice. Anna Fernandez, Health Informatics/Precision Medicine Lead at Booz Allen Hamilton told The Medical Futurist in 2017 that within three years, we would have many machine learning algorithms in active clinical pilot testing and approved use in the US.
Could the art of medicine be threatened?
As smart algorithms increasingly become better, many physicians are intimidated by their potential. Numerous experts voiced fears about A.I. taking their jobs and destroy the profession. They say the art of medicine, the creative process of understanding the uniqueness of each and every patient and tailoring treatments according to the arisen needs, as well as the highly value-added method of processing inputting information, finding the right response and treating patients accordingly, might disappear due to new technologies.
I’m completely sure that A.I. will transform the meaning of what it means to be a doctor: some tasks will disappear, while others will be added to the work routine. However, there will never be a situation where the embodiment of automation, either a robot or an algorithm will take the place of a doctor. Also, there is no reason to believe that A.I. would turn the very core of medicine, the art of healing, into a machine-like clockwork. Quite on the contrary. Let me show you the example of a documentary, AlphaGo, to prove you that the art of medicine is not going to vanish, in fact, it’s only getting started.
More potential configurations than atoms in the universe
The Chinese game, Go, is one of the most complex and elegant games ever invented. That’s why I never understood why computer scientists had for decades viewed chess as a meter stick for artificial intelligence. Unlike chess, where at any given moment in the game, the number of moves approximates thirty, on the 19×19-sized Go board, players can have around 200 potential moves. And the number of possible configurations on the board is more than the number of atoms in the universe. So if the victory of IBM’s Deep Blue over Garry Kasparov in 1997 was huge, then a win from Google Deepmind’s AlphaGo over the 18-times Go World Champion, Lee Sedol would be gigantic, colossal, massive, monumental, gargantuan? I’m searching for the right word, really, but it’s just beyond expression.
That is because Go requires more than computational power. It needs creativity and intuition, human skills that, many believe, are beyond current artificial intelligence technology. So, let’s set the stage: we are in the year of 2016, and Lee Sedol prepares to sit down for the first time against AlphaGo. Almost 80 million people gather around their TV screens to watch the match. The odds were against AlphaGo – everyone expected an easy win from Sedol. But the bookmaker’s shops lost tremendously. As the machine beat him, the world champion was visibly shaken. How did it happen?
The secret is reinforcement learning and neural networks
The key to success was in the way AlphaGo was trained to play the game. The machine that beat Kasparov learned everything from how humans play chess. Deep Blue had general-purpose supercomputer processors combined with chess accelerator chips. A software ran on the supercomputer to carry out part of a chess computation and then hand off the more complex parts of a move to the accelerator, which would then calculate possible moves and outcomes based on a pre-determined set of databases.
On the other hand, AlphaGo uses something of a combination of neural networks and the method of reinforcement learning. In a nutshell, the latter means that developers don’t tell the algorithm how to play the game. Instead, they describe the rules of the game and let the algorithm play millions of games against itself. Later, they tell the algorithm which outcomes were the desired ones (the matches it could win), and let the program find its own strategy and rules of how to achieve those outcomes.
It simply means that the cognitive limitations of people (in this case, the developers) are not programmed into the algorithm. This way, it might find brand new ways of playing the game even though it has been played by millions of people for thousands of years.
Over the long course of its evolution, humanity has accrued beliefs and mechanisms by which it views and understands its surroundings. Artificial intelligence will not be “burdened” by these preconceptions. Humans could discover that when they watched AlphaGo play. It was able to surprise everyone with a move that nobody else thought of. That’s because the advantage of reinforcement learning is that the cognitive limitation or the learning curves of humans are not built into its algorithm. So it didn’t inherit the boundaries and flaws of human intelligence. It was free to find new and exciting ways to beat the game, things we wouldn’t even think of.
Sedol and AlphaGo played four more times against each other and AlphaGo won three of those. And for example, in game two, AlphaGo made moves that left experts wonder.
Take away the potentials
How did Lee Sedol, the beaten champion react after the groundbreaking match? At first, he was deeply disappointed and felt embarrassed. Just imagine, how it must have been; his ego was shattered, his whole existence as an undisputed champion came into question. He had to play Go like it was his first time, he faced moves no human opponents would’ve played. And most importantly, he felt like he’s representing all humans in a losing battle against something otherworldly.
However, by the fifth game, they played together something happened. It was visible that Sedol had a newfound understanding of what AlphaGo is and who he is. The world champion started to analyze the new ways AlphaGo found to play this inventive and endlessly creative game. And by doing that he became a much better player. He basically stood on the shoulder of technology in order to develop. And that’s exactly how the medical community should look at the digital health revolution as a whole.
The physicians’ guide to the galaxy
In the next years, artificial intelligence will find new drugs, new treatments, and therapies through matching combinations that human physicians, pharma companies or medical innovators would never think of. As it won’t be limited by the traditional pathways and thought patterns used for centuries in medicine, artificial intelligence may come up with entirely new solutions – without telling humans how it figured these out. Similarly to the supercomputer in Douglas Adam’s The Hitchhiker’s Guide To The Galaxy, where the answer to life, to the universe and everything else is 42, smart algorithms might just spit out answers to questions without explanation. The real art of medicine will thus be the undertaking to figure out the logical path of how the A.I. arrived at a certain solution. That will definitely need the high levels of creativity, problem-solving and cognitive skills that the medical community possesses.
Thus, we are sure that A.I. is not going to replace us; it’s going to be the stethoscope of the 21st century. Digital health will give us more health data than ever before, and A.I will help us analyze it to find new ways to treat diseases, to cut down on administrative tasks, to streamline medical practices, to optimize both physicians’ and patients’ schedules. However, we should never forget that they are going to be tools in the hands of physicians – and not the other way around. Compassionate care, empathy, creativity, problem-solving and profound human connection will never cease to be the terrain of physicians, moreover, it will be enhanced by artificial intelligence.
As we said it before, if we embrace these tools, the real art of medicine begins with the era of artificial intelligence.