Translator: Gisela Giardino Reviewer: Sebastian Betti There is something that in the coming years, can change work completely as we know it today. There is something that in the future can improve substantially our quality of life and our health. But there is also something that could annihilate us as human beings.
Those three things, work, health and death, actually rely on only one: artificial intelligence. But what is that thing so many people talk about? What is artificial intelligence?
First, we should define what is human intelligence. And if I asked you what is human intelligence, I would probably have around 10,000 different answers. So, how do we define artificial intelligence if we don't even have a definition for human intelligence?
Alan Turing, one of the fathers of computer sciences, came forward with a very clever solution, more than 50 years ago: It doesn't matter what intelligence is, we will call it artificial intelligence when a computer manages to replicate human behavior, and does it to the point that when we interact with that computer, we will not be able tell if it's a person or a machine. This definition achieved something incredible. Generations and generations of researchers sought out to replicate human behavior in many different domains.
For example, if we play chess on the Internet, we can hardly know if our opponent is a machine or a person. Today, computers replicate human behavior both in their successes and errors. That is, computers play good or bad and adapt like humans.
But there are many cases where the computer even exceeds human performance. And chess could be an example of this. But let me tell you another example.
Not long ago, we believed that humans were unbeatable for detecting faces in photographs. If I showed you two photos of people, that could be shot in different lighting conditions, looking in different directions, even with different clothing. One with a hat and the other without, and I ask you: are these two people the same?
And I repeat these questions again and again with different pairs of photos, you would be right around 97 percent of the time. Do you know what? A few months ago a computer exceeded this mark, by achieving up to 99 percent accuracy.
What is it that makes computers so powerful, so smart? First, computing power. It's impressive the speed at which computers make calculations.
Second, memory. The computer remembers everything, and today the data abounds. And third, artificial intelligence software, having evolved a lot recently, now manage to detect similarities, differences, repetitions in large data volumes.
We used to say that computers were obedient, that they did strictly what we programmed them for, like following a cooking recipe. Today, these artificial intelligence programs succeed in finding results, conclusions, without being explicitly programmed but by performing an automatic analysis of the data. Can this be dangerous?
Can this artificial intelligence become aware as Terminator's Skynet and decide to destroy the human race? That is one of the pivotal debates today in the world. How far we come with the advance of artificial intelligence?
There are those who consider this a dangerous technology. Bill Gates, for example, the founder of Microsoft, or Elon Musk, one of the most relevant innovators nowadays, directly suggest stopping the development of this technology until we understand it and regulate it, and so prevent it from becoming unsafe and take over the control. On the other side, Mark Zuckerberg, the founder of Facebook, considers them irresponsible and says: "In no way it is insecure.
It would be absurd to stop the development of this technology that has such an impact in daily life. " This debate touches me particularly closely. I am a researcher in artificial intelligence applied to medicine.
And it's in medicine precisely where these three problems, work, health and death, lie in. Probably all of you or a relative, went to E. R.
for a severe headache. Imagine the situation. You feel awful, you decide to go to the hospital, ask for your turn and wait one, two, three hours in the waiting room.
Finally, the neurologist on duty interviews you. While he makes questions, writes down in a medical record and eventually gives you a diagnosis and proposes some treatment or further tests. But the doctor is human, he can go wrong.
We go wrong all the time. You could be the twenty third patient on a Wednesday at 3 am. There is where our system intervenes.
In association with computer neurologists we created an artificial intelligence system that reviews hundreds, thousands of medical records from many patients. And looks for repeated words, what symptoms are associated with each of the diagnoses for headaches. And so, when that doctor is writing down what he listens from the patient the system can validate that the diagnosis he is giving matches the words, the symptoms, that he is writing down in that clinical record.
And how accurate is our system? To validate that, we asked some neurologists to do the same computer work. That is, to review past medical records and guess the diagnosis from what was written down.
These neurologists reached a 80 percent precision in their diagnosis. Today, our artificial intelligence already exceeds 90 percent of diagnosis with respect to past medical records. And when we saw this, we said: let's go for more!
Could we also make the interview automatic? And then we tried a first version of a virtual doctor who asks a series of questions based on the diagnostic criteria that neurologists use -- and we make use of those three hours in the waiting room -- then tries to guess the diagnosis and later checks if matched or not with the doctor's. This very first version that was just released, is already guessing 85 percent of the cases.
And this raises again this vital debate: how far do we go with the advance of artificial intelligence? Notice that we had started with a warning to the doctor that it was obviously something that could help health and the quality of life, we followed with a virtual doctor and we are just one step away from suggesting a treatment automatically. And this keeps moving forward.
Are we going to replace doctors? I don't think so. I believe that in the coming years we'll see an impressive development in this kind of artificial intelligence systems that will give the doctors new tools to do their job.
And that their job is likely to change, to be probably much more oriented to the relationship with the patient, to that human characteristic that technology will never be able to replace. Can you imagine a doctor today diagnosing without a CT scan, without magnetic resonance or without X-rays? Perhaps, in a few years, we find it unimaginable that a doctor, a human being, could make a life or death diagnosis without the help of artificial intelligence.
We work every day to generate those kinds of technologies, safe and useful. And like with all powerful technology, this brings huge benefits, but also some risks. I don't know how this debate ends, but what I'm sure of, is that the game is played there.
Thank you very much.
