Millions of students have passed the exam to become certified doctor, so one more shouldn’t make the headlines, unless it happens to be a robot. Actually, Xiaoyi is the first robot to pass the exam. And it passed with flying colours, getting 456 points out of 600 (96 points above the minimum requirement to get the certification).
One should mention, however, that its first attempt to get the certification was a failure (it got 100 points out of 600). The researchers that designed it decided it had to do what any student failing an exam has to do: go back and study.
And study it did: it read dozens of medical text books, absorbed over 2 million medical records and perused over 400,000 medical articles to be up to date on the latest progress in medicine. All of this learning was used by its artificially intelligent brain to develop a way of reasoning (and perhaps a feeling for spotting a medical problem).
So, the second time it passed, and it took it just a fraction of the time it takes a human student to go through the questions.
To me is quite obvious that artificial intelligence is going to lead in the medical profession. And for a very simple reason. The amount of knowledge that is being created in the medical field is already exceeding the capabilities of a single human being. But this knowledge is available on the cyberspace. It is often the case, I am told by some medical doctor friends, that a patient may show an in depth knowledge of his specific pathology, sometimes pointing to some recent discovery that his doctor is not aware of.
According to MedLine the number of medical papers published in 2015 was 806,000, and although I was not able to find more recent statistics it is not an hazard to imagine than that number is larger now. 806,000 papers published in a year means 1.5 papers every single minutes, night and day, every single day. Assuming that it would take 30 minutes to read and digest a paper by a doctor, it means that once she is through reading the paper 45 new papers have become available…
It is obvious that specialisation has become a “must” with this avalanche of knowledge but even that cannot keep the pace of evolution. The only possible solution is to have an artificial intelligence that digest these increasing knowledge and to be smart enough to understand how and when to apply it.
We already have artificial intelligence applications that can outsmart expert clinicians in spotting cancers and identify the best clinical procedure for a cure. It is just a matter of time to get virtual doctors, and a little bit more to trust them. We are already trusting machines in performing blood analyses, scanning our body, make surgery …
Probably, we should aim for a symbioses between human doctor and AI to take the best of both.
By the way, Xiaoyi, in Chinese, means “Little Doctor”…