Artificial Intelligence: The Future of Medicine
Thumbnail from: Wix Media
As technology progresses, we see the rise of artificial intelligence; integrated into our phones, cars, thermostats, and even some house appliances, artificial intelligence appears to be the future of many businesses. Experimental technologies with AI are being developed all over the world to serve many different purposes, including for medicine.
To talk about the future of AI in medicine, I had the chance to interview aspiring computer science and AI student Emir Durakovic.
Do you want to write articles just like this one? Apply to become a contributor on the Contact & Apply page!
How is AI being used in medicine and daily life currently?
Artificial intelligence is currently being used for more accurate diagnoses and prognoses. Using past data, computers can accurately approximate the future health of a patient, or what the likely condition the patient has, given a list of symptoms. Using this, doctors can suggest better treatments or medication. Also, many computers can take in an input of symptoms and conditions and output an accurate model of the future well-being of someone.
In daily life, we have AI running our schedules like Amazon Alexas, Siri, etc. Nest and other companies have developed fire detectors and thermostats with AI that can sense temperatures and act accordingly. Researchers even point to, in the future, a so-called "internet of things", where every appliance, car, and device in the world will be connected to each other and to humans as well.
Here's what AI and computer science student Emir Durakovic had to say on artificial intelligence and its future uses in medicine.
What is AI, and why is it the future?
"The keyword is 'Machine learning'. In other words the AI is trained using different examples and it learns to adapt." AI is almost like a brain, being coded into existence by cutting edge languages such as Java, Python, Lisp, and more. At first, it will have almost no accuracy, but its accuracy will increase as it keeps being fed more data. The training process for an AI program varies depending on what the program needs to do. At some point, this accuracy will increase to a percentage close to 100%, which at that point you will have a computer that can take data and output what you want with a close to 0% error rate.
However, with such a foolproof machine, many worry about displacement of humans from their jobs, and replacement with machines. For example, data analysts have less and less work to do, as computers become smarter and smarter about organizing data. Is this a danger? How much do we need to worry about artificial intelligence?
What dangers and disadvantages come from using AI in daily life?
"It all depends on human morals. Basically, if the AI falls into the wrong hands, then bad things will happen. Other potential harms may include AI bias, loss of jobs, etc. " Emir, like the majority of people in the AI field, believes in the ideology that AI, like a weapon, is a tool, that can be swayed to malicious or benevolent intent through the person using it. He agrees that AI can create a large loss of jobs as it progresses, but feels that it is an unsure topic, and we will likely see how it will unfold in the coming years.
A rising issue in AI, however, is an interesting field called AI bias. AI, at its root, can only analyze data that is collected by human observers. When this happens, sometimes AI can show racial, gender, or other bias that was slightly expressed by human error in the data. It can be hard to get rid of AI bias when its root cause is from human error.
Some groups are advocating for AI restriction laws, feeling that AI, like a human, should be tried by the law if it makes a mistake and commits, for example, malpractice. Emir believes that AI shouldn't be restricted, but rather the people using it. For example, if people want to put a missile launch program completely in the hands of AI, there should be restrictions and humans to check and balance the power of AI.
However, if AI itself commits a malpractice or makes a dangerous mistake, Emir feels that it makes no sense to put a machine in a court of law. AI will never be as complex or capable as the human brain, and would never willfully make a mistake. "When giving an AI program pictures of a dog or a cat, it has to learn and improve, without being programmed, e.x. using logic trees. We definitely should restrict how we use AI, in order for it not to fall into the wrong hands. However it does not make sense to impose laws for a computer, as the only way it learns is from its mistakes."
AI in Medicine - How much can we trust a machine to do a doctor's job?
"We have to consider that an AI is not perfect, many errors can happen. In my opinion it would be better for a doctor to treat a patient. However, AI can be useful when finding, otherwise not predictable or seeable, problems. An example of this would be, researches in Bonn have found rare diseases with the use of neural networks and the AI found different treatment options." A neural network is a type of AI training model, that researches in Bonn, Germany were able to use to catch rare conditions. In this way, AI can aid the job of a doctor, but it will be many, many years until doctors are replaced. For more information on this, you can read this informative paper on the topic:
https://www.sciencedaily.com/releases/2019/06/190606133805.htm
In conclusion, as Nvidia, Google, Amazon, and more trailblaze a path to the future using artificial intelligence, it grows ever more important in our lives and in the future of business. We must learn to adapt and incorporate it into our society, and it will be a great tool for millions if used correctly.
Special thanks to: Emir Durakovic, for providing his insight in our interview on AI.
Bibliography:
https://news.mit.edu/2017/explained-neural-networks-deep-learning-0414