An introduction to ChatGPT: Will it change the world?
Editor’s note: This column is the first in a series about artificial intelligence, ChatGPT and medicine.
Several weeks ago I received an urgent call from my son Danny who is a Computer Science student. “Daddy, the world is about to change drastically,” he said, “it could be a catastrophe.” He was talking about ChatGPT–a new artificial intelligence chatbot.
What exactly is ChatGPT? Can it change the world? Will it be a change for the better or a calamity? And how will it affect medicine and the jobs doctors and other clinicians perform?
I have tried ChatGPT and have learned that it can answer most questions, write reasonably good essays, discuss both sides of almost any controversy, and string words into poetry. It can write and find mistakes in computer code. It can compose music. You can, for example, ask ChatGPT to write a short story about a love affair between an apple and an orange, or a love poem to your boyfriend. I have asked it to come up with a name for a book I am writing. It did all of that quite successfully. I could have asked ChatGTP to write this article, and you might have never noticed, but I chose not to because that would be cheating.
ChatGPT is first and foremost a chatbot, a computer program that is designed to simulate conversation with humans. You can ask it a question, present a problem, seek a solution. It isn’t a mere search engine, like Google or Bing, where, in response to a query, you are presented with a sea of websites where your can seek an answer. Instead, it answers questions with confidence, in a clear and authoritative way, as if, behind the computer screen, a group of experts has gathered for the sole purpose of assisting you in your pursuit of high productivity and a balanced life. One problem that might arise: the expert opinions and capabilities of ChatGTP may render your professional opinions and abilities obsolete.
How does ChatGPT work? It uses a few tools, but first and foremost it uses artificial neural networks which are modeled on biological neural networks such as the human brain. The goal is to allow the artificial neural network to serve as an “artificial brain” with the capabilities to learn and make decisions.
How does the human brain work? Here is an example: I am driving my Subaru through a small town in America and toward an intersection with a railroad. There is no traffic light, and no gate in front of the railroad, just a stop sign. My Subaru isn’t equipped with Auto-pilot, or self-driving capabilities like the one installed in Tesla, for example, and I am the one who has to make a decision whether to cross the railroad or to stop. My brain, a highly sophisticated biological neural network, is called into action, and there is only a simple output decision it has to generate: to drive on, or to stop. My brain is equipped with about 100 billions neurons. Each neuron is connected to 10,000 other neurons from which it receives information and to which it delivers information. The connections between the neurons take place within small gaps between the neurons called synapses. To make a decision, my brain has to rely on outside stimuli and on prior acquired knowledge. At the intersection, my eyes constantly collect inputs in the form of images (the stop sign, other vehicles, road conditions, the weather). My ears may search for a distant sound of an approaching train. In my mind, I can hear my driving teacher telling me, ‘Use extra caution when approaching a railroad.’ I might recognize the fear of a horrific accident, and at the same time, perhaps, realize that I am late for work. All that time, minute electrical currents travel along the neurons connecting my eyes, my ears, centers in my brains where my memory is stored, and areas charged with processing my emotions. At the end of this process, a group of cells within my brain will analyze and add up the information. To me, it would seem like a decision I made, to cross or to wait, but in reality, from a purely physical standpoint, the decision is a confluence of electrical currents and chemical reactions; inputs that are gathered from multiple sources, and processed in a “black box” which I do not fully understand, over which I have no control; an output in the form of a simple decision–stop, or proceed.
In April 2022, a Tesla vehicle crashed into a $3.5 million jet. At the time of the accident, the vehicle was using its “Smart Summon” feature which is designed to allow the vehicle to automatically find and drive toward its owner. It is offered (for $6,000 extra) only to those who have already purchased Tesla’s Autopilot, a highly sophisticated self-driving feature that is heavily reliant on artificial neural network. With such degree of sophistication, why didn’t the Tesla stop before it hit the jet? The answer has to do with the differences between biological and artificial neural networks. And it may be crucial to understanding the role ChatGPT will play in our lives. I will come back with more about the subject.
Dr. Shahar Madjar is a physician who specializes in urology.