AI and Healthcare
What if, at some point in the future, your next physical is done via a holographic physician? The hologram will be chosen specifically for you, based on age, gender, and other characteristics collected from your posts on social media, websites you’ve visited, music you’ve listened to, shows you’ve watched on television or whatever replaces that in the future. Even the voice will be one that’s chosen just for you. Within seconds, maybe a few minutes, you completely forget that you’re not talking to a real human physician; you’re talking to something created with artificial intelligence. He or she is now your “physician.”
Another article I read had nothing to do with the medical profession. It had everything to do with the development of artificial intelligence (AI) by Google. It could take decades to get to any semblance of AI at this point although there’s an acceleration in technology that seems to happen. The question is whether it will be a good thing or not.
Certainly, there will be a vast amount of data collected from you: blood and urine for sure. DNA as well as the microbiome and maybe things we haven’t considered before such as odors. Computers will be able to analyze billions of bits of data in milliseconds. They can scan obscure medical journals and everything ever written about any condition. But then comes the real trick: the ability of the machine to put that all into a diagnosis and if necessary, a treatment plan.
Could AI do all that? Sure, based on just numbers and data and probabilities. The problem would be this: “Wait a minute. What I’m seeing while examining this person doesn’t match these numbers.” Or “there’s more going on here than what I see in the numbers.”
That’s what a human physician, with years of training, will be able to do. I can’t see a computer being able to do that. Instinct or even a gut feeling just doesn’t work in the computer world. No matter how many calculations that they can perform, and even given that AI can learn, they just will not have the neural network to be able to do that. In addition to that, how do you program empathy? How can you have confidence in something that’s not real?
One more thing: How do you program, or in this case, teach a machine to ask why something doesn’t make sense? Riley can ask me that question a hundred times in a day, but a machine? How can they learn to do that?
For as something as complicated as this human body is, with trillions and more of interactions, we can use the best data possible. Let’s leave that to the machines because that’s what they can do. And other than cases where people are too remote to see a live doctor, let’s leave healthcare diagnosis and treatments to humans.
What are you prepared to do today?
Dr. Chet
References: Fast Company. October 2019.
