This piece, “Intelligence in Healthcare: Artificial and Not”, appeared on page 25 of the Summer 2023 issue of Chicago Life Magazine.
Whether artificial intelligence makes something better or worse depends, as it does for any tool, on how it’s used. The potential for intelligently deployed AI to improve health is enormous. But it is not about to replace human beings.
It can be viewed using this link: Chicago Life Magazine - Intelligence in Healthcare: Artificial and Not
--
I dare artificial intelligence software to generate the following dialog, which I was privileged to hear when I was a medical student. It happened during morning report, the traditional start of a doctor’s day in a training hospital. Each admission from the day before is presented to the gathered physicians and students, usually by a bleary-eyed, post-call resident.
Case presentations follow a general structure that is imprinted into the cerebrum of every doctor early in training. A typical recitation would start something like this, “This 37 year-old African-American male presented to the emergency room at 8PM complaining of abdominal pain of 12-hours duration.” Then the resident would move on to “History of present illness,” “Past medical history,” “Past surgical history,” and so forth.
I have not the slightest recollection of the facts of a new admission discussed one morning of my surgery rotation at the West Side Veterans Administration Hospital in Chicago. But I’ll never forget how the intern presented the case, totally routinely until he got to “military history,” a datum mandated for armed services practices that is rarely of any medical import, including in VA hospitals. Andy filled in the blank mumbling, “G.O.K.G.O.C.”
An alert attending physician stopped him, asking, “What was that, doctor?”
“G.O.K.G.O.C. God Only Knows. God Only Cares,” replied the humble intern. This got a big enough laugh to allow Andy to get through the rest of his presentation unscathed.
Maybe I’m naïve, but it’s inconceivable that any machine would ever come up with Andy’s answer, let alone deliver it with such perfect timing. Artificial intelligence doesn’t do humor so well, not yet anyway. Though it may fake them, AI doesn’t really do human relationships either.
On the other hand, these days AI is doing many other things better and better. Everybody’s talking about the potential of machine intelligence to revolutionize all sorts of scientific, business and artistic endeavors. On the dark side, the ever-more-convincing deep fakes of human writing, speech, images and videos that AI programs can generate pose a very real threat to the trusting human relationships that are the warp and woof of our lives and society.
I want to add my two cents to the conversation, not because I have much new to say about the promises and pitfalls of these amazing technologies, but because I want to put artificial intelligence in a larger context that might help us to understand and to harness it for healthcare in ways that could actually make us healthier.
I’ll start with my widest angle lens, philosophy. Plato, a philosopher who lived in Athens in the 5th and 4th centuries BC, railed against the written word. He understood what was lost when things were recorded for future reference rather than communicated face-to-face. Written language could dispense with the rhyme, melody, gesture and repetition of oral performance, crucial aids for remembering stories and instructions. Gone too was the juice of a live audience. (Nevertheless, Plato himself wrote a whole lot of stuff that people still study today.) Culture, politics, even human consciousness, were altered irrevocably by the written word. Nobody can say if we would have been better off if we’d stayed in the poetical, musical, intensely personal world that preceded writing. I do know that my life is unimaginably richer thanks to this medium that first appeared 5-1/2 millennia ago in Mesopotamia.
Let’s jump way ahead to moveable type, the telegraph, telephone, radio, movies, television, computers and the internet. Every one of these new communication technologies brought cataclysmic changes. And in each case the culture has incorporated the new medium, with mixed blessings. Consider, for example, the blow to live performance that came with the advent of radio, tv, movies and streaming on the one hand and on the other, the enormous range of artistic experiences available to us now that were simply beyond imagination before these technologies appeared. Today we’re faced with a new medium, artificial intelligence, in some ways more powerful than any that has come before; able to write text, for example, that keeps getting better at not only passing for human, but actually adding content that enhances both understanding and the power to communicate it.
There have to be ways to use AI to produce doctors’ reports, for example, that convey more than today’s documents, largely clumsily generated from templated checklists or copied-and-pasted from previous notes, full of tons of extraneous words, repetitions, misinformation and just plain nonsense. It would be of immeasurable value to have real, well organized, understandable medical information at the fingertips of every person who interacts with a patient; all of it deepened by machine learning that recognizes patterns and produces diagnostic insights and therapeutic suggestions that may never have been generated by the unaided brain of a human practitioner. As a physician, I will embrace every bit of help that artificial intelligence can give me.
The fears I may have about being replaced by a very smart machine don’t really originate with the gizmo itself. I’m more concerned with how the human beings who manage these technologies choose to employ them. AI should free clinicians from doing the tons of documentation that today’s electronic medical records require of us. It remains to be seen what practice administrators will ask us to do with our newfound time after, with the help of AI, our eyeballs have been unchained from the computer monitor; hopefully not just see more patients in a workday. To a person, my colleagues and I want to spend more time with patients, less distracted by computerized alerts and requests for data. That’s what patients want too, overwhelmingly.
As a society we have managed so far to use writing, the telegraph, telephone, radio, and television sometimes to impede but mostly to enhance our inter-connection. I expect clinicians eventually will learn how to relate better thanks to AI, not in spite of it. I’m hoping that not too far in the future artificial intelligence will make medical practice more powerful, more satisfying and more healing. But at every step of the way we’ll have to be vigilant that this technology is employed in ways that enhance the power of us humans to do what we do best, which is to really understand and care for each other.
We need to ask continuously what effect any information system, AI-driven or not, is having on practice and on relations with patients. And we can never accept “G.O.K.G.O.C.” for an answer.
