If you thought Gutenberg’s development of the printing press in the fifteenth century was the greatest advance in human technology, wait till you see what artificial intelligence is going to do.
Somewhere in the 1970s and ’80s, machine intelligence emerged merely as a curiosity in analyzing tons of complex data. By the mid-1980s, the field fell into disfavor only to emerge in the last few years as another outstanding technical advancement.
In short, our entire world is being transformed—and you may not even be aware of it.
Sure you see robots cooking french fries, and maybe you use tools such as spelling and writing checkers when you draft a letter on your computer. You also speak daily to bots answering phones and directing your call.
But I’m talking about the more scary kind of artificial intelligence that is pitching products, mirroring your voice on a video and even your image. A physician I know found himself online in a false pitch for dermatological products.
Fake is still fake, and none of this is what we might consider to be real. But that’s the dark side of AI.
Here’s another example. Amazon has created an array of voices for you to choose to read your book and create an audiobook. In seconds. My coauthor just chose an AI voice for her short book and had the audiobook version up on Amazon, for sale, within minutes. Totally free (compared to the cost of hiring a professional voice, unimaginable a year ago).
The reading voice is so good with intonation and pronunciation that most listeners would agree the production is “good enough.” Not a metallic robotic voice at all. You can even choose a British accent if you wish.
One more: ChatGPT added a million new users in just five days. Maybe these were high schoolers looking to AI to produce a term paper or a patient looking for treatment information in a much more sophisticated version of a Google search.
In my world of medicine, I saw a medical rock star, a leader in the field, address the issue of AI hallucinations. This was about a year ago.
He showed us, in the audience, what happens when the prompt was something like this: Can I use [drug X] to treat my insomnia? The drug name he entered was a commonly prescribed blood thinner. The AI did not know and the response came out as nonsense. But an uninformed patient could be led down a rabbit hole of disaster. At that time, we physicians felt, of course, that AI would not replace us.
That was then.
But now fast-forward a little more than a year from that silly demonstration. When the same question was asked of an AI platform, a very authoritative and carefully documented statement came out explaining that there was absolutely no role for a blood thinner in the treatment of insomnia. This is the kind of statement that could have taken considerable time for a human to write. A response generated by machine learning in seconds.
Let me give you another example. A global authority in a certain medical area quickly dictated a very credible case with age and gender of a patient with abnormal liver function studies. This was a demonstration at a medical conference. Laboratory values in the clinical scenario were analyzed by the software and the question was asked about what exactly the diagnosis was and what were the options and what could be the next step for the patient.
The information that came out of the smartphone was factual, realistic, and the visiting professor made the comment that this was equivalent to what we call a “curbside consultation,” which is what doctors do in medical centers over a cup of coffee or hallway chat.
Interestingly, when ChatGPT was compared to a group of clinicians, the computer was documented as showing more empathy compared to the human.
The day is here when complex decision-making can be clarified by putting into a large language model the specifics of a patient’s management, results from imaging interventions, and a blood study. Then the computer will provide a definitive diagnosis or treatment plan, and it may channel decision-making down a particular road, which a human may not appreciate because we have limited ability to process this information.
Why? Because each day we are “hit” with the equivalent of more than 150 daily newspapers.
The question always arises whether this technology replace the cognitive judgment of the physician. And the resounding reply is no. It will always take an experienced healthcare provider to read through the output from artificial intelligence to decide if the judgment is reasonable. Along with a hands-on examination of the patient.
With the feeding frenzy among small start-up companies, there is no doubt that this technology will transform the delivery of healthcare. Stay tuned, the future is exploding at a blistering pace, not just in medical care delivery, but in every aspect of our lives.
[This blog was written by a human, edited by a human, and titled by AI.]
Comments