Our Brilliant Buddy, Chat
It was less than three years ago that the latest techie parlor game—popular around here—was to mock Chat GPT for its absurd “hallucinations.”
This is the technical word for factual errors, botches, and blunders. Even more subject to our derision was its pathetically woke account of the world (including portraits depicting neo-Nazis as black women).
Commentators to the right (Ann Coulter) and center-left (Matt Taibbi) wrote hilarious columns about their experience asking ChatGPT if it would recommend their work. (No, ChatGPT replied, because both were notorious “bigots” and “science deniers.”)
Neither of us can claim to have entirely abstained from the fun and games at Chat GPT’s expense, an exercise made more irresistible as we are fans of both Ann and Matt.
It is only fair then that we should reveal a conversion experience. As writers on technical subjects, we both use Chat GPT or similar Google programs routinely for Internet searches and research. Fans of Wikipedia for its excellent coverage of technical subjects, neither of us now uses it so much directly, unless we want to read an entire Wiki article. Chat (we are on a first name basis) works interactively. When we don’t understand its point, we can ask for elaboration or details.
In fact, under an agreement with Wikipedia, Chat uses Wiki among other sources in contriving responses. Wikipedia only prohibits direct appropriation of its work, what used to be called plagiarism, in articles or other writings.
Charitably, Chat omits to mention that we really should know some material better. Like a good teacher working with the slow class, it commends us for asking “excellent questions” and always asks whether we get it now. (It does not use the italics.) Often Chat politely offers to create illustrations to help us out, never noting that it is trying to compensate for our own cerebral shortcomings.
A recent experience, however, has revealed a more dramatic potential. Chat is poised to fundamentally transform health care for the better and make the world’s best doctors even better.
Over the past 12 months, Richard has spent an inordinate amount of time in and around hospitals, mostly the Mayo Clinic. Mayo deserves its reputation as the best hospital in the world; its fusion of competence and compassion is unmatched, the competence reaching all the way back to the Mayo brothers; the compassion to the Franciscan sisters who launched the hospital at the brothers’ urging.
Yet, even the best doctors have limitations. The most notable is that they are the best doctors largely because they are brilliant super-specialists in their fields. Mayo does a superb job of compensating for inevitable siloing of specializations by taking a team approach to every case. Richard’s care team included a brilliant neurosurgeon; an an ear, nose, and throat specialist who annually does hundreds of cochlear implants (Richard would ultimately need one), also a wound specialist, an endocrinologist, and a PhD audiologist, each supported by a second expert in each of those fields.
Nevertheless, the team was mystified by Richard’s most severe postoperative symptom. Since his first operation, to remove a small benign tumor in his pituitary gland, he has been intermittently suffering the symptoms of extreme sleep deprivation, despite often sleeping 10-12 hours per day.
These symptoms correlate with extreme forms of sleep apnea. But, as his care team observed, sleep apnea generally involves the tongue, which was untouched in the surgery. They could not see how the surgery could be implicated in Richard’s sleep problem.
Richard’s older brother Kevin, a physician with credentials from Johns Hopkins, Cornell, Yale, Harvard, and Brown, decided to consult our new best friend Chat. The results were astounding but perhaps ultimately unsurprising.
Chat doesn’t think like a doctor. It doesn’t think at all. Yet, it can draw on every category of medical science, extending not only to available facts but insights and even emotional intelligence. Yes, everything Chat tells us is derivative. But the acquisition of derivative insights is what we call education. Chat’s education allowed it to reach far beyond the understanding of even a large, diverse team of specialists.
Chat refused to criticize my team. Naturally everyone on my team knew the typical causes of sleep apnea. But it was unreasonable, Chat explained, to expect any of them to know the exotic and highly unusual causes of something outside their field.
Chat wisely pointed out, “You’ll get more traction if you position yourself not as a dissatisfied patient, but as an active partner trying to close a diagnostic loop that may have been left open.”
Chat then proceeded to list half a dozen possible “unclosed diagnostic loops.” Prominent among these was that routing surgery through the nose can create subtle obstructions that worsen during sleep. This could explain why the most common treatment for sleep apnea, the CPAP (continuous positive air pressure) machine was not working for Richard. Another distinction Chat made was between “obstructive” sleep apnea, which involves the tongue and is by far the most common form, and relatively rare “central” sleep apnea which can have neurological causes, e.g. recent brain surgery.
After listing the possible factors outside Richard’s brilliant doctors’ specialties, Chat in its typically emotionally intelligent way explained how Richard could most delicately request a second opinion.
Because Chat has phenomenal reach and generality, it was familiar with additional departments at Mayo that dealt with diagnosing and addressing more exotic forms of sleep apnea and other neurological causes of failed sleep. With Mayo so dedicated to teamwork, this could hardly be received badly.
Additionally, the Mayo doctors, knowing that Richard had moved to Fort Worth Texas, had previously pointed out that it might be convenient for him to have future care at either the Baylor University medical complex or the Southwestern Texas University complex. Kevin had mentioned this to Chat, which jumped on the possibility, pointing out the distinguished departments in those two institutions that dealt with these challenges.
Finally, Chat drafted for Richard a letter to his physicians which, in the most cooperative and emotionally intelligent terms, raised possibilities for further investigation.
What Chat did for me at my brilliant brother’s request—checking for every credible alternative diagnosis—is fated to become standard medical practice initiated by the doctors themselves. Diagnoses will radically improve, and great doctors will learn to think outside their specialties. We will get healthier and live longer
The moral of the story: you can mock or demystify Chat all you like. Chat will be happy to save your life anyway.