The prospect of ChartGPT and how to optimize its application
By Bilyamin Abdulmumin
While toggling between TED videos, my mouse hovered over a talk titled “Can AI Catch What Doctors Miss?” which I decided to watch. The talk was given by a revered Cardiologist, Eric Topol. He espoused the possibility of AI surpassing expert physicians. To lend credence to his argument, he showed a picture of a retina to which he posed a question: can retina experts identify the gender by merely looking at this retina? He then replied that the expert chance of getting it correctly is only 50 per cent, while AI has an impressive chance of 97 per cent. So he argued if an AI could have such high precision compared to humans, how about relying on them to spot some diagnoses doctors have been missing? Dr Topol buttressed the idea with real examples.
A six-year-old boy suffering from an undiagnosed disease would shuttle between 17 experts, but for three years, his condition could not be diagnosed. So a decisive inspiration came to his mum, and she decided to consult ChartGpT by describing all the signs and symptoms she observed from the boy, and within a blink of an eye, the AI would successfully diagnose the condition: “occult spina bifida”. After the doctors worked on the result, the boy became “perfectly healthy well”.
Topol finally closes his remark by recounting his conversion with an apprenticeship student one day after seeing patients together. He reflects the student how lucky he was to practice during an AI era, as he will connect with patients in a way they could only imagine: the help of getting assistance for diagnosis.
This talk reminds me of my own mental note about ChartGPT—the prospect of the chatbot against the then-savior software like Grammarly, Quibolt, Google, etc.
Just as the advent of the phone rendered a plethora of gadgets obsolete—such as cameras, wristwatches, calendars, radios, recorders, calculators, tape players, speakers, torchlights, TVs, etc—the introduction of ChatGPT is poised to usher in a similar transformation. ChartGPT can make many tools redundant, including Grammarly, Quillbot, Turnitin, Google, Wikipedia, Britannica, and many more.
Crafting error-free text was a significant challenge for young writers: the intricacies of grammatical structures were daunting, especially for us bilingual individuals. But the advent of Grammarly, then, was revolutionary. With Grammarly, writers found relief from the hassles of English grammar intricacies. They could focus on generating a draft, confident that Grammarly would do the rest. But now, ChartGPT, in a similar vein, emerges as a game-changer.
Quillbolt is a six of one and a half dozen of another with Grammarly.
The quest for originality is paramount, mainly when composing a manuscript where maintaining a high degree of uniqueness—often limited to around a few per cent—is crucial. Authors submitting manuscripts endure a nerve-wracking moment as their work undergoes scrutiny through Turnitin software, hoping for a positive outcome. But with the introduction of Quillbot, much of this drama was alleviated. However, ChartGPT takes this capability to another level.
A mere decade ago, the idea that Google, with its dominance in information retrieval, could be overshadowed by a newer invention would have seemed far-fetched. Enter ChartGPT, and suddenly, the status quo is challenged.
While platforms like Wikipedia and Britannica offer ready-made articles, ChartGPT takes interaction to a whole new level. It doesn’t just provide answers; it collaborates with users to deliver precisely the information they seek.
Another hassling undertaking is programming. Coding used to be akin to a jewel in the crown or an elusive skill that felt like searching for a needle in a haystack. Few possessed the ability, and its intricate, time-consuming nature made those with such expertise exceptionally rare. However, for the proud coding masters, ChartGPT has emerged as a liberator. It can unravel the complexities of coding and teach it in a non-specialist manner. With ChartGPT, the once formidable barriers to coding knowledge are dismantled.
As Professor Topol reflected, I have also been in awe of ChartGPT transformations. Encountering it during my student years feels like a true blessing. Its impact on reshaping the educational landscape, especially in 2023, is remarkable. Being a student at this transformative juncture allows me to witness and reflect.
For optimal utilization of ChartGPT, however, a strategic approach should be involved. One has to initiate one’s efforts by crafting a draft before turning to AI for assistance. For instance, when comparing a request for a 250-word article on climate change without a draft to someone who provides a 50-word draft for improvement, the latter receives a more refined result. For even finer results, breaking down tasks into smaller components yields superior outcomes.
The same thing applies to coding with ChartGPT. If you ask ChartGPT, for instance, to generate a code for a specific task, there’s a likelihood that it might not run perfectly, leaving some details for you to fill in. However, if you start by writing your own code, even if it doesn’t run initially, and then turn to ChartGPT for assistance, it can swiftly identify and correct the errors.
Bilyamin Abdulmumin is a doctoral candidate in Chemical Engineering at ABU Zaria, a public affairs commentator, and a science writer.