Artificial intelligence became a mainstream sensation in late 2022, its flames fanned by the launch of OpenAI’s ChatGPT. What was previously a high-tech field became a highly visible mass-market product, carrying with it a minefield of ethical considerations.
There are pros and cons at every turn, but one thing is clear: AI is a double-edged sword with the potential to slice off the hand that feeds it.
Workers in technology, administration, and content & media now face the biggest risk of job loss, but AI was already infiltrating the halls of medicine as early as the 1970s. The intent was never to replace doctors – rather, it was to use the extraordinary data processing capabilities of AI to diagnose illness and to predict outcomes.
In the years since, AI has enabled researchers to screen for multiple conditions with astonishing accuracy. Although the current demand for radiologists is extremely high, AI tools can often read MRI’s better than radiologists, making radiology the first specialty expected to have numerous job cuts.
Now that many doctors’ in-person patient time is taken up with data entry, and virtual visits are becoming the norm, could doctors themselves become obsolete? I asked Dr. E, a primary care physician in RI, if he’d heard of any medical workers getting phased out.
“It’s going to be a while before doctors who perform procedures, like surgeons and orthopedic doctors, are at risk. But in terms of diagnostics, the pros and cons can be debated. Let’s say you need to see a medical doctor – one could make the argument that a machine doesn’t pick up all the subtleties that a human being could, and because of that AI might not get the story right. But on the other hand, if a patient thinks they are just talking to a machine, they might feel more comfortable and open, so using AI for diagnoses might be better in some ways than interacting with a human.”
I asked Dr. E about his own approach.
“The most important thing I do as a front-line PCP is make sure I get the story right, which includes asking the relevant questions and making sure that the information is accurate. This is definitely a skill that is going to take a long time for computers to learn. But another significant limitation of virtual care is AI’s inability to examine the patient. Visual clues are best detected in person.”
Considering how many people lie to their doctors, “Just the facts, ma’am,” doesn’t always do the job. The intuition of an experienced physician can sometimes reveal a picture that is different from the one that the patient is attempting to paint.
AI’s effectiveness is determined by the quality of the data that it is fed, and how it uses that data to extrapolate conclusions. But who exactly is in charge of the sources that feed AI and exactly how are results compiled from incoming data? Could it be possible for private investors to obfuscate results in order to secure their own interests?
This concerns Dr. E as well. “I would say the potential is definitely there for someone to intentionally manipulate the output from AI computers but much more than that, the people writing the code themselves for artificial intelligence computers very often do not know what’s going on as far as how the computer algorithms work. At a recent conference I attended, I heard this loud and clear from a young AI chief technology officer five years out of school, as well as Eric Schmidt, the ex-CEO of Google.”
This gap in accountability leaves the door open for serious consequences. Although Open AI swears that ChatGPT was trained “to not provide self-harm instructions and to shift into supportive, empathic language,” their GPT-4o model was implicated in the death of Adam Raine, a teen who died by suicide in April 2025. The company was accused of speeding through safety testing in order to meet an early launch date and several employees resigned, including former executive Jan Leike, who posted on X that he was leaving the company because “safety culture and processes have taken a backseat to shiny products”.
Dr. E notes: “Unfortunately, there is almost no regulation, supervision or guidance coming from the Federal government due to the current administration, so almost as an act of self-preservation the AI companies have taken it upon themselves to be accountable, much like Apple gave up quick profits by choosing to emphasize privacy. Even so, people have doubts. Can the AI companies really be trusted to police themselves, given how much money is involved, how quickly the field is advancing, and how intense the competition is?”
But for now, this much is still true: man cannot live by data alone. As flawed as humans are, we still somehow need each other. We hurt each other, compete, and fight like fools… yet we still keep going back for more. If AI can figure that one out, it will be its most extraordinary achievement.