It’s 1:58 pm inside the Pittsburgh Trauma Medical Center’s ER on the Fourth of July and Dr. Robby is exhausted.
The veteran physician and protagonist of HBO Max’s The Pitt, played by Noah Wyle, looks up at the board of nearly 60 patients being treated, running the gamut from a newborn baby abandoned in the hospital’s bathroom to a young law student experiencing a psychiatric breakdown, and a woman being examined after a rape. And then, in waltzes the hospital’s CEO.
Dr. Robby sighs, “This can’t be good.”
The CEO announces that two nearby hospitals have been hit by a cyber attack. The networks are being held hostage until a ransom is paid, something that happens more often in the real world than you might think. To prevent the attack from hitting the hospital, its IT department decides to preemptively shut down all of the computer systems before hackers get in. Anything connected to the internet will go dark.
The wrinkles around Dr. Robby’s eyes momentarily run ocean deep as the patient board disappears and the entire ED goes analog, reduced to dry erase boards, stacks of clipboards, and–get this–fax machines (“They still make those?” a Gen Z student doctor asks). Oh, and say goodbye to that fancy AI app used by Dr. Baran Al-Hashimi (Sepideh Moafi), the new tech-savvy attending physician soon to replace Dr. Robby after he goes on sabbatical at the end of yet another interminable 12-hour shift.
The IT shutdown, which takes place in episode 7, brings a question that’s been simmering throughout season 2 to a head: What does it mean to practice medicine in the age of AI?
The show argues that AI cannot replace the human mind in the field of medicine. The central tension is embodied in the drama between Dr. Robby and Dr. Al-Hashimi. Dr. Robby, the world-weary medical MacGyver, knows the rules and when to break them if it means saving someone’s life. Dr. Al-Hashimi is introduced this season as a “clinical informatics expert” who demonstrates for the student doctors how she uses a “generative AI app” to record doctor-patient interactions and then, in seconds, populates the patient chart. In episode 6, the app hallucinates an appendectomy, confuses medications, and mistakes “neurology” with “urology,” royally pissing off a surgeon operating off clinical slop.
When staff present a case earlier in the season, Dr. Robby asks, “What’s your gut telling you?”
Dr. Al-Hashimi interrupts, “Her gut?”
“It’s this thing AI will never have,” Dr. Robby responds.
The Pitt is lauded for being one of the most medically accurate TV dramas ever made. It’s also prescient on debates about medicine and AI flaring up right now. A recent study just found 60 percent of its sample used AI to ask health-related questions. This week, The New York Times Opinion section published a piece, “Take It From a Doctor: It’s OK if Your Medical Advice Comes From A.I.” Recently, Elon Musk proclaimed, in dubious fashion, that going to medical school is “pointless” because in a few years AI will be so much better at medicine than humans.
In the swirl of clinical vignettes, case after case shows how treating patients is still a deeply human endeavor that requires clinicians to rely on their own instincts, memories, and experience of the world to make on the fly decisions. In The Pitt, medicine is a blend of art and science.
Take one case led by fan favorite Dr. Melissa “Mel” King (Taylor Dearden), who clocks a young Black woman’s persistent cough as being caused by bulimia. Dr. King knows to inspect the back of the patient’s incisors to see if her tooth enamel is being worn down from all the stomach acid she vomits. Having thought she had an eating disorder in high school, Dr. King had obsessively done her own research on the subject.
“If it’s not in the differential, it won’t be in the diagnosis,” Dr. King says to a doctor she’s helping train. What she’s saying is, if you don’t ask the right questions, you won’t arrive at the right answer. Even if fed millions of case studies about eating disorders, an AI might have given the patient antibiotics for pneumonia and sent her home. But Dr. King is not processing information the way a computer does. She’s using her own experience to recognize the root cause of the illness.
While there are different types of AI being used across medicine today, personal experience is something large language models–no matter how many textbooks they’re fed–cannot mimic. LLMs lack instinct and have no experience of the world to draw on. And, crucially, LLMs have never suffered. Pain, anxiety, obsession–these are lovely traits and emotions relegated to the human experience. They can cause us harm, but they can also help us, like in Dr. King’s case, better understand each other.
While The Pitt’s larger point about AI is unsubtle this season, the show is also not blindly techphobic. We see novel uses of technology that help clinicians do their job, like when a young boy puts on a VR headset to play a game that distracts him while a doctor fishes out a small marble lodged in his nose.
The history of medicine is the history of technology reshaping what doctors can see, measure, and repair. The discovery and mass production of penicillin has saved countless lives from once-fatal infections. Medical imaging—from X-rays to CT scans and MRIs—allow physicians to look inside the body without cutting it open. Pacemakers, ventilators, and vaccines have stabilized failing hearts, breathed life into human bodies, and prevented deadly infectious diseases before they ever take hold.
Technology has always moved medicine forward and AI is already pushing boundaries. I asked an ER doctor if he uses AI at work and he said that he and his colleagues frequently use LLMs like OpenEvidence, an AI-trained on trusted peer-reviewed journals, to help aid in diagnosing and treating patients.
Using tech to streamline your work is one thing. Being unable to work without tech is something else. A cyber attack on hospitals is a reminder of how fragile these systems are. Think back to October, when Amazon Web Services crashed for 15 hours, leading to widespread outages across apps that millions of people need to do their work.
Will AIs help solve tedious workflows and make charting more efficient? Yes, it’s already happening. But if you’re ever at a hospital when the networks go down—and they surely will go down sometime—you will want that hospital staffed by smart, empathic humans who still know how to think without first consulting AI.
When the screen goes dark, we’ll be stuck with ourselves.
