Top Gun Medicine: Why Oncologists Deserve an AI Co-Pilot

“Superior pilots use their superior judgment to avoid situations requiring their superior skills.” — Unknown

It’s Thursday afternoon and you’re three patients behind schedule. The waiting room is full, your medical assistant keeps catching your eye with that look that means someone’s getting impatient, and you’re covering for a partner who’s on vacation. Which means half these patients you’ve never seen before.

Next up: a 52-year-old man with metastatic colorectal cancer, here for restaging scans after first-line oxaliplatin-based chemotherapy. You pull up the chart while walking to the exam room. The imaging report loads: progression in the liver, new lung nodules. He’s going to need second-line therapy. You’re thinking an irinotecan-containing regimen, standard approach. His performance status has held up well. Labs look reasonable.

But buried in the 47-page molecular report from six months ago—a report you’ve never seen because this isn’t your patient—is a detail that changes everything: UGT1A1*28 homozygosity. A pharmacogenomic variant that turns standard irinotecan dosing from life-extending therapy into life-threatening toxicity. It’s there in black and white, one line among thousands. Easy to see if you know to look for it. Easy to miss when you’re trying to synthesize imaging, pathology, labs, prior treatments, and a patient you don’t know well into a coherent plan in the next fifteen minutes.

This is modern oncology’s paradox. We’ve never had more sophisticated instruments at our disposal—genomic sequencing, liquid biopsies, pharmacogenomics, PET-CT fusion imaging. We’ve never had more ways to peer into the molecular machinery of cancer and personalize treatment accordingly. And we’ve never been more likely to miss critical details because the instrument panel has grown too complex for any human to monitor perfectly.

We’re flying increasingly sophisticated aircraft through increasingly complex airspace. But we’re still flying solo.

When Flying Solo Stopped Being Heroic

In 1977, two Boeing 747s collided on a foggy runway in Tenerife, killing 583 people. The cause wasn’t mechanical failure or weather—it was human error compounded by communication breakdown and cognitive overload. The tragedy became aviation’s inflection point, the moment the industry acknowledged that even the most skilled pilots needed systems designed around human limitations, not superhuman expectations.

What followed transformed flying forever. Not by replacing pilots, but by surrounding them with intelligence that watched what they couldn’t watch, remembered what they couldn’t remember, and spoke up when something critical needed attention. The result? Commercial aviation became exponentially safer, not because pilots became more skilled, but because they stopped flying alone.

Medicine stands at its own Tenerife moment. We’re asking oncologists to maintain perfect vigilance across an impossible landscape of information. To remember that UGT1A1 variants require irinotecan dose reduction. To track which patients qualify for which of hundreds of open clinical trials. To calculate drug interactions across polypharmacy regimens that change weekly. To synthesize molecular profiles against treatment algorithms that evolve monthly.

This isn’t sustainable. More importantly, it’s not necessary.

When Manuals Aren’t Enough

Every pilot studies their aircraft’s manual obsessively. Over a thousand pages of technical specifications, emergency protocols, and operational guidelines. They know the V-speeds, the fuel consumption curves, the precise sequence for engine failure procedures. This knowledge is essential, foundational, life-saving.

But manuals are static documents. They tell you what to do when the engine temperature reaches a certain threshold, but they don’t watch that gauge for you while you’re navigating through a thunderstorm. They explain the principles of lift and drag, but they don’t calculate wind shear in real-time as conditions change moment by moment.

In oncology, we have our own manuals. The NCCN guidelines—often numbering in the hundreds of pages per tumor type—updated constantly. Landmark papers from the Journal of Clinical Oncology that reshape treatment paradigms. Genomic databases that catalog the molecular signatures of thousands of tumors. These are our V-speeds, our emergency protocols, our operational guidelines.

But like aviation manuals, they’re snapshots of knowledge frozen in time. They tell us that patients with certain genetic variants require modified dosing, but they don’t actively monitor for those variants in real-time. They explain the principles of precision medicine, but they don’t synthesize a patient’s complete molecular profile against the latest clinical trial criteria while we’re sitting in the exam room—and certainly not when we’re four patients behind schedule after a particularly challenging office visit.

The gap between what we know and what we can actively remember and apply in the moment of decision has grown impossibly wide.

Your AI Co-Pilot Is Always Running

This is where artificial intelligence changes everything—not as a replacement for clinical judgment, but as real-time clinical decision support running continuously in the cockpit with you.

The AI co-pilot in oncology isn’t trying to practice medicine. It’s trying to do what aviation systems do: watch the instruments with perfect attention so you can focus on flying the plane. It’s not sitting in a seat beside you. It’s embedded in your workflow, monitoring every data point, cross-referencing every detail, running constantly in the background of your clinical encounter.

Picture the colorectal cancer case from earlier. You’ve reviewed the imaging showing progression, assessed performance status, and you’re ready to order an irinotecan-containing regimen. As you enter standard dosing, a discrete alert appears—not from a separate system requiring another login, but embedded right in your workflow:

“Alert: Patient homozygous for UGT1A1*28 variant. Standard irinotecan dosing associated with Grade 3-4 neutropenia and diarrhea in 50% of cases. FDA labeling recommends starting dose reduction to 200 mg/m². Consider 70% of standard dose per NCCN guidelines.”

You pause. Check the pharmacogenomics result from six months ago—one you’d never seen before today. There it is, buried on page 7. The AI didn’t make a medical decision—you’re still the one holding the yoke. But it prevented a potentially devastating mistake by continuously monitoring details across a patient record you couldn’t possibly have memorized while covering for a colleague.

This is augmented medicine. Enhanced clinical judgment. Partnership, not replacement.

Addressing the Fear in the Room

Let’s talk about what you’re really worried about. Not job security—you know medicine will always need doctors. It’s autonomy. The fear that algorithms will start driving your decisions. That some computer will second-guess your clinical judgment or override your expertise.

I get it. You didn’t spend decades learning to practice medicine only to have a machine tell you what to do. And here’s the crucial point: that’s not what this is about.

The AI co-pilot doesn’t hold a medical license. It doesn’t have clinical intuition. It can’t read between the lines of what a patient isn’t saying, or factor in family dynamics, or make the nuanced ethical decisions that define good doctoring. It can’t sit with someone through bad news or help them understand their prognosis or guide them through the valley of chemotherapy.

What it can do is ensure you never miss a drug interaction because you were thinking about three other patients. It can flag clinical trials you didn’t know existed. It can catch dosing errors before they happen. It handles the cognitive overhead so your brilliant mind can focus on the art of medicine.

You’re still the pilot. You’re still making every meaningful decision. You’re just not flying alone anymore.

The Liberation of Attention

When Chuck Yeager broke the sound barrier, he was flying solo in an experimental aircraft with minimal instrumentation. It was heroic, groundbreaking, and incredibly dangerous. Modern test pilots fly with teams, systems, and safety nets that would astonish Yeager—not because they’re less capable, but because we’ve learned that even extraordinary humans perform better with extraordinary support.

The same evolution is happening in oncology. The era of the lone-wolf physician—brilliant but isolated—is giving way to something better: collaborative intelligence that combines human wisdom with computational precision.

Your AI co-pilot doesn’t diminish your expertise; it amplifies it. It doesn’t reduce your autonomy; it enhances your ability to make informed decisions. It doesn’t replace your judgment; it creates space for deeper judgment by handling the mechanical aspects of medical vigilance.

When pilots trust their instruments are being monitored reliably, something profound happens: they can focus completely on flying. Not just maintaining altitude and heading, but reading the weather, anticipating problems, making strategic decisions about route and approach.

The same transformation is possible in oncology. When AI continuously monitors the data—the relentless checking and cross-referencing and pattern-matching—you become free to practice the medicine you trained for. To really listen instead of simultaneously running mental calculations. To notice the subtle signs that something’s wrong before it shows up in labs. To sit with difficult emotions instead of rushing to the next task. To make nuanced ethical decisions that no algorithm could navigate.

Superior pilots, as the old saying goes, use their superior judgment to avoid situations requiring their superior skills. The wisdom isn’t about having the best reflexes or the quickest hands on the controls. It’s about building systems that prevent emergencies before they happen. About creating environments where routine operations don’t demand heroic intervention.

Commercial aviation understood this decades ago. Every flight now operates with multiple layers of redundancy, automated monitoring, and real-time decision support. The result isn’t diminished pilots—it’s safer skies.

Medicine is finally catching up. The AI running continuously in your cockpit represents the same principle: use intelligence—both human and artificial—to avoid situations where patient safety depends on perfect recall under impossible conditions.

This isn’t about computers taking over medicine. It’s about doctors finally getting the support they deserve to practice the kind of medicine they envisioned when they first put on a white coat. Medicine focused on healing rather than information management. On connection rather than documentation. On wisdom rather than memory.

You’ve been flying solo long enough. You’ve maintained altitude through storms that would have grounded earlier generations. You’ve navigated complexity that keeps expanding beyond any individual’s capacity to manage alone.

But that patient with progressive disease—the one you’re seeing for a colleague, the one whose critical details are buried in reports from six months ago, the one who arrives on the day when you’re running behind and your attention is divided—they deserve more than hope that you’ll catch everything. They deserve a system designed to ensure nothing falls through the cracks.

Not because you’re not brilliant enough or careful enough. But because no human should have to be superhuman.

The AI is running. The instruments are being monitored. The cockpit is no longer a place of solitary vigilance.

Now you can focus on what only you can do: safely landing that plane.