1. AI as a tool for clinicians supporting patient decisions and care 

Oncology has tolerated a troubling truth for too long: where a patient receives care often shapes the care they are offered. AI should be welcomed only if it helps close that gap. Its highest purpose is not to dazzle academic centers with better dashboards, but to move hard-won expertise into community settings where most patients are treated. If AI merely reinforces the privilege of institutions already rich in knowledge, it will fail. If it raises the median standard of care, expands access to appropriate testing, and helps clinicians justify advanced treatment options, then it becomes more than a tool—it becomes a corrective to inequity. 

The promise of AI is often described in futuristic language, but its most immediate value may be refreshingly mundane: helping physicians escape the swamp of appeals, prior authorizations, documentation, and repetitive justification. In oncology, administrative burden is not a nuisance; it is a theft of clinical attention. Every hour spent fighting for treatment already supported by evidence is an hour taken from patients. AI should be designed to return that time, not create a new layer of busywork dressed up as innovation.  

AI will not replace the oncologist who can interpret uncertainty, hold competing risks, communicate honestly, and remain accountable to a patient sitting across the room. But it may replace parts of the job that should never have defined the profession in the first place. The physician’s value must move upward—from information retrieval to judgment, from documentation to relationship, from protocol recall to ethical navigation. Anxiety is understandable and real, but the answer is not resistance. It is a reinvention of what the role truly should and could be. 

Patients assume their care teams are aligned, yet too often, they are not. Oncology remains divided across specialties, systems, languages, and cultural expectations. AI could become a bridge across those divides, translating complexity into usable information for physicians, patients, and families. But this will not happen automatically. If AI tools are built only for billing workflows or institutional efficiency, they will deepen fragmentation. If they are built around patients’ lived realities, including language, culture, and health literacy, they could finally make coordinated care feel less aspirational and more real.  

2. AI as a tool for discovery   

AI also has applications before it reaches the clinic through drug discovery, trial design, patient identification, and digital pathology/imaging. In drug discovery, AI can screen molecular compounds, predict drug interactions, and reveal therapeutic patterns and mechanisms of action buried in massive datasets far faster than traditional methods. In trial design, AI tools can match patients to eligibility criteria and forecast enrollment bottlenecks prior to a trial starting. AI has been shown to more accurately detect cancers otherwise missed by human readers. It can rapidly identify patients at high risk for treatment-related adverse events and improve clinician response time, improving patient outcomes.  

These are exciting advances, but their discovery alone does not guarantee access and justice. A drug identified faster or a patient flagged sooner still needs to reach care. The more urgent question is whether AI can help patients find and access clinical trials before opportunity quietly passes them by. The oncology community should resist celebrating AI as a research engine unless it also becomes an access engine.  

The right posture toward AI in oncology is neither blind enthusiasm nor reflexive fear. It demands vigilant engagement and optimism. AI may fail to save time, magnify bad data, produce confident nonsense, or encode inconsistent definitions from the past and call them insight. But none of that means oncology should retreat from AI. Rather, the field must insist on evidence, shared definitions, patient-centered design, transparency, and accountability before allowing algorithms to shape the future of cancer care. 

3. AI Can Strengthen Advocacy, But It Cannot Replace Lived Experience  

AI not only offers a tool for better treatments, clinical trials, and care, but it can also spread knowledge faster than any conference, white paper, or advocacy training session. It can help advocates interpret evidence, identify policy gaps, and argue for better access to care. Yet the idea that AI could replace advocates misunderstands what advocacy is. Advocacy is not merely information transfer. It is moral pressure, personal witness, and persistence born from lived experience. Advocates and their caregivers are persons with lived experience. AI can arm advocates, but it cannot become one.  

This is why it is critical for us as advocates to educate and insert ourselves into the AI development and utilization process.  Recently, a non-profit organization was created to focus on ensuring that patient voices are heard in the progress and governance of AI in cancer.  The goal of the AICANcer Patient Advocacy Group (aicancerpatient.org) is to demand that AI is developed with principles of equality, integrity, empathy, and a patient-centered focus. It cannot happen around us. And it absolutely should not happen without us. 

Final thoughts

AI in oncology should not be measured by how intelligent it appears, but by whether it redistributes expertise, restores clinicians’ time, expands access to patients, and resists becoming another instrument of denial. The dreams of AI’s potential are worthy and achievable if the risks are confronted directly, with patients as central and active participants in how these tools are developed, evaluated, and ultimately used.