At ASCO, a cross-sectional survey of 75 adults with cancer at a safety-net oncology clinic found patient trust remains overwhelmingly with clinicians: 97.3% said they would trust their oncologist most to answer a serious question about their cancer’s future, while 0% chose ChatGPT or AI. The Dell Medical School at The University of Texas at Austin study suggests patient-facing AI may support oncology care, but not replace physician guidance for prognosis and treatment discussions.

Why It Matters To Oncology

• Trust is the gating factor for patient-facing AI in cancer care, especially in high-stakes conversations around prognosis, treatment decisions and serious illness communication.

• Only 52% of patients agreed that ChatGPT provides accurate responses, underscoring persistent concerns about hallucinations and reliability in oncology use cases.

• Comfort with technology tracked with comfort using AI for cancer questions, with a statistically significant Spearman correlation of 0.40.

• The findings point to a near-term role for AI as an adjunct for education, navigation and question preparation — not a substitute for oncologist-led counseling.

The Financials

• No funding, cost-effectiveness or commercial partnership details were provided in the ASCO presentation summary.

• For drug discovery and oncology developers, the practical takeaway is strategic: tools aimed at patients may face slower adoption than clinician-facing or workflow-focused AI products.

• That could shift investment interest toward decision support, trial matching, biomarker interpretation and operational uses rather than direct-to-patient cancer guidance.

What They're Saying

• The researchers concluded that physician trust remains central in oncology care even as AI tools become more accessible to patients.

• They flagged the 52% accuracy-confidence figure as significant and actionable, given that large language models can deliver confident but incorrect answers.

• Authors also suggested language may shape trust: Spanish-speaking patients were less likely than English-speaking patients to agree they would trust a doctor more than AI, 71.9% vs. 87.2%.

• A similar directional pattern appeared by age, with patients younger than 55 less likely than older patients to say they would trust a doctor more than AI, 73.3% vs. 84.4%.

What's Next

• The authors called for prospective, multicentre and longitudinal studies to test whether these findings generalize across oncology settings.

• Future work will need to track how trust in AI changes across the cancer care continuum, from diagnosis through treatment and survivorship.

• For innovators in oncology and drug discovery, the next opportunity may be building AI that strengthens the clinician-patient relationship rather than trying to displace it.

• Outside oncology, doctors in a separate FirstWord snap-poll saw nearer-term promise for patient-facing AI in primary care and diabetes/metabolic care, where routine follow-up and education are more standardized.