The AI Moment in Radiation Oncology and What Comes Next

A response to ASTRO's Spring 2026 ASTROnews feature on AI in radiation oncology

ASTRO’s Spring 2026 ASTROnews feature on AI in radiation oncology is one of the most grounded assessments we’ve seen from a major professional society. It doesn’t oversell. It doesn’t dismiss. And it asks the right question — not “Is AI coming?” but “Where exactly are we, what’s already in use, and what actually comes next?”

From our perspective, the article feels less like a preview of the future and more like a clear description of the present moment radiation oncology is already navigating. That’s why we wanted to respond. Not to restate ASTRO’s conclusions, but to engage with the challenges it names and extend the conversation a step further.

Documentation: where general solutions fall short

ASTRO is right to call out clinical documentation as one of the most immediate and impactful applications of AI in medicine today. The burden of after‑hours charting is real, and the fact that health systems are investing in ambient documentation tools even without direct reimbursement speaks volumes about the operational pressure clinicians are under.

What’s important is that documentation in radiation oncology is uniquely complex. A consult note isn’t just a record of a clinical conversation. It spans data from multiple systems, activates disease‑specific workflows, supports treatment planning, and must remain usable for multidisciplinary teams and downstream processes.

Generic tools can help, but specialty‑specific workflows demand specialty‑specific infrastructure. Without that, even powerful AI risks delivering results that don’t fully reflect how radiation oncology actually works.

Community practice realities can’t be an afterthought

One of the most important sections of ASTRO’s article addresses the reality of community practice. Most radiation oncology practices don’t have dedicated informatics teams or internal AI expertise, and that matters deeply for how new tools get evaluated, adopted, and trusted.

Slowing innovation isn’t the solution. Designing tools that are accessible, interoperable, and practical across the full range of practice settings is. Safe and equitable adoption depends not just on guidance, but on products that reflect real‑world constraints — limited staffing, limited time, and limited tolerance for long implementation cycles.

Community practices experience the same workforce pressures as academic centers, often more acutely. Any serious conversation about AI in radiation oncology has to start there.

From point solutions to connected workflows

ASTRO’s framing of the next phase of AI is particularly insightful: moving from isolated features toward coordinated, multi‑step workflows — systems that prepare structured work products, surface the right context, and hand decisions back to humans at the right moment.

This is where architecture matters. Siloed tools can each perform well at a single task while still leaving friction between steps untouched. The real gains come when information flows cleanly across the workflow — when documentation informs planning, planning informs analytics, and leadership can actually act on what the system is producing.

AI’s value compounds when it’s designed to work with the workflow, not alongside it.

Responsible adoption is an operational challenge

ASTRO’s emphasis on transparency, evaluation, and ongoing monitoring reflects lessons the field has already learned, particularly from earlier waves of automation like auto‑contouring. AI output should initiate human review, not replace it.

Responsible adoption shows up in design choices: grounding automation in evidence‑based standards, maintaining clear approval points, and ensuring that accountability always rests with clinicians. AI should support decisions, not obscure them.

This is less about technical sophistication and more about discipline in how tools are built and deployed.

What we’re taking away

ASTRO’s framing lands: we are past the “AI is coming” moment, and still early in the work that actually matters.

The next chapter in radiation oncology won’t be defined by who has the most advanced models, but by who designs workflows that keep patients safe, keep humans accountable, and make quality visible rather than theoretical.

We’re glad ASTRO is naming these challenges at a societal level. The conversation that’s starting about trust, implementation, and real‑world impact is exactly the one the field needs to continue.

This post reflects on themes raised in ASTRO’s Spring 2026 AI in Radiation Oncology feature. Read the full ASTRO article here.