Patients are not waiting for healthcare organizations to decide how AI should fit into their care.
They are already using it.
That was one of the clearest messages from “AI Meets Empathy,” a session at the 2026 MAPA Medical Affairs Summit featuring Monisha Nair, Chief Commercial Officer, APAC at RxPx, and Luan Lawrenson-Woods, award-winning patient advocate recognized by Patients Australia, women’s health advocate, speaker, and host of the podcasts Rewritten Me and Regarding Me.
Rather than starting with what AI can do, the conversation started with the moments patients actually experience: the shock after diagnosis, the uncertainty during treatment, the questions that surface in the middle of the night, and the ongoing mental load of navigating care long after active treatment ends.
Across those moments, one question kept returning:
What would genuinely help the patient here?
Three themes stood out.
- Patients Are Already Using AI
At 3 a.m., when fear hits and a patient’s healthcare team is not available until 8 a.m., patients still have questions.
And increasingly, AI is where they are turning for answers.
During the session, Mon highlighted research suggesting that around half of patients are already using AI for healthcare questions and to prepare for medical appointments.
The question for the industry, then, is no longer whether patients will use AI. It is whether the information and support they receive will be safe, accurate, and appropriate for their individual situation.
Luan brought that reality to life through her own experience with breast cancer.
After being diagnosed with a rare form of breast cancer in 2017, before today’s AI tools were available, she searched online for information that could help her understand what her diagnosis meant. Much of the credible information available was difficult to understand. One accessible source painted an extremely dire prognosis, only revealing at the bottom of the page that the information was outdated.
The disclaimer was there.
But by then, the damage had already been done.
“I couldn’t unsee it,” Luan told the audience.
Today, AI can make complex health information more accessible and conversational. It can help patients prepare questions for healthcare professionals, explain unfamiliar terminology, and provide support when another person may not be available.
But accessibility alone is not enough.
During the session, Luan tested a general-purpose AI tool using her own diagnosis. Much of the response was accurate, but it also described a likely treatment pathway that did not reflect her actual experience. For a frightened patient, generalized information can quickly become an expectation about what should happen next.
That reveals an important limitation: patients may experience an AI response as personal even when the information behind it is not.
The conversation also produced an unexpected example of that tension after the Summit.
Luan used Gemini to turn the session transcript into a 10-minute podcast-style recap of the discussion. Within minutes, AI had transformed a live conversation into an accessible new format, demonstrating how quickly the technology can help make information available in different ways.
But the output was not perfect. The AI made assumptions that did not fully reflect the original conversation.
That is precisely the point.
Generative AI can produce content that sounds polished, confident, and credible while still getting important details wrong. In healthcare, that distinction matters. Whether AI is answering a patient’s question or interpreting a conversation about patient care, a convincing response should never be mistaken for a verified one.
We have shared the AI-generated recap as an example of both the opportunity and the limitation: what AI can create quickly, and why human review, appropriate governance, and transparency still matter.
Listen:
About this recording: This podcast-style recap was generated by Gemini from the AI Meets Empathy session transcript. It is shared as an example of generative AI in practice and has not been presented as a verbatim or fully accurate representation of the discussion. The generated content includes assumptions that do not fully reflect the original conversation.
And that tension extends beyond the information AI provides. It also raises questions about where in the patient journey we choose to use it.
Patient support has traditionally focused on the period after a therapy is prescribed. Yet diagnosis, treatment decision-making, and the period immediately following diagnosis can be among the most vulnerable points in the patient journey.
AI may create new opportunities to support patients earlier, when questions and uncertainty are often at their highest. The challenge is ensuring that industry structures, governance, and support models evolve alongside the technology.
- Governance Is Care
When we talk about empathy in AI, it can be tempting to focus on tone.
Does the response sound compassionate? Does it acknowledge fear? Does it communicate in language the patient understands?
Those things matter. But empathy in healthcare AI needs to go further.
It means understanding the patient’s real context and responding safely within it.
During active treatment, Luan described managing chemotherapy side effects, contradictory information, temperature monitoring, and the exhaustion of trying to remember what mattered and when she needed to act.
At one point, her temperature began rising overnight. She had clear instructions from her oncologist about when to seek emergency care, but in the moment, she hesitated because she did not want to bother anyone.
It is an important reminder that giving patients information does not always translate into action.
Patients become tired. They normalize symptoms. They lose track of what they have experienced. They may hesitate to contact their care team or struggle to determine which information matters.
Used appropriately, AI could help reduce some of that burden.
It could personalize treatment education, support symptom tracking, surface relevant information, help patients prepare for appointments, identify changes in engagement, and connect digital interactions with human support. It can also adapt information to different languages and potentially incorporate information from wearables to create a more complete picture of what is happening outside the clinic.
But the more personalized and influential AI becomes, the more important its guardrails become.
Patients need to understand where their data goes, who can access it, how it is being used, whether they can opt in or out, and how the information they receive is validated.
For organizations, that means AI-enabled patient programs need to be secure, transparent, auditable, and designed with clear controls around what information can reach patients and when human intervention is required.
In a regulated industry, governance is not paperwork around the patient experience. It is part of the care itself.
The goal should not be an AI that simply sounds empathetic.
It should be an experience built around the patient’s actual treatment context, needs, and preferences, with guardrails they can trust.
- AI Should See the Whole Patient, Not Just the Treatment
The need for support does not disappear when active treatment ends.
Luan described life after cancer treatment as a fragmented landscape of healthcare professionals, scans, appointments, apps, chronic conditions, and information she often has to carry between providers herself.
At one point, she described feeling like a “carrier pigeon” between healthcare professionals.
That description captures a problem technology alone cannot solve: patients are often expected to connect a healthcare ecosystem that was never designed around them.
This is where the conversation about AI becomes bigger than any individual chatbot, app, or feature.
The opportunity is to build a connected support ecosystem around the patient.
AI could help organize questions before appointments, recognize patterns across symptoms and behaviors, reduce administrative burden, personalize education, and identify moments when human intervention may be needed.
But doing that effectively requires designing around the whole patient journey rather than an isolated product interaction.
As Mon discussed during the session, organizations need to connect functions, co-create programs with patients, design for safety, and close the loop back to care.
For Medical Affairs, that creates an important leadership opportunity.
Scientific credibility, patient insight, evidence generation, safety, and cross-functional collaboration all intersect in AI-enabled patient engagement. Medical Affairs can help ensure organizations look beyond what technology is technically capable of delivering and ask a more important question:
What should AI do for this patient, at this moment, and what needs to be true for us to trust it?
From “Could” to “Will”
Luan closed the session with a challenge for the room.
Not what could you do?
What will you change on Monday?
It is an important distinction.
Healthcare does not need more abstract conversations about the potential of AI. Patients are already using it to interpret symptoms, understand diagnoses, prepare for appointments, and find answers when their healthcare teams are unavailable.
The opportunity for life sciences is not simply to introduce more AI into that experience.
It is to make the experience safer, more useful, more connected, and more human.
That starts with understanding what patients are doing when no one else is in the room.
It means co-creating with patients from the beginning, rather than validating solutions after they have already been built.
It means treating governance, transparency, accuracy, and connection to human care as fundamental parts of patient-centered design.
Because the most important question for AI in healthcare may no longer be what the technology could do.
It is what we will do to ensure it meets patients with empathy when they need it most.
