How patients combine AI chatbots with trusted health sources

Patients may not be choosing between AI and trusted health information

Major healthcare organizations have responded to consumer-facing AI chatbots by launching branded AI tools, in part to maintain patient relationships and continuity of care. That response treats control of the AI interaction as strategically important. Luth Research’s findings point to another behavior that is already common: patients can use consumer AI and established health information in the same research process, giving each source a different job.

Consumer chatbots are still in their nascency, so this behavior changes the strategic question for healthcare leaders deciding how to respond. The formulation “AI chatbots don’t replace online research for patients. They add to it.” captures the central finding: AI can become another part of patient education while health-system websites and other established resources continue to play a role. Healthcare organizations can use that distinction to understand where their information fits before, during and after an AI interaction.

The dominant pattern is hybrid research

The strongest evidence for this view comes from Luth Research’s assessment of more than 1,500 healthcare consumers. Among people who use AI to learn about their health, 80% also use other information sources. The report authors call these consumers “hybrid researchers,” a useful term for people who combine AI with established ways of researching health questions.

The hybrid finding concerns information-seeking behavior, so its scope is narrower than the wider debate about AI and clinical care. Some experts characterize patients seeking health advice from AI chatbots as a severe safety risk, while others see a role for chatbots where access to healthcare is difficult. Luth Research’s findings address a different question: where AI sits among the information sources patients use.

That narrower question still matters because it challenges simple models of channel adoption. Patients who use AI can continue using conventional online resources, so healthcare organizations can remain part of their research even when the AI interaction happens elsewhere. The finding therefore makes the complete research journey a more useful unit of analysis for digital strategy than any single visit or channel.

For many surveyed AI users, that journey crosses several kinds of information source as their needs change. A channel visit alone cannot explain the patient’s purpose because the same person may want a quick explanation at one moment and validation at another. Understanding hybrid research therefore means looking at the jobs patients give each source.

Patients give AI and trusted sources different jobs

Those jobs are clearest when patients want quick information or help understanding something complex. Among hybrid researchers, Luth Research found that 47% use AI to get a plain-language explanation of something their doctor told them. The research also found that 44% use it because contacting their doctor is harder. AI therefore fills specific gaps in the information experience: interpreting an existing clinical conversation and providing an accessible answer when reaching a clinician is difficult.

Those gaps help explain why AI can be useful within a multi-source journey. After using AI, hybrid researchers favor online publications from sources such as health systems and leading trade groups as secondary resources; some regard these resources as more reputable. The later visit has a different purpose because patients can arrive with an explanation or initial answer in hand, then use established health information to assess it.

Janeen Hazel, senior executive vice president of client solutions at Luth Research, describes this division of labor explicitly. Luth Research has a commercial interest in organizations using its consumer-research work, so Hazel’s interpretation also reflects that business perspective. “What surprised us most is that AI is not pulling patients away from trusted health sources, it is sending them there better informed,” Hazel said in a press release. “Patients lean on AI for speed, then turn to trusted sources to validate their findings before acting.”

Hazel’s interpretation gives the later visit a specific role: AI accelerates initial understanding, while the established resource supports a judgment about whether the information should influence what the patient does. For healthcare organizations that publish patient information, a pageview after an AI conversation can therefore represent a developed research task. The consumer may arrive with a specific claim to check, keeping that content inside the decision process even though another service handled the initial question.

The order changes with the patient’s task. When hybrid researchers prepare for an upcoming medical appointment, they rely first and foremost on health information websites. In that setting, the established website becomes the initial research destination, showing how the patient’s immediate purpose can determine which source comes first.

The appointment example makes source order part of the behavior being studied. AI can offer speed, accessibility, simpler wording and reinterpretation of information received from a doctor, while health information websites can provide validation or lead the research before an appointment. Movement across sources follows changes in the patient’s task.

Those changing tasks also show the practical pressures behind the choice of source. Difficulty contacting a doctor creates demand for an immediately available answer, while a previous clinical conversation creates demand for interpretation. Later checking creates a separate validation task, so digital leaders need to read channel usage in the context of the patient need that produced it.

Higher-stakes, longer-term research reinforces the multi-source pattern

The same multi-source behavior appears in research about chronic conditions, where patients can spend much more time gathering information. Luth Research found that hybrid researchers spend three times as much time researching chronic conditions as people who use AI alone. Greater research intensity is therefore associated with using multiple resources, showing that substantial engagement with a health issue can coexist with AI use.

Duration of illness adds another association. Luth Research found that people who have been sick for longer are more likely to use resources in addition to AI, and people who have had a chronic disease diagnosis for more than 11 years are 27% more likely to be hybrid researchers. These figures establish an association between extended experience with chronic disease and multi-source research; they do not establish that longer illness causes the behavior.

That association matters for digital strategy because AI usage alone reveals little about the patient’s broader research activity. Someone can be an active AI user while also investing substantially more time across health information sources. For healthcare organizations, chronic-condition research therefore reinforces the value of examining journeys across sources when deciding how their information should serve patients.

The strategic goal is to support the hybrid journey

That journey creates a useful tension with healthcare organizations’ move toward branded chatbots. Building an organization-owned AI tool can help maintain patient relationships and continuity of care. The survey adds another strategic path: healthcare information can remain involved when patients begin or continue their research through an outside AI service.

Hazel frames that path around what happens after an AI interaction: “The takeaway for the health information space is not to compete with AI, but to be the trusted destination AI users come to next,” she said. Her recommendation reflects Luth Research’s commercial perspective, while the underlying behavior gives established health resources a continuing role because validation can follow an interaction that began elsewhere. The pre-appointment findings extend that role because trusted health websites can also come first when a patient is preparing to see a clinician.

Healthcare organizations can use these patterns to design for hybrid information seeking. Organizations with branded AI can consider how it connects with established patient information, while organizations without the resources or appetite to create branded AI can still serve patients through trusted health content. In either case, the substantive design problem is the handoff across a changing research journey: a patient may arrive seeking interpretation, validation or preparation for an upcoming medical appointment.

Key takeaways for decision-makers

• Design for hybrid research: Among surveyed health consumers who use AI, 80% also use other information sources. Healthcare organizations can plan patient education around journeys that move across AI and established health resources.
• Match content to the patient’s task: Patients use AI for quick explanations and access, while trusted health sources support validation and appointment preparation. Digital teams can design content and handoffs around these distinct needs.
• Account for deeper chronic-condition research: Hybrid researchers spend three times as long researching chronic conditions as AI-only users, and longer-term patients are more likely to use multiple sources. Healthcare organizations serving these patients can prioritize credible content that supports sustained research across channels.
• Support the handoffs across sources: Branded chatbots are one way to maintain patient relationships, while trusted health content can serve patients who use external AI services. Digital leaders can connect AI experiences and established resources around interpretation, validation and appointment preparation.