The Paradox of Digital Rapport in Preventative Care
The integration of artificial intelligence into primary care is often framed as a solution to administrative bottlenecks and patient disengagement. However, as healthcare providers deploy AI chatbots to drive preventative health measures, a critical tension has emerged between the desire for “human-like” interaction and the institutional requirement for clinical transparency and safety.
Research conducted at the University of Surrey suggests that while anthropomorphism-the attribution of human characteristics to non-human entities-is often intended to reduce anxiety and build rapport, it can frequently produce the opposite effect in a medical context. When AI attempts to mimic human intimacy or empathy, it can trigger skepticism among patients who prioritize clarity and professional boundaries over simulated warmth.
Dr Doris Dippold, lead author of the study and Associate Professor in Intercultural Communication at the University of Surrey, said: “Our analysis shows that anthropomorphism is not universally positive. Human-like features can build rapport – but when they clash with patients’ expectations for transparency in a healthcare setting, they undermine exactly the trust the chatbot is trying to build.”
This friction is particularly acute when the technology attempts to blur the lines between software and clinician. For many users, the chatbot’s statement that they could “chat to me as if I am a real person” backfired, appearing suspicious rather than reassuring. In a system where consent, accountability and clinical governance depend on knowing who-or what-is giving advice, ambiguity about the nature of the “speaker” is not simply a user-experience flaw; it is a structural risk.
Systemic Barriers and Patient Friction
The deployment of digital health tools often overlooks the intersection of technology design and patient vulnerability. In a trial of the chatbot Asa, designed to assist with cervical screening uptake in a diverse and socioeconomically deprived community in Islington, several distinct friction points were identified that go beyond interface preferences and into questions of power and pressure.
For patients navigating complex life circumstances, the perceived “helpfulness” of an AI can be interpreted as intrusive or aggressive. The study found that follow-up messages delivered within 24 hours were often viewed as overbearing, particularly when they arrived at times of day that clashed with work, childcare or rest. Furthermore, imperative language such as “Let’s book you in” was perceived as a demand rather than a supportive prompt, especially when it did not acknowledge that a patient might have reasons to delay or decline.
These design flaws have disproportionate impacts on specific population groups:
- Neurodivergent patients: High-pressure communication, notifications that cluster within short time windows, and rigid response timelines can create sensory or cognitive overwhelm and prompt withdrawal from the service altogether.
- Caregivers: Those with demanding caring responsibilities may find rapid-fire automated reminders unfair or stressful, reinforcing a perception that the system does not understand the constraints of unpaid care.
- Mental health patients: Patients managing psychological challenges may perceive “pushy” AI interactions as a lack of empathy or understanding of their current capacity, compounding existing mistrust in institutions.
What begins as a micro-level design decision-tone, timing, frequency-therefore becomes a macro-level access issue. When digital-first communication is embedded into screening programmes, patients who disengage from the chatbot may in practice be disengaging from the health system itself.
Public Health Implications for Cancer Screening
The failure of digital communication tools is not merely a matter of user experience; it is a matter of clinical outcome and public health planning. Preventative screenings are the cornerstone of early cancer detection, yet uptake rates are currently facing systemic declines at the very moment when many health services are banking on AI to close gaps in care.
The necessity for effective, equitable communication is underscored by recent trends in population health:
| Metric | Impact / Trend |
|---|---|
| Cervical screening uptake (UK) | 5.3% decrease during 2023-24 |
| Demographic gap | Consistent underrepresentation of ethnic minority groups |
| Risk factor | Digital disengagement leading to total loss of patient contact |
In underserved areas, where trust in institutional healthcare may already be fragile due to historic underinvestment or cultural barriers, an untrustworthy or aggressive digital interface can act as a final barrier to access. This is especially critical for cervical screening, where consistent participation is essential for the prevention of cervical cancer and where invitation, reminder and follow-up systems are now largely automated.
The policy stakes are high. National screening programmes are commissioned and performance-managed on the assumption that eligible populations can be reliably reached and nudged. If AI-mediated outreach causes a subset of patients-often those at highest risk-to opt out of digital contact altogether, it undermines both equity targets and headline coverage rates. For health system leaders, chatbot tone, pacing and transparency are therefore not peripheral product details but determinants of whether statutory screening standards can realistically be met.
Frameworks for Ethical AI Integration
To prevent the alienation of vulnerable populations, the transition to AI-driven health communication must move away from mere simulation of human personality toward a framework of functional transparency and respect that is aligned with emerging regulatory expectations. The University of Surrey study indicates that the efficacy of health AI depends on its adherence to a specific set of patient-centric principles, many of which echo the UK government’s cross-sector approach to regulating AI set out in its pro‑innovation AI regulation framework.
The recommended design standards for healthcare chatbots include:
- Goal orientation: Focusing on helping people achieve their health goals without unnecessary friction, while being honest about what the system can and cannot do.
- Patient agency: Ensuring users maintain full control over decisions and timing, including clear options to pause, reschedule or opt out of reminders without penalty.
- Adaptive response: Responding appropriately to the specific needs and capacities of the individual-for example, slowing the pace of messages or changing tone when signs of distress or hesitation appear.
- Dignity: Treating all users with professional respect, avoiding language that trivialises concerns, and recognising that declining a screening appointment can be a rational choice at a particular time.
- Equity: Ensuring fairness in how the technology is deployed and accessed across different demographics, including communities with limited digital literacy or intermittent connectivity.
- Operational transparency: Being explicitly clear about how the technology works, what data it collects, whether responses are automated or clinician-reviewed, and how clinical responsibility is allocated.
The goal of health AI is not to replace the human element of care, but to facilitate the pathway to it-getting the right information to the right person at the right time, and making it easier rather than harder to reach a clinician when needed. When digital tools fail to recognize the nuances of patient trust, they risk becoming an obstacle rather than an entry point, forcing health services to invest yet more resource in repairing relationships offline.
Dr Dippold continued: “Feeling seen, appreciated and emotionally supported is not a luxury feature in health AI – it is a condition of access. If patients disengage because a chatbot feels pushy or untrustworthy, the health service loses them entirely.”
As health systems move toward greater AI regulation and integration, the priority for policymakers, regulators and providers must remain the preservation of the patient-provider relationship, ensuring that innovation does not come at the cost of equity or trust. For national screening leads and hospital boards now piloting or procuring AI tools, that means treating conversational design and transparency not as cosmetic add-ons, but as core governance questions that determine whether preventative care becomes more inclusive-or more exclusionary-in the digital age.
Related reading
