Healthcare May Have Misdiagnosed the Clinician Technology Problem
- 6 days ago
- 10 min read
Doctors and nurses are not necessarily resisting AI and digital tools. The evidence suggests they are becoming increasingly selective about the technology they allow into clinical work.
Healthcare has spent years trying to solve the problem of technology adoption.
Hospitals have digitised records, introduced clinical information systems, expanded virtual care, experimented with artificial intelligence and added a growing collection of digital tools around the clinical workflow. Yet adoption frequently remains difficult.
The conventional explanation is often cultural: healthcare professionals are conservative, clinical environments resist change, and doctors and nurses are reluctant to trust technology.
The evidence increasingly suggests that explanation is incomplete.
Healthcare professionals appear considerably more receptive to technology than the resistance narrative implies. What they are far less receptive to is technology that adds work without creating proportional clinical value, interrupts established workflows, produces information they cannot confidently verify, or attempts to substitute professional judgement without sufficient transparency. That distinction matters. Because if clinician resistance is primarily a cultural problem, the solution is better change management.
If it is primarily a technology-design and operating-model problem, the solution is very different.
Clinicians appear more receptive to AI than commonly assumed
One of the clearest signals comes from the NHS.
01
The Health Foundation surveyed 1,292 NHS staff alongside 7,201 members of the UK public. Among NHS staff, 76% supported the use of AI for patient care and 81% supported its use for administrative purposes. Fifty-seven percent said they looked forward to using AI as part of their work, compared with 17% who disagreed.
02
The Stanford Center for Digital Health found a similar pattern in the United States.
Its 2024–25 study surveyed 1,115 healthcare workers at Stanford Health Care. AI use was still relatively limited: 32% had not used AI at work, while more than half reported using it less than monthly or rarely.
Yet 68% still considered AI applications useful, and 60% expected AI to increase how much they could accomplish during a typical working day over the following five years.
That gap is revealing.
Low usage did not necessarily mean low perceived value.
It suggests there is a meaningful difference between believing in the potential of a technology and having that technology successfully embedded into everyday clinical practice. This is where healthcare technology strategy becomes more interesting.
The challenge may increasingly be less about convincing clinicians that technology can help and more about converting technological capability into something clinicians actually want to use.
Healthcare is beginning to redesign itself around intelligence rather than institutions.
What clinicians want technology to do is equally revealing
The Stanford research provides another important clue.
03
Healthcare workers were most excited about AI's potential for scheduling (51%), writing electronic health record notes (47%), reducing healthcare costs (46%) and improving access to healthcare (43%).
Notice what sits near the top.
Scheduling
Documentation
Efficiency
Access
These are not primarily visions of autonomous medicine. They are areas where technology can remove friction around the delivery of care. The Health Foundation found a similar distinction. NHS staff were more supportive of administrative AI than patient-facing applications, while concerns increased around inaccurate decisions and the potential loss of human qualities such as empathy and kindness. This points toward an important emerging principle:
Clinicians appear highly receptive to technology that removes work. They are more cautious about technology that removes judgement.
Those two forms of automation should not be treated as equivalent.
Automating appointment scheduling is fundamentally different from automating diagnosis.
Generating a first draft of a clinical note is different from independently recommending treatment. Retrieving relevant information is different from deciding what that information means for an individual patient.
Healthcare organisations and HealthTech companies therefore need a more sophisticated model of clinical technology adoption than simply measuring whether clinicians are “positive about AI.” The relevant question is which responsibilities clinicians are willing to delegate and under what conditions.
Digital technology can improve clinical performance
The argument for technology is not based solely on perception.
There is evidence that appropriately deployed digital tools can improve healthcare-worker performance.
04
WHO/Europe highlighted an umbrella review covering 123 studies and approximately 250,000 healthcare providers globally. It found positive effects associated with mobile technologies, telemedicine and digital decision-support tools, including improvements in health-worker performance, skills and competencies. WHO also highlighted evidence of improvements in decision accuracy, task completion, productivity, communication and access to real-time information.
That creates an important distinction. The strategic debate should not be framed as:
Technology versus clinicians.
The more useful question is:
Which combination of technology and human expertise produces better clinical work?
That moves the conversation away from replacement and toward augmentation and it changes how technology should be evaluated.
The problem begins when technology enters the workflow
A digital tool can demonstrate impressive performance in isolation and still create limited value inside a hospital. Clinical work does not happen in isolation. A doctor may move between the EHR, imaging systems, laboratory results, communication platforms, scheduling systems, decision-support tools and administrative systems during a single episode of care. A nurse may simultaneously manage documentation, medication, monitoring, communication, escalation, coordination and direct patient interaction.
Every additional interface competes for attention. Every new login creates friction. Every alert demands cognitive processing. Every piece of information that cannot move between systems creates another manual step.
This is why workflow integration appears repeatedly in the research literature.
05
A 2026 systematic review in Oxford Open Digital Health examining healthcare professionals' perspectives on clinical AI identified recurring facilitators including education and hands-on training, explainability, clinician involvement and effective integration into clinical workflows. Barriers included skills gaps, legal and ethical uncertainty, infrastructure limitations and poor workflow integration.
The broader systematic-review literature on digital-health adoption reaches a similar conclusion: technical capability alone does not determine whether healthcare professionals adopt a technology. Organisational conditions, usability, workflow compatibility, training and implementation all matter.
This leads to one of the central conclusions of this analysis:
A technically successful healthcare technology can still be a clinically unsuccessful product.
Healthcare's previous digital transformation matters
AI is not entering healthcare with a blank reputation.
Doctors and nurses have already lived through decades of healthcare digitisation.
Electronic health records made clinical information more accessible and created important infrastructure for modern healthcare. But the experience also introduced documentation burden, information overload, fragmented systems and new forms of administrative work.
06
EY's Global Voices in Health Care research, based on more than 100 interviews with clinicians and health-system executives across 11 countries, illustrates the tension.
Clinicians described practical benefits from technologies such as voice dictation and remote access to medical information. But they also described frustration with siloed applications, repeated logins and difficulty finding the information they needed inside existing digital environments.This history matters because every new AI tool enters an environment already shaped by those experiences.
Clinicians are not evaluating AI solely against what AI promises. They are also evaluating it against what previous healthcare technology delivered.
That produces a strategic challenge for the next generation of HealthTech companies:
Do not simply add another digital layer to an already fragmented clinical environment.
More data is not necessarily more intelligence
Healthcare's digital transformation has also created an information paradox. Healthcare organisations now generate extraordinary volumes of data. But information abundance does not guarantee decision usefulness. Clinical professionals need the right information, in the right context, at the moment a decision is being made. AI could potentially close part of this gap by retrieving, summarising, prioritising and interpreting information.
But it could also make the problem worse. A system that generates another alert, recommendation or dashboard without reducing existing information burden may increase cognitive load rather than reduce it.
This means healthcare should distinguish between two very different outcomes:
More information
and
better decisions.
The objective of clinical technology should increasingly be the second.
The hidden metric: clinical time returned
Healthcare technology business cases frequently focus on cost reduction, throughput, accuracy or revenue.
There may be another metric worth elevating:
clinical time returned.
The Stanford study found 60% of healthcare workers expected AI to increase what they could accomplish during their working day. WHO's review similarly identified faster task completion and productivity benefits from digital tools.
But simply increasing output may be too narrow a definition of success. In a healthcare system experiencing workforce shortages and clinician burnout, the strategic value of technology may lie in reallocating scarce human attention. If AI removes documentation, what happens to those minutes? If automation eliminates searching, what does the nurse do with the recovered time? If decision support accelerates information retrieval, does the physician see another patient or spend longer discussing the decision with the existing one?
We define this potential value as the Clinical Time Dividend:
The value created when technology removes low-value work and reallocates scarce clinical attention toward activities where human expertise matters more.
This moves the discussion beyond labour substitution.
The objective is not necessarily fewer clinicians.
It is clinicians spending less time doing work that does not require a clinician.
The Clinical Technology Paradox
Taken together, the evidence points toward what we define as the Clinical Technology Paradox.
Healthcare professionals increasingly recognise the potential of digital technology and AI.
But willingness to use technology depends heavily on what happens after that technology enters the workflow.
Does it reduce work?
Does it improve the decision?
Can the output be trusted?
Does it integrate with existing systems?
Does it preserve professional judgement?
Does it create more time for patients?
Or does it create another login, another dashboard, another alert, another verification task and another layer of complexity?
This distinction explains why deployment statistics alone tell us relatively little about digital transformation. A hospital can purchase a technology. IT can integrate it. Management can mandate it. Clinicians can receive training and the technology can still fail to create meaningful value.
Because:
Deployment is not adoption.
And adoption is not value.
The complete journey is closer to:
Capability → Deployment → Workflow Fit → Trust → Routine Use → Clinical Value
Each transition creates another opportunity for technology to fail.
What healthcare leaders should do differently
For health systems, the evidence suggests technology evaluation should begin with clinical work rather than technological capability. Before asking “What can this AI do?”, leaders should understand where clinicians lose time, where information becomes fragmented, where decisions are delayed and where administrative work consumes expensive clinical capacity. Technology can then be applied to those problems. Clinicians should also be involved earlier. The systematic-review evidence increasingly supports training, end-user participation and workflow integration as important facilitators of adoption and technology ROI should extend beyond financial measures.
Executives should measure:
clinical time returned;
workflow steps removed;
system switching reduced;
documentation burden;
verification burden;
clinician adoption after the pilot;
patient-facing time;
clinical outcomes where applicable.
These metrics bring technology evaluation closer to the reality of clinical work.
What HealthTech companies should learn
The implications for technology companies may be even more important. Healthcare does not necessarily need another application. It needs fewer reasons for clinicians to leave the workflow. That changes product strategy. Integration becomes a feature. Trust becomes part of UX. Evidence becomes part of the product. Time saved becomes a commercial metric and understanding the clinical workflow becomes as important as improving the underlying model.
The strongest value proposition may therefore not be:
“Look what our AI can do.”
It may be:
“Look what your clinicians no longer have to do.”
The Bolgarz View
Healthcare may have spent too much time asking whether clinicians are ready for technology.
The evidence suggests another question deserves equal attention:
Is healthcare technology ready for clinicians?
Doctors, nurses and frontline teams operate in environments where attention is scarce, decisions carry consequences and additional friction has a real cost. They appear increasingly willing to embrace technology that makes that environment better.
What they are less willing to accept is technology that requires them to work around the technology itself.
That distinction may define the next phase of digital health. The winners will not necessarily be the companies deploying the most AI or the hospitals accumulating the largest number of digital tools. They may be the organisations that understand something much simpler:
The future of clinical technology will be determined less by how much the technology can do and more by how much unnecessary work it allows clinicians to stop doing.
And perhaps that should have been the objective of healthcare technology all along.
Sources
This Bolgarz Intelligence Brief synthesises findings from peer-reviewed research and institutional studies published by Stanford Center for Digital Health, WHO/Europe, The Health Foundation, Oxford Academic, npj Digital Medicine, npj Health Systems, EY and additional healthcare workforce research. Findings from different studies represent different populations, geographies and methodologies and should not be interpreted as a single dataset.
Academic & peer-reviewed research
BMC Health Services Research / PubMed Central — Healthcare professionals’ perspectives on artificial intelligence in patient care: a systematic review of hindering and facilitating factors to AI adoption
Oxford Open Digital Health, Oxford Academic (2026) — Healthcare professionals’ perspectives on artificial intelligence in clinical practice: a systematic review of facilitators and challenges
A particularly important source for our analysis of workflow integration, training, explainability, legal/ethical uncertainty and clinician participation in AI implementation.
npj Health Systems (2025) — Spotlighting healthcare frontline workers’ perceptions on artificial intelligence across the globe
Mixed-methods research involving frontline healthcare workers interacting with AI across eight countries. Among evaluated responses, 75.4% were classified as enthusiastic, 21.6% as practical and 3.0% as sceptical, while qualitative findings exposed important concerns around accuracy, language, cultural context, privacy and human validation.
npj Digital Medicine (2023) — Barriers and facilitators to utilizing digital health technologies by healthcare professionals: a systematic review
Important broader evidence for understanding why clinical technology adoption depends on factors beyond technical capability, including usability, workflow, organisational conditions and implementation.
PubMed / National Library of Medicine — Research indexed under PMID 40613096
PubMed Central — Additional research on healthcare professionals and digital-health adoption
Patient Education and Counseling / ScienceDirect (2025) — Research examining healthcare-professional perspectives relevant to digital/AI-supported care
Health-system & institutional research
The Health Foundation — AI in health care: what do the public and NHS staff think?
One of the most useful sources for the brief. Its survey included 1,292 NHS staff and provides evidence on support for administrative and patient-care AI, expectations, perceived advantages, concerns and the importance of preserving human care.
Stanford Center for Digital Health — Artificial Intelligence in the Healthcare Workforce
Survey research involving 1,115 Stanford Health Care workers, covering current AI use, perceived usefulness, expected productivity effects, trust and preferred applications. This is the source behind several of the headline statistics used in our article.
World Health Organization Regional Office for Europe (2023) — Digital tools positively impact health workers’ performance, new WHO study shows
WHO reports on an umbrella review covering 123 studies and approximately 250,000 healthcare providers, examining the effects of digital technologies on health-worker performance and competencies.
Industry & workforce intelligence
EY — Global Voices in Health Care
Research based on more than 100 in-depth interviews with clinicians and health-system executives across 11 countries, examining workforce pressure, autonomy, digital transformation, technology experience and the future clinical operating environment. This is particularly important for the wider Bolgarz research programme.
Medical Economics — What health care workers really think about tech tools
Additional industry perspective on healthcare-worker experiences and attitudes toward technology.
eSkill — Can healthcare professionals keep pace with the digital age?
Supporting workforce perspective on digital skills and technology readiness.
Comments