Artificial Intelligence in Healthcare: What Clinicians in Ireland Are Saying

Physicians in Ireland are already using AI in routine care. A new RCPI and EY report reveals rapid uptake, clear benefits, and urgent gaps in training and governance.

In brief

• AI is already used by 58% of physicians in Ireland, mainly through generative tools supporting documentation, clinical reasoning and administrative work in routine practice.
• Clinicians see gains in efficiency and decision‑making, alongside concerns about safety, data security and growing reliance.
• A clear gap exists between use and readiness, with strong demand for training, clearer governance and coordinated implementation as AI increasingly shapes care and patient behaviour.

RCPI and EY Physician Survey on AI in Healthcare. April 2026

In late 2025, EY Ireland, in collaboration with the Royal College of Physicians of Ireland (RCPI) undertook a survey of over 500 physicians working across a broad range of specialties and career stages. It aimed to assess the behaviours, perceptions and attitudes of physicians regarding the use of AI in healthcare. The findings offer new insight into how AI is being used in clinical practice, where clinicians see the greatest opportunities, and the concerns that remain. It highlights the rapid uptake of generative AI, the growing influence of AI on patient behaviour, and the emerging impact on clinical decision making and the doctor-patient relationship.

The finding points to a system in transition. While clinicians recognize clear potential benefits – particularly in reducing administrative burden and improving efficiency – there is also a need for greater clarity, guidance and support to ensure AI is used safely and effectively in practice.

As Ireland progresses its AI for Care Strategy and wider digital health agenda, this report provides timely evidence to inform policy, clinical practice and future implementation.

Executive Summary

Artificial intelligence (AI) is already embedded in clinical practice in Ireland, with 58% of physicians reporting use within the past year. Among those using AI, engagement is frequent, with 60% using these tools at least weekly and more than one in five using them daily.

Adoption is being driven primarily by accessible, general-purpose tools – particularly generative AI – which clinicians are using to support clinical reasoning, documentation and administrative work. This reflects broader international trends, where AI is being incorporated into everyday workflows, often ahead of formal implementation frameworks.

Clinicians recognise clear potential benefits in efficiency, documentation, and decision-making. At the same time, concerns relating to safety, data security and over-dependence on AI highlight that confidence in its use is still developing. This combination of rapid uptake and ongoing uncertainty suggests that AI is already influencing clinical practice, but that its integration remains uneven.

The findings also point to emerging changes in how care is delivered. AI is beginning to influence clinical decision-making and is increasingly present within consultations, where both clinicians and patients are engaging with AI-generated information, with 32% of clinicians reporting that patients presented with AI-generated information in the previous month. This signals a shift towards more AI-enabled models of care, with implications for professional judgement, the doctor-patient relationship and healthcare delivery more broadly.

A consistent theme across the findings is the gap between adoption and preparedness. 93% of clinicians report a need for additional training and there is a clear requirement to support the workforce and future clinicians in developing the skills needed to work effectively in AI-enabled environments.

Taken together, the findings suggest that clinicians are moving beyond early, exploratory use of AI towards more routine integration into clinical practice, albeit in a largely unstructured way. In this context, Ireland is well positioned – through the AI for Care Strategy, the Digital for Care agenda and wider health system reform – to support a more coordinated approach to AI use that is safe, effective and which supports high-quality clinical practice.

To build on this position, the focus now should be on translating early adoption into structured implementation. This includes developing practical guidance for clinicians, strengthening overall clinical governance and oversight, supporting both current and future workforce capability, and ensuring that the impact of AI on patient behaviour and healthcare utilisation is understood and appropriately managed.

Respondent Profile

Consultants accounted for 49% of respondents, with specialist registrars representing 20%, alongside other training and non-training grades (16%), with the remainder in non-clinical roles or having retired from clinical practice. The sample reflects a diverse clinical workforce, with representation across faculties and institutes.

Respondents were distributed across age groups, with the largest proportion aged 35-49 years (39%), followed by 20-34 years (27%); 50-64 years (26%) and 65+ (8%). 53% of respondents were female, 45% male and 2% preferred not to say.

Key Findings

1. AI is already embedded in clinical practice

Over half of respondents (58%) reported using AI in clinical practice within the past 12 months.

The use of AI in clinical practice varied by age group, ranging from 49% and 63% of respondents aged 20-34 years and 35-49 years, respectively, to 64% in those aged 50-64 years, before declining to 38% in those aged 65 years or older. Similarly, use of AI varied by career stage, with 60% of consultants reporting use, compared with 56% of Specialist Registrars and 54% of doctors in other training and non-training grades.  AI use in clinical practice did also vary by gender with 53% of females and 63% of males reporting use in clinical practice.

This indicates that AI is not an emerging technology, but is already an established part of routine clinical workflows for many physicians.

2. Generative AI is driving adoption

Generative AI tools dominate current use:

• Of those using AI in their clinical practice, 86% reported using generative AI systems

While this pattern suggests that adoption is being driven by accessible, flexible tools, it may also raise questions about consistency, oversight and appropriate use.

3. Other AI applications are less widely embedded

While generative AI is driving much of current adoption, other AI use cases are integrated less consistently across clinical practice.

This suggests that, while a range of AI tools are in use, they are not yet as widely embedded as generative tools. In contrast to generative AI, which is often accessible and flexible, these applications may be more dependent on specific systems, infrastructure or organisational implementation.  The relatively low reporting of some tools, such as ambient scribes, may also reflect differences in how these systems are understood or classified by respondents within the survey.

4. Physicians are broadly optimistic about AI, but concerns remain

The combination of optimism and caution suggest that clinician trust in AI is still being established. Addressing issues such as data security, safety and over-reliance will be critical to building confidence and supporting appropriate adoption.

5. Views on equity are mixed

6. Professional impacts remain uncertain

Clinicians expressed mixed views on AI will affect their roles, particularly in relation to autonomy and decision-making.

These results suggest that, while views on autonomy remain uncertain, clinicians anticipate that AI may exert a growing influence on clinical decision-making that they find difficult to override.

7. AI is already shaping patient behaviour

These results indicate that AI is already a factor in shaping patterns of healthcare utilisation and patient decision-making.

8. Replacement by AI

While most clinicians do not see AI as a replacement for them, a notable minority view this as a realistic future possibility:

9. There is a clear and urgent demand for training

This highlights a significant gap between adoption and preparedness. It also highlights the need for clinicians to develop the skills required to interpret, communicate and contextualise AI-generated information in clinical settings.

Discussion

The findings suggest that AI is already embedded in clinical practice in Ireland, but its integration remains uneven and is evolving in real time. This reflects a broader international pattern, where the rapid emergence of accessible AI tools – particularly generative systems – has enabled widespread informal adoption by clinicians, often ahead of formal regulatory, governance and implementation frameworks.

Across multiple healthcare systems, including in Europe, the UK and North America, similar findings are now being reported. Clinicians are incorporating AI tools into routine workflows for documentation, decision support and information synthesis, while policymakers and regulators are working to establish clearer guidance on safe and appropriate use. In Ireland, this aligns with the current policy direction as set out in the AI for Care Strategy and the wider Digital for Care agenda, both of which signal a shift towards more coordinated and structured approaches to digital transformation and AI-enabled care.

Adoption appears to be driven primarily by accessibility and immediate practical utility, rather than by structured system level deployment. This aligns with international evidence suggesting that generative AI, in particular, is being adopted as a ‘general purpose’ tool within clinical environments, often accessed informally by clinicians, rather than as a formally implemented medical technology. As other jurisdictions move towards more coordinated approaches to governance, training and clinical integration, there is a clear opportunity to shape a similarly structured approach within the Irish healthcare system in a way that supports consistency, safety and effectiveness.

There is a clear balance between opportunity and risk. Clinicians recognise the potential for meaningful gains in efficiency, documentation, and aspects of clinical decision-making, which suggests that AI is already delivering value in practice. At the same time, clinicians expressed concerns about safety, dependence and data security, mirroring increasing international focus on issues of oversight, accountability and safe deployment. These shared concerns reinforce the importance of aligning national implementation efforts with emerging international standards and regulatory frameworks, including the EU AI Act.

The results indicate the emergence of what might be described as augmented clinical judgement. As AI tools become more embedded in practice, clinical decision-making may increasingly involve interaction between clinician expertise and AI-generated insights. This reflects a wider international shift towards understanding AI, not as a replacement for clinical expertise, but as a technology which reshapes how judgement is formed and applied. This has implications for professional standards, clinical responsibility, and the governance of AI-supported care.

There is also evidence of evolving clinical dynamics. The reported difficulty in disregarding AI recommendations, alongside growing patient use of AI, suggests that both clinical decision-making and the patient–clinician relationship are beginning to shift. Similar patterns are being observed internationally, particularly as patients increasingly access AI tools directly for health information and advice, while clinicians are incorporating AI into decision-making and documentation. This convergence represents a broader transformation in how healthcare interactions are mediated by digital technologies, with implications for how information is interpreted, how shared understanding is achieved, and how clinical authority is exercised. Appropriate guidance and support will be needed to ensure that this evolution strengthens, rather than undermines, clinical judgement and patient care.

Overall, the findings reflect a system in transition. Internationally, healthcare systems are beginning to move from early, informal adoption towards more deliberate and coordinated approaches to AI integration. Ireland is well positioned through the Digital for Care agenda and the AI for Care Strategy which together provide a clear foundation on which to build a safe, structured and effective approach to the use of AI in clinical practice.

Implications

1. Implementation and Guidance

AI is already being used in routine clinical practice by clinicians in Ireland, but its integration remains largely unstructured. As set out in Ireland’s AI for Care Strategy, there is an opportunity to move from informal adoption towards more consistent and coordinated use through practical guidance that supports safe, effective and context appropriate application, particularly as generative AI becomes more embedded in day-to-day care. This should also consider how AI can be incorporated into clinical pathways and organisational workflows, beyond individual or ad hoc use.

2. Safety, Trust, and Governance

While clinicians recognise the benefits of AI, concerns around safety, data security and over-reliance indicate that confidence in its use is still developing. These concerns mirror those emerging internationally and reinforce the importance of establishing clear governance frameworks, aligned with evolving regulatory approaches, to support safe deployment and maintain trust.

3. Clinical Decision-Making and Autonomy

The findings suggest that AI is beginning to influence clinical decision-making in practice. The reported difficulty in disregarding AI recommendations highlights the importance of maintaining professional judgement and ensuring clear accountability in AI-supported care. With appropriate guidance and safeguards, there is an opportunity to support clinicians in using AI as a complement to, rather than as a replacement for, clinical expertise. In particular, there is a need to translate emerging governance principles into practical frameworks that support clinicians in navigating AI-supported decision-making in real-world settings. This also raises broader questions about how clinical governance structures may need to evolve as AI becomes more embedded in routine care.

4. Patient Behaviour, Healthcare Utilisation and the Doctor-Patient Relationship

AI is starting to shape patient behaviour, with implications for how and when patients decide to seek care.

Given that demand, access and waiting times are already significant pressures in Ireland, some patients may turn to these tools as a substitute for clinical consultation. On the one hand, many of these tools hold the potential to support patients in accessing information, preparing for consultations and engaging more actively in their care. However, there is also a need to consider the potential risks associated with patients using AI tools in place of, rather than alongside, medical advice. These include the possibility of delayed presentation, inappropriate self management, misinterpretation of AI generated information, or reliance on information that may be inaccurate or incomplete. Addressing these risks will require a combination of clear public guidance on the appropriate use of AI and ongoing evaluation of how these tools are influencing patient behaviour and patterns of healthcare utilisation.

As consultations increasingly involve both clinician and patient engagement with AI-generated information, there is also a need to support clinicians in responding to the changing nature of these interactions. This includes helping clinicians manage a shifting doctor-patient relationship, including how AI generated information is interpreted and integrated into shared decision-making. It also includes how clinicians respond when patients present with information that may be more recent, differently framed or of variable quality, as well as managing expectations, maintaining trust and supporting shared understanding.

5. Training and Capability

The strong demand for training highlights a clear gap between adoption and preparedness. As Ireland progresses its Digital for Care agenda, there is a significant opportunity to build capability across the clinical workforce. This includes not only technical understanding, but also the skills required to interpret, contextualise and communicate AI outputs within clinical decision-making.  

There is also a need to consider how these capabilities are developed within medical education and training pathways. Ensuring that undergraduate and postgraduate training programmes equip future clinicians with the skills to work effectively in an AI-enabled environment will be increasingly important.

6. Strategic Direction

Taken together, the findings suggest that the priority for policymakers and healthcare organisations in Ireland is not whether AI will be used, but how its use can be supported in a way that is safe, effective, and aligned with clinical standards. In line with international trends, this involves moving from early, informal adoption towards a more structured and deliberate approach to integration. The current policy landscape - including the AI for Care strategy and wider digital transformation within the health system - provides a strong foundation to guide this transition in a coordinated and sustainable way.

Acknowledgements

We extend our sincere thanks to those who responded to this survey.

This report reflects the input and expertise of a wide range of contributors across the RCPI and EY. We would particularly like to thank the following for their input and support:

• Janet O’Farrell, Louise O’Gogain, Professor Trevor Duffy, Professor Catherine Quinlan, Niamh O' Sullivan, Rebecca McEntee, Orla Woods, Audrey Houlihan and Dr Diarmuid O’Shea – RCPI
• Dr Ana de Chevillier, Rebecca Williams, Liyans Matthews and Cliodhna Daly – EY
• Dr Kasia Domanska, Dr Odharna Ni Dhomhnallain and Dr Brian McCarthy – EY RCPI Management Fellows