The majority of the respondents were younger in age and rated themselves as ‘tech savvy’, believed that they understood the concept of AI and were aware of AI being used in healthcare. The participants in this study expressed excitement and openness towards AI in healthcare, highlighting the potential benefits during the diagnostic and analytic phases of hearing healthcare delivery. However, they also raised concerns about ethical considerations, including the potential for biases, data misuse, a lack of clarity about legal liability and the need to address the impacts and potential harms associated with AI technologies. Healthcare professionals consider fairness and equity in the development and deployment of AI as important factors in ensuring existing health inequalities are not widened.

• Clinician acceptance impacts technology adoption, yet little is known about the attitudes of UK hearing health professionals towards artificial intelligence (AI) in hearing healthcare

• A web-based survey was conducted with UK ENT specialists, audiologists and general practitioners to assess their attitudes to AI in hearing healthcare

• The findings highlight diagnostic quality, time efficiency and user-centred design as important considerations for AI adoption and implementation

• Understanding the sociotechnical context of the local healthcare system and the emotional element of attitudes to AI technologies is vital to avoid implementation failures

• Future studies should focus on understanding and addressing specific concerns and challenges within each speciality to ensure safe and responsible integration of AI technologies in hearing healthcare

While healthcare professionals providing hearing healthcare noted the potential advantages of using AI technologies, in their response to questions regarding the impact of AI technologies on the patient–clinician relationship they highlighted the need for AI technologies not to replace the human interpersonal aspect of the patient–clinician relationship and for patient preferences to remain central to hearing healthcare delivery. Participants also identified some practical considerations and limitations associated with the integration of AI in hearing healthcare. Adequate training and resources for healthcare professionals were emphasised as essential for the effective use of AI, while the time pressures often faced in primary care settings was recognised as a potential limitation to adoption.

Participants believed that changes in AI technologies would occur gradually as the technologies mature and a balanced perspective is required. There was scepticism about the utility of AI in certain interpersonal situations, underscoring the importance of careful consideration and evaluation when implementing AI technologies in hearing healthcare. These insights emphasise the need to address practical barriers and ensure successful integration of AI into hearing healthcare workflows while considering the unique dynamics of clinical practice.

It was interesting to note that there were no significant differences between the professions in their opinions about AI. This contrasts with attitudes in the 1970s, when general practitioners were at the forefront of the revolution in computerised health service provision while secondary care professionals were slow to adopt it.Reference Benson21 In addition, the age of the participants showed only weak correlations with agreement on the survey statements. This suggests that there may be broad agreement on how AI technology should be developed and deployed.

This is the first study investigating attitudes to AI amongst a range of healthcare professionals providing hearing healthcare. No studies have specifically investigated attitudes amongst ENT specialists or audiologists, highlighting the gap in the literature that this study fills.

The only previous survey investigating attitudes to AI amongst UK general practitioners was published by Blease et al. and included 1474 general practitioner respondents.Reference Blease, Bernstein, Gaab, Kaptchuk, Kossowsky and Mandl22 It revealed that general practitioners expressed considerable optimism about the likelihood that clinical documentation would be fully automated soon, with the majority of participants (79 per cent) believing this would occur within 10 years. In the same study, respondents believed it unlikely that technology will ever be able to fully replace physicians when it comes to diagnosing patients (68 per cent) and delivering empathic care (94 per cent). Artificial intelligence technologies have significantly advanced since the Blease et al. paper was published, yet the present study also found that hearing health professionals did not believe that AI technologies would replace human clinicians entirely. In addition, this study found that hearing health professionals believed the human component of the patient–clinician relationship would not easily be replicated by AI technologies.

Two qualitative studies have been conducted with general practitioners exploring their opinions on AI in healthcare. In a follow up to their questionnaire, Blease et al. conducted a web-based survey of 720 UK general practitioners to gauge their opinions on the likelihood of future technology fully replacing general practitioners in performing 6 key primary care tasks and, if respondents considered replacement for a particular task likely, their estimate of how soon the technological capacity might emerge.Reference Blease, Kaptchuk, Bernstein, Mandl, Halamka and Desroches23 The survey identified three major themes: the limitations of future technology, the potential benefits of future technology, and social and ethical concerns. The limitations included the belief that AI will not be able to replicate the human ability to show empathy. The perceived benefits were improved efficiency and a reduced burden on clinicians. Social and ethical concerns were varied and included the acceptability of AI to patients, the potential to cause harm and, tangentially, AI's potential impact on understaffing.

Buck et al. interviewed 18 general practitioners from Germany in 2020 to better understand general practitioners’ attitudes to AI-enabled systems.Reference Buck, Doctor, Hennrich, Jöhnk and Eymann24 They found three determinants of their attitudes: concerns, expectations and minimum requirements. Concerns regarding AI-enabled systems included existential anxiety, the potential change in the physician–patient relationship, misuse of data and diagnostic bias. Some participants expressed fears of AI taking over their tasks and feeling replaceable. There were concerns about the impact of AI on the physician–patient relationship, including the potential loss of personal attention and standardised interactions.

Misuse of data and violation of privacy were also major concerns, with participants worried about data being intercepted and used to the disadvantage of patients. Diagnostic bias was another concern, as AI systems could influence decision-making and lead to incorrect diagnoses and overexpansion of treatment services.

General practitioners had positive expectations regarding the benefits of AI in terms of diagnostic quality, diagnostic efficiency and legal liability, but they also had concerns about AI's lack of human competencies and the potential increase in time expenditure. Environmental influences, such as changing working conditions, stakeholder opinions, media coverage and information technology infrastructure, also shaped general practitioners' attitudes. Individual characteristics, such as age and affinity with technology, influenced the participants’ perspectives.

The minimum requirements for AI-enabled systems, as identified from the interviews, included time efficiency, diagnostic quality, data security, economic viability, transparency and autonomy. General practitioners expect AI systems to be fast and easy to use, provide accurate diagnoses, ensure data privacy, be affordable, be transparent in their functioning and allow physicians to maintain their autonomy in decision-making. It is important to note that in some respects, clinicians’ autonomy in decision-making has been progressively limited over recent decades by clinical guidelines and targets (e.g. quality and outcomes framework indicators), so AI technologies used for decision-support would not be spoiling a ‘perfect picture’.

This study's findings share similar themes with those published in the literature, including the perceived benefits of AI in hearing healthcare, its limitations, and social and ethical considerations. However, in contrast to the findings of Buck et al.,Reference Buck, Doctor, Hennrich, Jöhnk and Eymann24 we did not find a significant association between age and the attitudes expressed by general practitioners.

One common theme across all studies was the belief that AI technologies could not replicate the human clinician's ability to interact empathetically with their patients. However, recent advancements in large language models, such as ChatGPTTM, have demonstrated the ability of AI to generate quality and empathetic responses to patient questions. Evaluators in one study preferred the chatbot's responses to physicians in 78.6 per cent of the 585 evaluations (95 per cent confidence interval, 75.0–81.8 per cent) and rated the chatbot's responses as significantly more empathetic than physician responses (t = 18.9, p < 0.001).Reference Ayers, Poliak, Dredze, Leas, Zhu and Kelley25 Although further research is required to explore this application of AI in healthcare, it has clear implications for the future of the patient–clinician relationship.

These findings highlight the importance of addressing the emotional component of attitudes to AI technologies, and have identified diagnostic quality and time efficiency as crucial factors for healthcare professionals to consider in adopting AI systems. User-centred design and proactive promotion of AI-enabled systems are essential for the successful integration of AI systems into everyday use in hearing healthcare. Ignoring the surrounding sociotechnical context of the complex circumstances of the healthcare system could lead to implementation failure (e.g. unintended effects, non-adoption or abandonment).Reference Han, Carcillo, Venkataraman, Clark, Watson and Nguyen26,Reference Greenhalgh, Wherton, Papoutsi, Lynch, Hughes and A'Court27