OH practitioners are beginning to use artificial intelligence (AI) within their practice. But, as a survey for iOH has shown, concerns remain about how it will affect provision and practice and where this potentially transformational technology will lead the profession.

In 2024, artificial intelligence (AI) transitioned from a niche tech concept to a mainstream tool embraced by businesses and the public, including occupational health (OH) respondents.

Technology and health

Half admit to self-diagnosing health online – poll

Robotics can transform health and safety as well as efficiency – report

Artificial intelligence set to revolutionise future occupational health delivery

The introduction of AI tools in a large organisation sparked a lively discussion in a UK OH social media forum, leading to a voluntary special interest group (SIG) – the ‘AI in OH’ group – being established. This group, comprising OH respondents and academics, delved into the potential and challenges of AI in OH, as this article intends to show.

Early discussions were a blend of excitement and concern. Enthusiasm centred on AI’s potential to enhance OH access, reduce administrative burdens, allow respondents more time to spend on direct client care and focus on strategic, impactful work.

However, this optimism was tempered by worries about job displacement, ethical considerations, and data security. The lack of specific guidance on AI’s safe and effective use in OH became evident, highlighting the need for industry direction.

In a pioneering effort, the SIG issued a consensus statement defining its mission to ‘Collaboratively enhance the health of the working population through the effective, ethical, and safe integration of AI in occupational health strategies and practices’. This is supported by five key workstreams:

• Advocate for worker’s wellbeing
• AI risk management
• Enhance clinical value with AI
• AI education for OH clinicians, and
• Leverage AI opportunities – human touch, efficiency, information gaps, and data insights.

To kickstart their mission, the SIG surveyed UK OH respondents to gauge current and planned use of AI.

The survey, which ran for 35 days and garnered 86 responses, aimed to assess AI tools and software in use, their purposes, governance processes, funding arrangements, and perceived challenges and opportunities. The findings from this survey are briefly discussed below and will guide the workstreams’ efforts.

Considering the rapid integration of AI into daily and professional life and recognising the dynamic nature of AI, the group plans to repeat the survey regularly to track the evolution of AI usage in OH.

This ongoing effort underscores the importance for OH professionals to be curious, informed, and proactive in leveraging AI to enhance practice intelligently, safely, and effectively.

Current and future use of AI tools

More than half of our survey respondents said they use AI within various aspects of their OH practice. However, interestingly, far fewer had plans to use AI (see graphs). Of those who are using and planning to use, uses include:

• Research, writing and editing
• Rewriting emails and reports for conciseness
• Drafting and clarifying reports
• Writing clinical notes
• Checking language and tone in communications, mainly emails
• Minute-taking during virtual meetings
• Writing policies, procedures and questionnaires
• Research and summarise online educational material and regulatory documents
• Researching rehabilitation plans
• Summarising medical conditions
• Writing employee leaflets
• Training material and media
• Data collection and analysis with predictions
• Regulation and compliance with auditing, risk assessment, policy and procedure development.

Although not indicative of all within the OH workforce, these themes were consistent across all respondents. Understanding these common themes (such as report writing, communication, administration, data analysis and research) provides us, as a working group with insight and direction.

Nerys (2023) predicted a far wider use of AI within OH, for example, risk reduction and risk management within the workplace, which does not appear to have played out yet. Perhaps this is due to ethical concerns noted in the survey and discussed below, alongside the insurance considerations discussed by Kenyon (2023).

The Future Vision: Survey respondents highlighted a number of potential benefits of AI in occupational health (OH) that, if realised, could transform OH practice and delivery. Common thoughts included:

• widening access to OH which will support SOM’s call for universal access (2020);
• standardisation, consistency and quality, thereby reducing criticism of OH, such as that from the Disability Forum (2023);
• identifying missing information, in other words, gaps in data not easily seen by people and analysing health surveillance data to “identify and interrupt occupational disease pathways early”, the respondents said.

This last point moves the use of AI from monitoring risks in the workplace which El-Helaly (2024) discusses, to reducing risks for employees, such as that seen in musculoskeletal disorders (Mileso, 2024).

Time-saving and improved workflow seem to be the most consistent forecast, which could assist with the workforce deficit in OH (DWP, 2023). Predicting improvement in evidence-based OH will be a boon for workplace health and necessary as Michell (2024) tells us. Other suggestions are support for those with neurodiversity and, thought-provokingly, a virtual workplace health assistant! Educating employers is another novel suggestion which could significantly improve the effective utilisation of OH (DWP, 2023).

Reading these suggestions and understanding current uses bodes well for OH if we retain the human touch and manage governance and obstacles.

Governance and obstacles

The survey revealed significant gaps in governance, policies, and procedures for AI use in OH. Most respondents highlighted the absence of clear ethical and regulatory frameworks as a critical barrier to adoption. Concerns centred on confidentiality and data security, with respondents expressing unease about potential breaches, misuse of sensitive health information, and inadequate safeguards in existing systems.

Many [respondents] emphasised the risk of ‘dehumanising’ OH care, stressing the need for AI to complement rather than replace expert human judgement.”

Ethical considerations emerged as another prominent theme. Respondents underscored the importance of ensuring fairness, transparency, and explainability in AI-driven decision-making. Many emphasised the risk of ‘dehumanising’ OH care, stressing the need for AI to complement rather than replace expert human judgement.

Reliability was another major concern, with respondents wary of errors, inaccuracies, and ‘hallucinations’ in AI outputs. The lack of consistent performance standards and robust validation processes further compounded this issue.

The survey also identified practical obstacles, including limited access to AI tools, insufficient training, and high implementation costs. Many respondents pointed to the need for comprehensive governance frameworks that address both ethical (soft) and regulatory (hard) requirements. These frameworks should include data protection protocols, rigorous evaluation processes, and education initiatives to build trust and confidence in AI systems.

Addressing these concerns is essential for the safe and effective integration of AI in OH. The literature emphasises the crucial role of soft (ethical) and hard (regulatory) governance in the healthcare system’s adoption of new technologies, including AI.

OH tools and platforms

The survey shed light on AI tools used in the OH sector, which included ChatGPT, Microsoft Copilot, Perplexity, Grammarly, and Apple Intelligence. Survey participants appear to rely on these tools for tasks such as report writing, email refinement, and data analysis.

In OH, there are digital software platforms that use AI for risk assessments, predicting health risks, engaging employees in digital reporting and ensuring compliance with regulations. Other AI-powered tools like Genny AI and Wellby automate documentation and risk analysis.

What we have done, is open doors to conversation and identify, to some extent, what the thinking is around AI in the general OH population.”

Despite these advances, only 9% of those who are not using AI or are unsure, are actively planning to use AI in the future, citing cost, data protection, and a lack of governance as significant barriers.

The UK supports AI in medical software through significant funding for diagnostics, research and innovation, targeting healthcare providers, researchers, and tech firms.

In 2024, the UK government allocated £1.5m through its Innovate UK’s Small Business Research Initiative (SBRI) to support AI-driven OH software development. This funding backs projects like cloud-based occupational health platforms using AI to monitor health trends and detect early signs of illness.

More recently, prime minister Keir Starmer outlined how the government wanted to use AI to “turbocharge” economic growth in the UK and kickstart “a decade of national renewal”.

Awareness of UK government funding among OH professionals, such as Innovate UK grants, remains limited, however.

While uptake grows, challenges include resource constraints, regulatory hurdles, and scepticism. Successful adoption requires clear frameworks, practical integration, and sustained collaboration among stakeholders to unlock AI’s full potential in healthcare.

Survey overview

Despite the survey population being varied in experience, role, and occupational health provision (for example encompassing in-house, OH provider, and self-employed), the feelings towards AI appear similar.

Some similarity was seen in job type, role, service provision, and length of experience. Across the different groups, obstacles and opportunities can be grouped according to the two groups: those currently using, or planning to use AI, and those who are not or who are unsure.

The clustering shows a complex web of concerns in a backdrop of fast-paced change.

Conclusions

Our conclusion is our beginning. As a working group, we haven’t even begun to reach our goal. What we have done, is open doors to conversation and identify, to some extent, what the thinking is around AI in the general OH population.

Further surveys and further directed discussion are needed, concentrating on reaching the aims outlined in the consensus statement and addressing that complex web of concerns.

Sign up to our weekly round-up of HR news and guidance

Receive the Personnel Today Direct e-newsletter every Wednesday

Email(Required) OptOut From time to time, we will send you emails about selected products, events and services from Personnel Today and OHW+ - but you can choose to opt-out at any time. If you do not wish to receive these emails, please tick this box.EmailThis field is for validation purposes and should be left unchanged.

Δdocument.getElementById( "ak_js_2" ).setAttribute( "value", ( new Date() ).getTime() );

/* = 0;if(!is_postback){return;}var form_content = jQuery(this).contents().find('#gform_wrapper_129');var is_confirmation = jQuery(this).contents().find('#gform_confirmation_wrapper_129').length > 0;var is_redirect = contents.indexOf('gformRedirect(){') >= 0;var is_form = form_content.length > 0 && ! is_redirect && ! is_confirmation;var mt = parseInt(jQuery('html').css('margin-top'), 10) + parseInt(jQuery('body').css('margin-top'), 10) + 100;if(is_form){jQuery('#gform_wrapper_129').html(form_content.html());if(form_content.hasClass('gform_validation_error')){jQuery('#gform_wrapper_129').addClass('gform_validation_error');} else {jQuery('#gform_wrapper_129').removeClass('gform_validation_error');}setTimeout( function() { /* delay the scroll by 50 milliseconds to fix a bug in chrome */ }, 50 );if(window['gformInitDatepicker']) {gformInitDatepicker();}if(window['gformInitPriceFields']) {gformInitPriceFields();}var current_page = jQuery('#gform_source_page_number_129').val();gformInitSpinner( 129, 'https://www.personneltoday.com/wp-content/plugins/gravityforms/images/spinner.svg', true );jQuery(document).trigger('gform_page_loaded', [129, current_page]);window['gf_submitting_129'] = false;}else if(!is_redirect){var confirmation_content = jQuery(this).contents().find('.GF_AJAX_POSTBACK').html();if(!confirmation_content){confirmation_content = contents;}jQuery('#gform_wrapper_129').replaceWith(confirmation_content);jQuery(document).trigger('gform_confirmation_loaded', [129]);window['gf_submitting_129'] = false;wp.a11y.speak(jQuery('#gform_confirmation_message_129').text());}else{jQuery('#gform_129').append(contents);if(window['gformRedirect']) {gformRedirect();}}jQuery(document).trigger("gform_pre_post_render", [{ formId: "129", currentPage: "current_page", abort: function() { this.preventDefault(); } }]); if (event && event.defaultPrevented) { return; } const gformWrapperDiv = document.getElementById( "gform_wrapper_129" ); if ( gformWrapperDiv ) { const visibilitySpan = document.createElement( "span" ); visibilitySpan.id = "gform_visibility_test_129"; gformWrapperDiv.insertAdjacentElement( "afterend", visibilitySpan ); } const visibilityTestDiv = document.getElementById( "gform_visibility_test_129" ); let postRenderFired = false; function triggerPostRender() { if ( postRenderFired ) { return; } postRenderFired = true; jQuery( document ).trigger( 'gform_post_render', [129, current_page] ); gform.utils.trigger( { event: 'gform/postRender', native: false, data: { formId: 129, currentPage: current_page } } ); if ( visibilityTestDiv ) { visibilityTestDiv.parentNode.removeChild( visibilityTestDiv ); } } function debounce( func, wait, immediate ) { var timeout; return function() { var context = this, args = arguments; var later = function() { timeout = null; if ( !immediate ) func.apply( context, args ); }; var callNow = immediate && !timeout; clearTimeout( timeout ); timeout = setTimeout( later, wait ); if ( callNow ) func.apply( context, args ); }; } const debouncedTriggerPostRender = debounce( function() { triggerPostRender(); }, 200 ); if ( visibilityTestDiv && visibilityTestDiv.offsetParent === null ) { const observer = new MutationObserver( ( mutations ) => { mutations.forEach( ( mutation ) => { if ( mutation.type === 'attributes' && visibilityTestDiv.offsetParent !== null ) { debouncedTriggerPostRender(); observer.disconnect(); } }); }); observer.observe( document.body, { attributes: true, childList: false, subtree: true, attributeFilter: [ 'style', 'class' ], }); } else { triggerPostRender(); } } );} ); /* ]]> */