Responsible AI in the NHS: ethics lead the way

How often have you thought about ethics in the last year? This week at NHS ConfedExpo in Manchester, the buzz around artificial intelligence was impossible to ignore. From the main stage session “Are we ready for AI?” to the AI learning theatre discussions on workforce transformation and productivity, leaders grappled with a shared concern: technology is advancing faster than our systems, people, and governance can keep up. In light of this, I wanted to share something to help stimulate conversations you can have with your boards and people teams about the need for ethics in how we engage with AI. 

As someone who has spent most of my career in NHS HR and workforce leadership, I left the conference both energised by the potential of AI and firmly convinced that ethics cannot be an afterthought.

Indeed, ethics must be at the heart of how we adopt AI in workforce management. 

Through many different conversations, I know that AI is occupying the thoughts of many of you, and the technology is already making inroads into scheduling, recruitment, performance analytics, staff allocation, and even predictive workforce planning. In an NHS facing persistent staffing pressures and the ambitions of the 10-Year Health Plan, tools that improve efficiencies matter. Yet the conversations at ConfedExpo repeatedly returned to the same critical questions: Who owns the data? How do we prevent bias? And crucially, how do we ensure AI supports rather than erodes the human core of healthcare, as I wrote about in my previous blog.

Ethics in AI workforce applications starts with transparency. Staff need to understand how algorithms influence shift patterns, promotion shortlists, or development opportunities. During one session on non-clinical AI, delegates at ConfedExpo highlighted real examples where opaque systems risked deskilling or alienating teams. When a clinician or administrator feels their professional judgement is being overridden by a black-box model, trust diminishes. We have seen this before with other technologies in the NHS, and where psychological safety and engagement directly impact patient care, we cannot afford that erosion of trust.  

We must be on our guard. I do worry about what this does to entry-level roles and early career learning opportunities. 

Bias is perhaps the most pressing ethical risk troubling people professionals. Historical workforce data in the NHS reflects long-standing patterns of inequality in career progression, disciplinary outcomes, and access to flexible working, particularly for ethnic minority staff and those with disabilities. Feeding that data into AI without rigorous scrutiny, we risk amplifying these disparities. An algorithm trained on past recruitment might inadvertently favour certain educational backgrounds or career paths that do not reflect the diverse talent we need for modern, community-based care models.  

At ConfedExpo, panel discussions emphasised the need for diverse development teams, regular bias audits, and ongoing equality impact assessments. This is not bureaucracy; it is basic fairness and good risk management. 

Then there is the human element. AI can reduce administrative drudgery (thank goodness!) freeing time for patient care and leadership, but it must not become a vehicle for intensified surveillance or unrealistic performance targets. Workforce analytics should empower managers to have better conversations, not replace them. We must guard against algorithmic management that strips away the relational aspects of people leadership. The NHS People Promise speaks to belonging, compassion, and recognition. AI systems must be designed to reinforce these values, not undermine them. 

I never thought I would say this as someone with a reasonably high-risk appetite, but governance is essential. NHS organisations need clear frameworks for responsible AI adoption: ethics committees or boards with people teams, clinical, digital, and staff-side representation; transparent procurement that demands explainability; and investment in digital and data literacy for leaders and front-line staff alike.

Regulation and national guidance are catching up, but local accountability cannot wait.  

None of this means slowing down innovation (again, thankfully!). The NHS workforce challenges are too acute for that. Done ethically, AI can help us reskill teams for new care models, improve retention through better work-life balance, and attract digital-native talent. Ethics is not a brake, it is the steering wheel. It ensures we harness AI to build a more sustainable, inclusive, and compassionate workforce. 

As I reflect on the energy and honest debate at ConfedExpo, my message to people leaders is clear: it is time for us to step up. We are uniquely placed to champion responsible AI. Work with your digital and clinical colleagues (in fact, make them your best friends!) to embed ethical principles from day one. Engage staff and unions early. Measure success not just by efficiency gains but by staff experience, fairness, and patient outcomes. 

The technology will keep evolving, but our values, equity, dignity, compassion and public service must remain constant. If we get the ethics right, I believe that AI can be a genuine force for good in tackling the workforce pressures we all face.

Get it wrong, and we risk damaging the very thing that makes the NHS special: its people. 

We are keen to showcase and share learning and good practice from across the NHS on how organisations are approaching AI transformation. If you have examples or insights you would like to share, please contact our communications team.