AI in Your Hospice: Why Every Leader Needs a Framework, Not Just an Opinion

Ask a room full of hospice leaders what they think about AI and you'll get a full spectrum of answers. Some will tell you it's the most exciting efficiency opportunity they've seen in years. Others will tell you it's a threat to the human, compassionate culture that hospices are built on. Most will probably tell you they haven't really thought about it in depth yet. There's a service to run and people to care for. 

All of these reactions are understandable. But here's the honest truth. Many of your colleagues have probably already started using AI, whether you've thought about it or not. 

There are real challenges facing the Hospice sector workforce, a number of which are covered in this blog post on ‘Rethinking How Hospices Attract and Keep Early-Career Talent’. Across the sector we have a workforce under real pressure with rising demand for care, competition for talent with the NHS, constrained resources etc. Against that backdrop, it's no surprise that AI is creeping into hospice workplaces from the bottom up, one ChatGPT tab at a time. Often this is happening well before any leadership team has decided what "good" looks like. 

This post isn't a technical guide to generative AI. It's a challenge to hospice HR teams and senior leaders to get ahead of this rather than let it happen to you. That means building a clear, consistent and ethical framework - one where AI is informed, not led, by the people who use it. 

Do we need to use AI? 

This is the question lots of leaders may ask. Often with a slightly hopeful tone that the answer might be "no”. The honest answer is that the decision has largely already been made for you. The 2026 Charity Digital Skills Report found that 79% of UK charities now use AI in some form, up from 76% the year before. However, only 4% describe it as fully embedded in their strategy, with a further 34% using it in pockets without any real plan. In other words, almost everyone is doing it, almost nobody is managing it. [Charity Digital Skills Report 2026] 

That gap could really matter in a hospice. We're working with people's most sensitive personal information, at the most sensitive point in their lives, with a workforce that is already stretched. Demand for palliative and end-of-life care is only going to grow. None of that is a reason to avoid AI but it is a reason to be deliberate rather than accidental about it. 

So "do we need to use AI?" is probably the wrong question. The better one is “are we going to decide how AI gets used here, or are we going to find out after the fact?” 

Used well, AI won't replace the compassion at the heart of hospice care but it might give your people back some of the time that compassion needs. Time spent redrafting a policy, writing up meeting notes or building a job advert from scratch is time that could be spent with a colleague, a volunteer or a family. 

How do you engage colleagues in the conversation? 

Once you accept that AI is already in your organisation in some form, the next job is to bring it into the open. Some useful questions to put to your teams:

• Are you already using AI tools like ChatGPT or Copilot? What's your experience been?
• How could AI genuinely help you do your job better? 
• Is there anything that worries you about how AI might affect your work, or your colleagues' jobs? 
• Have you seen AI used well anywhere (inside or outside the hospice)? Could we learn from that? 
• Are there areas of our work where you think it would be inappropriate, or unethical, to use AI? 
• What support would help you feel more confident using it well? 

An ethical framework for AI in the workplace 

This is where it might be worth every hospice leadership team slowing down and doing some proper thinking. Right now the sector has a governance gap that should worry trustees as much as HR Directors. 

Charity Commission research into trust in charities found that only 3% of trustees believed their charity used AI (rising to 8% in larger charities). The real figure for actual use is closer to four in five. It means, in most charities, AI decisions are being made informally by individual staff, with almost no visibility at board level, at exactly the point where trustees remain legally accountable for data protection, confidentiality and bias risk. 

A workable ethical framework doesn't need to be a fifty-page policy document. It needs to answer a handful of practical questions clearly enough that every member of staff and every volunteer knows where they stand: 

• Confidentiality first. What patient, family, staff or donor data must never go into a public AI tool? Given the nature of the information hospices hold, we can’t assume that "people will use common sense." 
• A human always decides. AI can help draft, summarise or suggest. It should never make the final call on a recruitment decision, a disciplinary outcome, a clinical matter or anything touching on a family's care. That distinction, AI-informed, not AI-led could well be the single most important line in any framework. 
• Be transparent. If AI has materially shaped a piece of communication, a policy or a decision, say so. 
• Consistency over individual choice. Right now, AI use in most organisations is a patchwork of personal decisions. A framework should apply the same standards whether you're in fundraising, clinical services or the CEO's office. 
• Keep it under review. This technology is moving quickly enough that a framework written today will need revisiting within the year. Build that review point in now. 

Building skills and confidence across your team 

Even with the right framework in place, none of it matters if your people don't feel confident using it. CIPD's most recent research found that while 61% of UK employers now permit staff to use generative AI at work, only 35% actually provide any training to help them do it well. CIPD, Generative AI at work: can it deliver the productivity boost UK employers need?

A few principles that tend to work: 

• Start with awareness, not expertise. Not everyone needs to become a power user. Everyone does need to understand what the organisation's framework allows, what it doesn't and why. 
• Protect time to experiment. Asking already-stretched hospice staff to "just go and learn AI" on top of a full workload won’t work. Give people permission and some protected time to try it on something low-risk. 
• Learn from your own early adopters. Peer-led show-and-tell sessions, where a colleague demonstrates something that actually saved them time, is likely to land far better than a mandated e-learning module. 
• Don't forget the board. Given the governance gap mentioned above, trustees need their own, tailored version of this confidence-building. They need enough information to be able to ask good questions and hold the organisation to accoun

Graham Atkinson 

Atkinson HR Consulting Ltd