How can generative AI be used in a way that is both genuinely useful and ethically sound?

Last week’s webinar, hosted by Grace Tattersall and featuring Jeremy Worrell – former board and C-Suite advisor, Head of Digital & Data, and author – explored the question many NFP organisations are now grappling with: how can generative AI be used in a way that is both genuinely useful and ethically sound?

Refraining from presenting AI as a perfect one-size-fits-all solution, the session focused on what organisations are working to achieve, the obstacles they face, and how AI might fit into that picture.

What “AI” Really Means in Practice

One of the first points made was that when people talk about AI today, they’re usually referring to a very specific subset: generative AI. Tools like ChatGPT, Gemini and Copilot fall into this category. They are designed to perform tasks that normally require human input—writing, summarising, analysing or generating ideas.

While AI itself has existed for decades, generative AI is relatively new and has become widespread largely because it is easy to use. Unlike older systems, it doesn’t produce identical outputs each time. Two identical prompts can result in two different responses. This built-in variability makes it flexible, but also introduces an element of unpredictability that organisations must take seriously—particularly those working with vulnerable groups or sensitive information.

Where the Opportunities Lie

AI is not as a technology problem, it is a value question. Where does it actually help?

Improving efficiency. Many everyday tasks such as writing grant applications, summarising reports, or preparing communications can be partly automated. The impact here is not about replacing people, but about freeing up time. AI can take on preliminary work, leaving staff to review, refine and make final decisions.

Enabling work that never quite gets done. Most organisations have activities that are recognised as important but repeatedly deprioritised such as deeper data analysis, improved reporting, or better donor engagement. AI can lower the effort needed to start tackling these areas.

Simplification. Some processes exist purely because of how work has traditionally been done. AI creates the opportunity to remove or redesign parts of these processes altogether. That might mean reducing duplication, eliminating manual steps or even replacing certain outsourced services.

Transformation. This is about rethinking how services are delivered. It’s where AI could have the greatest long-term impact, potentially improving how beneficiaries are supported, but it is also the hardest area to get right. It involves more uncertainty, more risk and often more investment.

The Main Barriers to Adoption

Despite the clear opportunities, adoption is far from straightforward and there are several barriers that more organisations seem to come up against.

One of the most significant is a lack of focus. For many organisations, AI is something individuals use casually but as a wider organisation there is often little coordination and no clear prioritisation. This feeds into the need for leadership. Even in smaller organisations, assigning someone responsibility for AI can make a measurable difference. Without that, progress tends to stall.

Skills are often raised as a concern, Jeremy acknowledged, but he also dismantled that perception, sharing that many modern tools require little or no coding knowledge. In fact, the level of technical skill needed to get started is lower than many expect. The bigger issue is often confidence rather than capability.

Data, too, is sometimes seen as a blocker. While it’s true that some use cases depend on high-quality data, not all do. Many applications rely on general knowledge or external information. In some cases, AI can even help identify and fix data quality issues. So while data matters, it doesn’t need to prevent organisations from getting started.

But there is one barrier that stands out the most and seems to be the hardest to tackle: ethics, especially in this sector.

Navigating Ethical Challenges

For not-for-profits, ethical considerations are central. The use of AI raises concerns about bias, accuracy, transparency and trust. There are also broader questions about environmental impact and how AI-generated outputs are perceived by donors and beneficiaries.

Staff attitudes vary widely. Some are enthusiastic, others cautious, and some instinctively uncomfortable without being able to clearly articulate why.

Jeremy emphasised that these concerns should not be dismissed but instead quite the opposite, they need to be openly discussed to determine how risks can be mitigated, perhaps, for example, by carefully reviewing outputs, asking AI to explain its reasoning, or limiting its use in sensitive contexts. In other cases, organisations may decide certain uses are simply not appropriate.

Rather than treating AI as something to either fully embrace or completely avoid, organisations can position themselves along a spectrum. Most will sit somewhere in the middle, balancing opportunity with risk. Once your organisation’s position is clear, it will then become easier to put governance in place. Policies, controls and safeguards can then be aligned with the organisation’s chosen approach.

Governance and Practical Steps

Once you have that approach in place, it is important that governance goes beyond simply having an AI policy. While policies have value, they are largely defensive. They help manage risk, but they don’t, on their own, create benefit.

More meaningful progress comes from combining governance with action. This includes:

• Appointing someone to coordinate AI activity
• Identifying a small number of priority use cases
• Building internal understanding and confidence through training
• Ensuring clear oversight and accountability

All of this can only be ensured through training, but providing access to tools is not enough. Staff need to understand how and when to use them, and to feel comfortable doing so. Because human involvement in using AI is imperative, as Jeremy said, “the co‑pilot is not the pilot”. AI supports decisions, but responsibility remains with people.

Looking Ahead

Closing out the session, Jeremy speculated on the broader reflection on the role AI is likely to play in the future. Compared with previous waves of technology, such as mobile or cloud computing, generative AI is expected to have a more fundamental impact.

This is partly due to its ability to replicate aspects of human work and partly because it is already accelerating its own development. The pace of change is unlikely to slow, which makes it difficult for organisations to stay fully up to date.

For not-for-profits, this presents a dilemma. Ignoring AI entirely may mean missing opportunities to improve impact. At the same time, adopting it without careful thought introduces risks.

For organisations in the not-for-profit sector, the priority remains the same: delivering value to beneficiaries. AI should be considered in that light. If it helps extend reach, improve effectiveness or free up resources, it is worth exploring. If not, it can wait.

The challenge now is not whether to engage with AI, but how to do so in a way that is thoughtful, responsible and aligned with each organisation’s values.

Jeremy Worrell now dedicates himself to working with non-profits to help them adopt AI technologies for good. In practice, this is about coaching, training and ethics … even ahead of technology. In 2025, Jeremy helped the RSPB, Mind, Save the Children and the Marine Conservation Society to form robust plans, circumventing common blockers. He’s now leading AI for the National Deaf Children’s Society. To learn more about Jeremy’s work, you can connect with him on LinkedIn or email him.

Women in Tech: Thriving in the Digital Age | AQ