Artificial intelligence – how will it change our work, our roles and our charities?
Are we missing out on transformative opportunities or could it be a solution to some of our most stubborn problems?
After working with cross-sector technology consultant Marcel Britsch recently, I learnt more about the opportunities and challenges that AI presents to Cycling UK and to charities more broadly.
Be wary of the hype
AI is all over the news as the industry makes a play for investment. Everyone, including our trustees and staff, grasp at AI as a way of doing away with menial admin work or gaining a competitive edge.
But you will have seen this bandwagon-jumping before – remember crypto? Just two years ago a trustee was trying to persuade me that cryptocurrencies were the next great thing for fundraising but how many charities do you know taking donations in cryptocurrencies? The risk and complexity involved have kept it niche several years on.
Gartner’s “AI hype cycle” reflects our view – that we’re actually still at an early stage in its development.
Identify the problem you want to solve
Through talking with Marcel, one of the things we realised at Cycling UK was that we aren’t always precise enough about the specific problem that we want to solve.
Similarly, charities too often jump to a technological solution, perhaps conflating several issues into one, or failing to see a potential opportunity to provide new value.
This can mean we miss the right solution or a great opportunity because we haven’t properly identified and defined the problem. Some problems may not have a tech or AI solution or may just not be all that strategically important.
Focus on your strategic priorities
AI solutions to relatively low-order problems may seem to be a quick win, so they get shunted up the priority list. But are these our organisational priorities?
For example, at Cycling UK we might love the idea of using AI to boost our content creation, but if our strategic priorities are around maximising unrestricted income, how much will that help move us forward?
As we prioritise, we need to find the sweet spot between the value a solution will provide vs its cost, risk and constraints. With the emergence of large language models such as chatGPT, content can be created quicker, but are we sure that the quality is what we want to put in front of our audience?
Test and learn
As with any tech development, but specifically with something as new and emergent as AI, it’s very hard and risky to commit to a final, big solution early on.
While it might be tempting to make sweeping changes, we have seen far better outcomes by exploring small, short-term initiatives.
These baby steps are not only financially safer, but more importantly allow an organisation to better define the problem to be solved, and the solution to be delivered and evaluated.
Assess the broader impact
AI is not a cost-free option, both financially and ethically.
The question of bias in datasets is well publicised, but the environmental cost of AI should also be considered – how many equivalent transatlantic flights of carbon is it worth to run casual queries or generate a poem about your annual report?
Once you have identified the problem and established that AI is a potential solution, you should assess the broader, non-functional impact of AI and feed that into your business case.
Don’t underestimate the need for data
The insights AI can generate are only as good as the data it’s trained on and the questions we ask it.
Why do we think most customer support chatbots are so bad? Largely because the data that is given to them does not reflect real users’ needs. This is obvious and fixable where we talk about narrow problems, but becomes an impossibly hard undertaking where we expect AI to have the same knowledge and context as us human workers.
Humans – at least for now – are far superior at absorbing and distilling information, and synthesising insights in complex and variable environments than any current technological system. This puts hard constraints on what AI, at this point in time, can do.
Be realistic
AI isn’t a solution to all our woes. In fact, what AI can really help with presently is often quite counter-intuitive: At the start of 2025, AI is surprisingly good at tasks we humans intuitively find hard, such as detecting patterns and applying them.
However, AI still can’t manage what we humans think of simple, menial tasks. While AI can identify faces and generate images, it won’t become your PA or provide you with that perfect project plan any time soon.
At Cycling UK, we found that many of the issues we wanted to tackle with AI were actually ones we could tackle with existing technology, or through changes to our work practices. The areas where AI could potentially be most useful to us were in analysing our big datasets in monitoring and evaluation work.
Of course, this is a fast-moving area and we’ll be keeping an eye on the technology as it develops and perhaps becomes more relevant to us and our priorities. The major outcome for me as a charity leader was feeling far more confident about having explored the options and being better equipped to quickly and simply assess potential opportunities, based on our strategic priorities.
Sarah Mitchell is chief executive of Cycling UK
Related articles
