AI raised big questions for our team. Here is what we learned, what we built, and what we are still wrestling with. By Leah Milner-Campbell, outgoing Director of People and Resources
Like many organisations, we’ve been grappling with artificial intelligence. Not just what it can do, but what it means for how we work, what we value, and who we are as an organisation. The Finance Innovation Lab team came to AI from very different places. Some colleagues had been exploring tools independently for months, learning through trial and error what worked and what did not. Others had serious ethical concerns and were not sure where to start.
So in September 2025 we ran a whole-team workshop exploring our AI expertise and concerns. We asked ‘what guidance and support do you need from the Lab to use AI responsibly?’
Through this, four key themes emerged.
The first was accuracy. As one colleague put it, it “can be totally wrong, yet sound so right.” Because of this, we must apply the same critical thinking we use with any other source. Just as you would not copy and paste Wikipedia, the same logic applies here.
The second was data protection and transparency. Colleagues raised practical concerns about GDPR compliance and how to handle AI use in external work. Our principles outline the importance of only using GDPR-compliant tools with storage in Europe for any sensitive or confidential data. The principles also outline how we will always tell each other when we have used AI, and be open with partners and stakeholders too. FYI, AI helped to write this blog!
The third was justice and equity. AI models are trained on data that reflects existing inequities, particularly Western cultural perspectives and current power structures in finance. AI replicates what already exists. It will not produce transformative thinking unprompted. We have to bring the critical lens ourselves.
The fourth was the environment. Our work focuses on environmental justice, and we cannot ignore the huge energy costs of AI tools. We did not resolve this tension, because it cannot be fully resolved. We recognise that other IT usage costs energy too. We committed to honesty about the contradiction, and using AI purposefully rather than wastefully.
From those conversations, we have started to develop a set of principles to help us learn to use AI responsibly and effectively. They cover seven areas: learning and experimentation, critical engagement and quality control, transparency, data protection, justice and equity, environmental responsibility, and maintaining human skills and connections.
We have used our organisational values to shape our reflections. Ambition, for us, means bravery and learning, and that includes being willing to engage with AI even when you feel uncertain or sceptical. Experimentation is how we grow. Collaboration meant we could not have some colleagues racing ahead while others felt left behind or unheard. We needed shared expectations, not individual choices made in isolation. And empowerment meant making sure everyone on the team, regardless of their role or starting point, had what they needed to engage critically and confidently.
AI is also reshaping the financial system we exist to transform. We are integrating this into our strategies and workplans. Our 2023 report on AI in financial services identified emerging risks including systemic financial instability. Those risks are growing. AI-driven decision-making is concentrating power, accelerating market dynamics, and creating new vectors for crisis. We believe regulators need a precautionary approach, acting ahead of harm rather than responding to it.
For several colleagues, concerns about AI are still palpable. Colleagues worry that over-reliance on AI could diminish our own capabilities, and the collaborative relationships that make our work possible. AI will also reinforce your existing perspectives rather than challenge them. For systems change work, that is a significant limitation. Diverse human judgement remains irreplaceable.
We are sharing this because the questions we wrestled with are ones many purpose-driven organisations are facing right now. We do not claim to have got this right. We are making progress, and being transparent about that process feels consistent with who we are trying to be.
If you are working through something similar, we would love to hear from you.
*Parts of this blog were edited using AI software