As part of Banking of Values Day 2025, the Global Alliance for Banking on Values (GABV) is exploring the relationship between one of the defining technologies of our time–generative AI–and banking. We’re conscious that for good or bad, AI will shape so much of how our future looks and also how people understand it.
This year, GABV began by thinking about how AI describes banking to the rest of the world. They paid particularly close attention to the way that, when someone asks a generative AI chatbot (like ChatGPT or Claude) about banking, it responds with information that reflects conventional banking, with little about banking on values.
For example, we might ask it: what is a bank for?
And it might respond:
“A bank is for maximising profits for its shareholders while maintaining financial stability and regulatory compliance.”
While financial institutions like the ones in the GABV network would probably say:
Financial institutions, such as credit unions, can drive positive change in society, while providing essential financial services.
Generative AI has these kinds of limited responses to banking because of the materials it draws from, which are often dated news articles and ads, academic papers, and other sources. Its understanding is constrained and, in this way, many people’s understanding of banking is constrained too. What AI says now really matters, for so many it’s becoming the primary source for information–about banking and about so much else.
AI search doesn’t always show the truth - it searches for what is statistically likely. In practice, this means that ''AI doesn’t find the best, it finds the most mentioned."
Researchers warn that this is producing two epistemic risks:
- Synthetic consensus: A false sense of agreement created not by people thinking the same thing, but by AI systems echoing the same claims again and again. When the same phrase or assertion keeps appearing, it acquires the feeling of consensus — not because it is accurate, but because it has been algorithmically repeated.
- Knowledge collapse: Generative systems tend to converge on the “average” of what they have seen, pushing minority, niche, or emerging perspectives out of view. This erodes nuance and originality and can slowly distort public understanding as visibility drifts toward whatever is most conventional and most repeated — not necessarily what is most correct.
Search Again: ChatGABV
Since generative AI is the next search engine - a one-stop-shop for learning for people all around the world - what it thinks banking is matters a great deal. The Global Alliance for Banking on Values wants to improve the dataset AI has to work with–getting more coverage for the kind of values-led banking that GABV members do.
To explore what “rewriting” AI and rethinking banking could look like, GABV made ‘ChatGABV’ - an interactive experience that considers how a wiser AI would answer questions about banking compared to a more conventional AI. You can explore that experience here.
Help AI Rethink Banking
AI and banking have more in common than you might think. Each helps people around the world generate new things from old. Each has the potential to do a lot of good for the world–but only if human beings intentionally steer them. That’s what GABV is interested in doing–using both these processes for the good of people and the planet.
You can find out more on how to shape AI in our guide.
Tell us what you think about AI and Banking.