Building on his previous article about practical steps to AI production, Ali Bovis at Version 1 delves into the importance of establishing essential trust in public sector AI projects
Advanced technology represents one of the most impactful opportunities to improve public service delivery in our generation. The IMF estimates that AI can boost productivity by 1.5% annually, potentially worth £47 billion to the UK over a decade. However, the public sector faces a fundamental challenge: the absolute necessity of trust before commitment.
Unlike revenue-draining private sector failures, AI failures in public services directly impact vulnerable citizens’ lives. These failures are also played out in public, with outsized media interest compared to private sector projects. As such, integrating this technology into public services demands a fundamentally different approach, one that prioritises responsible development and trust-building as the foundation for expanded capability.
Establishing trust: The Cafcass example
The UK ranks third in the Government AI Readiness Index; however, ranking highly and successfully implementing AI responsibly are two distinct challenges. Cafcass (Children and Family Court Advisory and Support Service) demonstrates how public sector organisations can harness AI’s power whilst building essential trust.
It faced a significant challenge: 1,500 Family Court Advisers supporting over 140,000 children annually, producing 80,000 letters monthly to help children understand complex legal processes. The opportunity was clear, but the imperative to implement safely was paramount. Communication errors with vulnerable families were not an option.
Working closely with Family Court Advisers, and in partnership with Version 1, we co-designed Scribe: a focused solution embedded directly into their ChildFirst case management system. This collaborative methodology was crucial, as involving practitioners from the outset addressed real needs whilst building user confidence.
Scribe’s capabilities were carefully scoped to deliver value whilst maintaining strict quality controls. The system automatically drafts communications from information about children’s cases, adapts content to children’s reading levels, translates into over 100 languages, and creates audio versions. Most importantly, these capabilities operate within defined parameters, with human oversight maintaining ultimate responsibility.
The results demonstrate responsible AI implementation: Scribe is estimated to save up to 60,000 advisor hours annually, all released back to frontline work. More significantly, it does this whilst providing clearer, more accessible communication that reduces stress for children and families.
Expanding capability through proven trust
Scribe has started to establish crucial trust amongst advisors, management, and families. This foundation is proving invaluable as our team works with Cafcass to explore new opportunities to improve.
Cafcass is exploring multiple other areas where AI can support and bring efficiencies into its operations. The strong foundation established by Scribe demonstrates a crucial principle: once trust is earned and foundations are in place, organisations can progressively add new AI capabilities to amplify their impact. Starting with focused, manageable challenges rather than attempting to solve complex problems immediately allows building confidence and expertise incrementally.
Realising AI’s opportunities
PublicFirst highlighted 61% of public administration workers report increased overwork, with 70% saying morale has decreased. Meanwhile, two-thirds of managers agree AI will change public sector operations forever, but just 12% have significantly deployed AI tools. This gap represents both challenge and opportunity.
Our approach on the collaborative Cafcass project demonstrates sustainable AI implementation. The same principles that made Scribe successful – collaborative design with practitioners, integration with existing systems, human oversight, and prioritising trust – are being applied to document summarisation and genogram creation.
Both capabilities can leverage the same secure, scalable AI infrastructure developed for Scribe, with similar authentication, data handling, and quality processes. This exemplifies systematic capability building.
Building trust for transformation
This project demonstrates that AI’s power in public services extends beyond technology. It’s about building sustainable trust that enables continuous innovation. By starting with focused, well-executed implementations, the foundation for expanding AI capabilities while maintaining responsible approaches can be built.
Beyond efficiency gains, AI can make public services more accessible, personalised, and responsive. From multi-language communications breaking down barriers, to intelligent document analysis surfacing critical insights, and automated processes freeing professionals for direct citizen engagement – the potential is extraordinary.
This partnership offers a blueprint for responsible AI implementation across the public sector. Rather than wholesale transformation, success lies in identifying specific, high-value use cases, implementing them with rigorous attention to user needs and safety, and using early successes to build trust for broader adoption.
The old saying that trust is earned is never truer than in relation to delivering essential public services. Those bodies that successfully harness AI’s power will prioritise trust building alongside technological advancement. The two are intrinsically linked. This is the most effective way to get public perception onside and deliver tangible improvement. Through responsible, trust-centred approaches like the partnership detailed above, we, as a technology community, can directly contribute to delivering better outcomes for citizens whilst building confidence in the full potential of AI to positively transform UK public services.
Your trust building journey starts now
The Cafcass example proves that responsible AI implementation isn’t just possible, it’s transformative. But trust isn’t built through strategy documents or pilot programmes alone. It requires action.
Take these three steps within the next 30 days:
- Identify your Cafcass moment: Look for a single, high-volume administrative task that impacts citizens or your teams. Map the human hours involved and the potential for improvement. This is your trust-building foundation.
- Convene your trust team: Bring together your most respected frontline practitioners, your most cautious risk managers, and your most ambitious digital leaders. If they can’t co-design a solution together, your citizens won’t trust it either.
- Define your success differently: Set metrics that matter to citizens, not just efficiency targets. Cafcass didn’t just save time and money; they reduced stress for children in need. What equivalent citizen outcome could transform your service’s reputation?
Your organisation’s AI journey doesn’t need to start with transformation. It needs to start with trust. And trust starts with taking the first, measured step.
This work is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International.
