As artificial intelligence becomes more embedded in our daily work, we’re taking deliberate steps to ensure it’s used responsibly, transparently, and in ways that reflect our values. That’s why we’ve set up an internal AI steering group at CDPS – to guide our approach, shape policy, and support the CEOs with recommendations on AI governance. 

Why this matters 

AI tools are already helping us summarise, translate, and draft content more efficiently. But with that power comes responsibility. Our internal usage survey showed that while enthusiasm for AI is high, there’s a clear need for shared standards, better visibility of what’s being used, and more support for ethical and inclusive use. 

We wanted to make sure everyone at CDPS regardless of role or experience, feels confident using AI in a way that’s safe, fair, and aligned with our service standards and the Welsh context. 

Introducing the AI Steering Group 

We invited colleagues from across CDPS to join a new group. We looked for a mix of experience levels and disciplines because good governance needs diverse perspectives. We have representatives from all areas of the organisation. 

The group has met a few times (once a month) and: 

  • Defined usage guidelines: Clarify what AI tools and services can (and can’t) be used, and why (our simple rules are below). 
  • Agreed we will maintain an AI inventory, to keep a live record of who’s using what (and review it regularly).  
  • Laid the foundations for governance: Set the tone for responsible AI use across CDPS and beyond. 
  • Manages risk: Identifying and mitigating corporate risks related to AI. 
  • Acts as an internal regulatory panel: We will ensure transparency, reinforce the AI inventory, and clarify the group’s role. 
  • Promotes responsible adoption: Encourages positive engagement with AI, while considering the Welsh language, and service standards. 
  • Aims to understand the environmental impact: Exploring ways to measure the environmental impact involved in the use of AI tools.   
  • Embeds training: Make AI literacy part of our mandatory staff training — just like GDPR. 
  • Communicates clearly: We share what we’re doing, why it matters, and how others can get involved. 
  • Agreed we will invite external review: We’ll work with a critical friend (like colleagues from The Alan Turing Institute with whom we already have a good relationship) to challenge and improve our approach.

We discuss hot topics, providing an internal critical voice so we can take measured steps with AI – we often play cynics to keep a clear view on the topic. 

What we’re learning 

Our internal survey surfaced some key insights: 

  • ¾ of staff are already using AI (70% of staff completed the survey). 
  • Common tools include Chat GPT, Copilot, and Gemini (Deep Seek/Perplexity). 
  • 84% use free versions of these tools. 
  • Most began using AI tools months/years ago. 
  • 96% say AI tools save them at least 1-2 hours a week (max was 5hrs!). 
  • People reported using AI for:  
  • Increased efficiency and productivity 
  • Enhanced content quality and consistency 
  • Improved decision-making and data analysis 
  • Enhanced learning and understanding 
  • Reduced workload and stress 
  • Enhanced creativity and innovation 
  • Improved communication and collaboration, Cost reduction and return on investment (ROI). 
  • Concerns include over-reliance on AI, ethical considerations, and environmental impact. 
  • 84% of line managers knew about their direct reports’ AI usage. 
  • 80% of respondents indicated they had not received any training on AI tools.

Our rules

  • Is the info public or general knowledge? Free tools OK. 
  • Does it contain personal or business-confidential data? Use paid Copilot only.
  • Unsure? Ask a #bwrdd-ai member or your friendly DPO. 

What’s next 

We’re now: 

  • Opting for a wider rollout of MS Copilot – a 12-month beta with all staff who would like a license (everyone has to outline their purpose and intent and commit to reporting back) 
  • Planning communications 
  • Exploring partnerships for external review. 

We’re just getting started and we want this to be a shared journey. If you’re interested in contributing ideas, we’d love to hear from you. 

Let’s work in the open 

AI in some form is here to stay. Let’s make sure we’re using it in ways that are thoughtful, inclusive, and aligned with our mission to serve the people of Wales.