Let's get smarter with AI implementation

In this article, Ross Murray talks directly to local authority leaders and teams that have started on their AI journey. He explores common pitfalls and sets out four essential principles for implementing AI effectively, ensuring councils can harness its potential for cost savings and improved services while avoiding the risks of rushed or poorly planned deployment.

At the moment, it seems like wherever you look people are talking about AI. Local government is no different, with councils across the country exploring new tools and piloting them. AI undoubtedly presents a massive opportunity to address the growing financial pressures and increasing demand for public services.

Reports indicate that strategically implementing AI across local government could generate productivity gains worth up to £8 billion annually across England and Wales. That equates to approximately £325 per household. Many see the potential of AI as a necessity to clear existing backlogs, cut operational costs, and improve service delivery.  

There is a but…

While there is potential for cost savings and service transformations, there are also risks. The adoption of AI carries profound ethical, legal, operational, and human implications.

Councils should avoid falling into the trap of procuring AI tools in the hope that they alone will solve all of the major challenges faced by the public sector. There's a real rush to snap up the latest tools, but without taking a step back to think about strategy, priorities, or whether the groundwork has been laid.

There is surely a better way!

Just like with most major transformations, there are evidence-based implementation principles that put people, purpose and learning at the heart of AI transformation. Unfortunately, my experience and research suggest that many organisations are skipping these steps.

Common pitfalls in AI adoption

Many local authorities find themselves stalled or struggling to reap the promised benefits of AI. The challenges often revolve around resources, readiness, and power dynamics:

1. A recent Local Government Association (LGA) survey highlighted that the primary hurdles are rooted in resources. A lack of funding, staff capacity, and staff capabilities consistently emerge as the top three barriers to wider AI deployment. With budgets under pressure, LAs find it difficult to justify significant upfront investment, and existing workforces often lack the specialist skills needed to manage and maintain complex AI systems.

   
   

2. AI is only as good as the data it is trained on, and councils often face legacy data silos, inconsistent formats, and historical data sets that are inherently biased. Rushing to implement AI without first cleaning and structuring this data can lead to inaccurate, unfair, and potentially discriminatory outcomes in critical service areas.

   
   

3. Similarly, LAs are buying different products but are not always taking their workforce on the journey with them. Successful adoption depends on engaging staff, understanding their needs, and co-designing solutions.

   
   

4. On top of everything else, there’s a power imbalance when it comes to procuring new tech. Many councils simply don’t have the expertise in-house to challenge the big claims made by tech companies. This often means buying ‘off-the-shelf’ solutions that aren’t tailored to what the council or local community needs. Without the right expertise, councils can end up locked into costly contracts and rolling out technology that doesn’t meet local requirements.

This situation can be avoided. Councils need to take a considered approach and follow the evidence. The rest of this article outlines four principles designed to help authorities develop a meaningful strategy for AI adoption and embed the tools to make it a reality.

Ensure you've got the right enabling environment

This is a theme we talk about a lot at Mutual Ventures. Simply put, it means putting in place the foundations that set you up for success. Before starting to think about pilots or procurement, assess your organisation’s readiness for AI.

It is more than just checking your data infrastructure (though this is important). It’s about understanding your digital maturity, workforce skills, and organisational culture. Are your teams digitally literate and confident? Do you have clear data governance policies? Are staff and stakeholders engaged and informed?

If the answer is no to these questions, then some groundwork needs to be done before an AI tool is implemented so that you set it up to succeed and your organisation can get the most out of it.

For many, this will mean improving the way that data is collected and managed, as well as ensuring the workforce is ready to use AI in a safe way.

Plan with purpose

This may seem obvious, but many councils are buying and implementing tools and then thinking about effectiveness. In an increasingly crowded market, this isn't a sensible way to procure. Councils don’t usually just go to the market and buy a mix of services to see which one works, and they shouldn’t do this for AI either.

Instead, take a strategic approach and diagnose your problem first. Review where things are going wrong or could be improved and then consider a range of potential tools that address this.

Focus on people and process

It's easy to get distracted by shiny new technology, but it's important to keep the focus on two key areas: people and processes.

This requires a human-centric and structured change management approach. At Mutual Ventures, we believe that AI should enhance human capabilities, not replace them. There is a need to ensure that the processes AI touches are clearly defined and aligned with broader goals. Remember to share any plans with staff early on to bring them on the journey and build confidence. Provide training and support so that tools can be adopted responsibly. And finally, ensure there are practical governance structures which include oversight to ensure tools are managed appropriately. There’s little benefit in paying lots of money for something new if it isn’t actively used, or worse, people are using it incorrectly.

Shockingly, I’ve already witnessed local authority staff entering sensitive, person-level data into public AI chatboxes - a mistake which is easy to make, but represents a serious data breach and violation of GDPR. Local government needs to get on top of issues like this before they negatively impact public trust.

Evaluate, learn and refine

Organisations should take a test and learn approach. Create measures to assess whether the product you're using is achieving your desired goals. As with all good evaluations, this should use both qualitative and quantitative measures. If the product isn't doing what you want, go back to step two!

There's a need for continuous work to upskill your workforce to make the most of any new products. As tools get upgrades and new versions, training for the workforce also needs to be updated to further learning.

By embracing a culture of ongoing reflection and adaptation, LAs can ensure that AI continues to deliver meaningful impact and supports the evolving needs of their communities.

Success from the use of AI in local government isn't going to come from the technology alone. Organisational readiness and strategy are key from the start. The steps outlined in this article provide a starting point to ensure AI is implemented in a smart way and acts as a partner in service delivery, instead of being a risk. 

Ready to get smarter with the way you implement AI products? If the answer to this question is yes and you'd like to discuss how we at Mutual Ventures can help with those ambitions, get in touch with Ross at ross.murray@mutualventures.co.uk