By Emma Thwaites 

AI is a rapidly growing technology that’s here to stay. Recent events (models acting autonomously and in dangerous ways, tech bosses and others warning about the risks of superintelligence) have underlined the need for caution and careful consideration in adoption. For public bodies, this will be essential for maintaining public trust. As one speaker at the recent DigiGov conference reminded us, this is a journey, not a race. Digital literacy and data quality should be priorities, ensuring people, skills, and processes develop alongside the technology that is changing our everyday lives. 

This year I had the pleasure of representing Allegory’s client, the Open Data Institute (ODI), on the GovTech stage at DigiGov Expo 2026. Sharing a stage with a wide variety of experts means you get the full benefit of their knowledge, and this year’s panels and talks did not disappoint. From asking how the government assures the AI it’s already using, to whether the productivity promised from AI is actually happening, we posed a series of challenging questions and received some thought-provoking responses. 

Which problems to focus on

The day began with an overview from Chris Pannell of Salesforce, who asked whether we are focusing on the wrong problems. He gave examples from across the NHS, the Crown Courts, and prisons, and suggested that while technology use is increasing, backlogs and delays are too. He described humans as the ‘middleware’, bridging the gap between systems, teams, and processes. Chris highlighted NHS SBS as an example of an organisation that is handling the change in the right way. The company worked with Salesforce to redesign how enquiries flow between teams rather than simply seeking quick fixes. By using context and data to address root issues over time, they transformed the entire hand-off process. As a result, they cut the average time to raise a query from 12 to 3 minutes, and reduced average handling time by 20%. Chris posited that AI transformation should be approached in the same way as any transformation within an organisation; in the early days, “we all just got excited and forgot the rules!” 

This set the themes for the day, as speakers touched on how AI implementations require consideration of skills, people, and processes. 

AI assurance

Panellists in the AI assurance session stressed the importance of scaling literacy across entire organisations. As Paul Neville from The Pensions Regulator said, just as you would never want a Board that doesn’t understand finance, you shouldn’t have one that doesn’t understand AI. We must educate ourselves and our colleagues on reliable, productive ways to use AI at every level of the organisational hierarchy, he added. Alec Thomas from the Defence AI Centre (DAIC) emphasised bridging the gap between organisational leadership and AI professionals by encouraging a mutual understanding of risk and opportunity. All panellists encouraged a more individual approach to building skills and awareness by asking what each person needs, as well as what the organisation requires as a whole. 

Transparency

Alongside AI literacy, public trust is essential, and people only trust what they can understand. While emerging technologies are notoriously difficult to explain, organisations must commit to genuine transparency with their staff and end users. That applies internationally, in how global AI regulation develops, and locally, in how individual teams take a risk-based approach to adoption. As Alec Thomas noted, no one has agreed on a universal blueprint for what ‘good’ looks like when change happens this fast. Instead, the focus must be on building international standards, policies, and governance collaboratively so that citizens retain control over where their data lives and how it is used. Pointing to the DAIC as a practical example, Alec explained that they published their internal AI policy for public viewing. Ultimately, honesty is the best policy when explaining how organisations use AI, and, crucially, how they handle data.

Data quality

Data quality was another major theme throughout the data, a topic particularly close to our hearts at Allegory given our work with the ODI. If poor data goes into AI models, flawed outcomes inevitably follow. The ODI’s Resham Kotecha highlighted how easily bias creeps into datasets and scales through AI systems, warning of real-world risks like LLMs delivering faulty legal counsel or health recommendations that ignore key demographic factors. Catherine Nikolovski from the GovTech Global Alliance and Civic Software Foundation argued that we need to understand what AI amplifies when the underlying data is flawed.

What comes next?

No one can predict how technology will look 70 years from now, and anyone trying in 1956 would probably have missed the web entirely! Something that came through loud and clear on the GovTech stage was the importance of admitting humility when considering what the future may hold: there are many new skills to learn and much new knowledge to acquire. However, we can decide now how we will use these technologies by building AI and data literacy across organisations, keeping policy transparent and honest, and carefully considering data quality to maintain public trust and mitigate harms, as we adopt new and exciting technologies in the decades to come. 

Get in touch to find out more about our specialist strategic communications service for AI and data organisations by emailing team@allegoryagency.co.uk.