The public sector increasingly recognises that legacy systems don’t just pose operational challenges – they’re critical barriers to effective AI adoption.
How to tackle the issue was the topic of a key panel session at this week’s Think Data for Government event in London (pictured). There, digital leaders highlighted the urgent need to modernise infrastructure and data foundations to unlock AI’s potential.
Chad Bond, director of strategy and innovation at Zaizi, framed the challenge as the “great legacy escape.”
“AI can’t thrive on creeping legacy systems operating on spreadsheets,” he said, pointing out the widespread use of inefficient data management practices across government departments.
The panellists unanimously agreed that successful AI implementation requires far more than purchasing cutting-edge technology. Richard Appiah, head of data strategy at Migrations and Borders Group Home Office, stressed the importance of getting “the basics right” by focusing on three fundamental elements: data, people, and culture.
Data maturity as a critical starting point
Elsewhere, Sian Thomas MBE, chief data officer at the Department for Business and Trade, underscored the importance of infrastructure improvements, noting that organisations must carefully justify funding for both innovative technologies and underlying system upgrades.
“If the legacy is so old that it costs more to get off it than the value it provides, you’re in a challenging position,” she said.
Meanwhile, Dr Ravinder Singh Zandu, head of digital and systems team, Cabinet Office, noted that that AI is not a magic solution, but a tool that requires careful preparation. “AI will solve problems only if the machine model has learned,” he said, emphasising the need for a strategic, measured approach to technology adoption.
The panellists highlighted several key strategies for effective legacy modernisation:
If you liked this content…
- Start small and demonstrate value
Appiah recommended spinning up small proof-of-concept projects to demonstrate tangible results to senior stakeholders. “They just want to see results,” he said, suggesting a gradual progression approach.
- Develop data literacy
The Home Office’s approach included hosting “data awareness” events to illustrate the consequences of poor data management. By creating simulated scenarios, it raised understanding of data’s critical importance across different organisational levels.
- Establish clear accountability
Dr. Singh Zandu stressed the importance of creating clear accountability mechanisms and developing a comprehensive data asset register.
“We need to identify new rules and make people accountable for specific steps,” he said.
Ethical considerations and human oversight
An often-overlooked aspect of AI strategy is ethical implementation. Thomas highlighted the crucial role of governance frameworks, stating, “Just because you can do something doesn’t necessarily mean you should.”
The panel agreed that human insight remains paramount. According to Thomas, the goal is not to create “smart tech” but “smarter people and businesses.” This means developing workforce capabilities to work alongside AI, not be replaced by it.
Practical recommendations
For public sector organisations looking to modernise and prepare for AI adoption, the panel offered several key recommendations:
- Conduct a comprehensive capability assessment
- Develop a clear target operating model
- Create robust data governance frameworks
- Invest in data literacy across all organisational levels
- Start with small, manageable AI pilot projects
- Maintain a strong focus on ethics and human oversight
