The Public Accounts Committee Urges Government to Address Legacy IT Issues to Succeed with AI
The Public Accounts Committee (PAC) has emphasized the importance of fixing legacy IT issues in order for the government to effectively implement artificial intelligence (AI) initiatives within the public sector.
In their report on the Use of AI in government, the PAC highlighted the critical role of high-quality data in AI applications. However, the Department for Science, Innovation and Technology (DSIT) revealed that government data is often of poor quality and stored in outdated legacy IT systems.
Approximately 28% of the government’s IT systems are classified as legacy systems, encompassing both software and hardware. According to DSIT, these systems have reached end-of-life, lack supplier support, and are unable to be updated. The government’s Digital and Data Roadmap 2022-25 aims to address these issues, with plans to remediate all 72 highest-risk legacy systems. Despite progress, 21 systems still lack funding for remediation.
The PAC stressed the urgency of resolving legacy IT issues not only for AI implementation but also to mitigate cybersecurity vulnerabilities. DSIT plans to prioritize systems with valuable data and high security risks.
Regarding legacy hardware, DSIT views cloud services as a solution to move away from maintaining large data centers. However, the PAC highlighted the need to balance concentration for cost-effectiveness with diversification for resilience and security.
The PAC also raised concerns about data sharing barriers in government. Efforts are underway to implement standards, frameworks, and tools to facilitate data sharing, including the establishment of a national data library.
Addressing the digital skills gap, the PAC recommended establishing mechanisms for evaluating AI pilots and supporting the development of common and scalable AI products.
In terms of AI procurement, the report urged DSIT and the Cabinet Office to ensure fair opportunities for all suppliers in a market dominated by a few global technology companies.
