Ordnance Survey CEO: Location is the obvious way to connect data in AI era

Bolton, whose background is in computer vision, noted that OS has used AI for more than a decade, beginning with automatic feature extraction from aerial images. The agency scans the whole of Britain every three years, producing high-resolution imagery that Bolton said is ideal for training neural networks. While acknowledging the recent buzz around large language models and transformer networks, he insisted AI has existed far longer and that OS sees it primarily as a tool for automation rather than human replacement.

The organization is now exploring agentic AI and natural language interfaces to offer its vast data repository to public and private partners. Bolton emphasized that trust remains paramount, pointing to OS's AI Charter as a framework for secure and responsible usage. He stressed the need for human checks on AI-generated decisions, ensuring every insight can be traced back to concrete data. "Making sure we go slowly, preserving that this decision or that insight can be traced back to this set of data means therefore I can have a high confidence in it," he said.

OS data is already used across government agencies, utility networks, and construction firms. Bolton described location as a "general purpose technology" similar to semiconductors or pneumatic tyres, with applications too diverse for any single organization to own the entire stack. He said OS's mission is to produce the best data and collaborate with others to embed it into various solutions. "Everyone has got 'where?' questions," he added, highlighting the universal relevance of geography and location insights.

The CEO acknowledged that AI adoption is an organic process, with the institution taking a cautious approach to preserve its authoritative and trustworthy reputation. He said secure access and data attribution are critical as OS navigates this transformation.