Is your municipality data-ready for a Generative AI copilot?
The use of machine learning (ML) and artificial intelligence (AI) in local government isn’t new, despite the recent surge in attention due to generative AI’s public availability. Traditionally, ML/AI has been the domain of specialized data analytics teams, requiring specific skills and tools for narrow applications. In contrast, generative AI (GenAI) is easily accessible, user-friendly, can be applied to a broad range of questions, and produces human-like outputs just like a professional artist, analyst, or copywriter. Now integrated into fundamental business productivity platforms like Microsoft Office (via CoPilot) and Google Workspace (via Duet), GenAI poses a question for municipalities: should they broadly adopt it?
Before jumping on the bandwagon to demonstrate innovation, it’s crucial to evaluate whether these tools align with your municipality’s current capabilities and needs. Is there a problem only GenAI can uniquely solve? If so, what data and file repositories are you willing to expose it to? This brings data quality and governance to the forefront of the AI readiness conversation. Does your government have a program to properly evaluate the quality, completeness, and sensitivity of its data for the problem GenAI might tackle?
The challenge lies in the vast amounts of semi-structured (e.g., spreadsheets) and unstructured (e.g., PDFs, images, videos) data stored in platforms like SharePoint or Google Workspace. Unlike administrative databases for permitting, taxation, or HR operations, these files are rarely classified as public vs. non-public or high vs. low sensitivity. Without strict regulatory frameworks like HIPAA, these repositories — which vendors tout as ripe for business intelligence — often remain entirely unclassified. Without consistent and reliable classification, appropriate use of the files is very difficult to determine, especially at an enterprise scale.
While GenAI offers the potential for analyzing large volumes of documents and answering questions, the accuracy and appropriateness of its outputs aren’t guaranteed. AI-generated insights require validation of input data’s accuracy, completeness, currency, and sensitivity. There’s also a risk of GenAI producing responses containing regulatory and privacy-protected content.
Data professionals understand the pitfalls of poor data quality, but business leaders and elected officials often underestimate the importance of data hygiene and governance. This is largely because these topics aren’t exactly scintillating. Hold a public meeting on affordable housing and you’ll get a crowd, hold one on data governance and that box of donuts you brought to the meeting will go uneaten. I’m arguing that the first step in AI readiness is to assess the extent of your municipality’s data governance even before identifying problems GenAI can solve. After all, your data will always be at the heart of any AI solution. With GenAI plugged into an enterprise business suite, all records are potentially fair game for utilization and this is very different from just accessing enterprise databases, which tend to have established privacy and quality controls in place.
