Data First: The Critical Foundation for Your GenAI Launchpad

Building your GenAI Launchpad

According to NASCIO, the leading advocate for technology policy at all levels of government, only 27 percent of state CIOs expressed that they have mature governance over enterprise information management - a very foundational element for GenAI implementation. GenAI only thrives when the information it's reasoning over is thriving.  That information has to be packaged and fed to GenAI in formats that it can easily understand; reiterating that structured, digitized and accessible data being able to be used for analyzing, sharing and informing decision-making is key.

A strong data foundation makes the ultimate impact on fueling innovation. This strong data foundation is the fundament for developing a powerful and dependable GenAI model. Learn about the four keys to establishing a solid data foundation below:

1. Digitize and Standardize Data:
   1. Digitize Paper Records: Convert paper records into digital formats using optical character recognition (OCR) to automatically generate meaningful metadata.
   2. Standardize Digital Data: Digitized records and existing digital data should be ingested into a centralized repository. Standardization requires organizing data in a uniform manner, such as using a common file format, naming conventions, and data structure.
2. Implement Data Quality Control:
   1. Cleanse and Validate Data: Identify and rectify errors, inconsistencies, and inaccuracies in your data. Proper digitization will provide this automatically, with a human in the loop to ensure accuracy.
   2. Enforce Data Entry Standards: Establish guidelines for entering new data, including validation rules, review processes, and day-forward paper digitization to prevent errors at the source.
3. Create Metadata and Taxonomies:
   1. Generate Metadata: Go beyond record-level metadata to develop metadata for each dataset, including information about data origin, creation date, and any relevant contextual details. This metadata is crucial for understanding the data's meaning and lineage.
   2. Build Taxonomies: Create structured classifications or taxonomies that define relationships and categorizations within your data. This helps organize information and facilitates more efficient searching and analysis.
4. Implement Data Governance:
   1. Define Data Ownership and Responsibility: Clearly assign ownership and responsibility for different datasets within your organization. This ensures accountability and helps maintain data quality.
   2. Establish Data Governance Policies: Develop and enforce policies regarding data access, security, and privacy. This includes defining who has permission to modify or delete data and setting guidelines for data sharing.
      

Specific AI roadmap needs of each state will vary depending on their individual priorities and strategic plans, including the four important considerations above. Making sure each point is considered and thought out is crucial in ensuring a solid data foundation that seamlessly integrates AI into state IT initiatives. As we witness more states start to implement GenAI strategies, we learn improved best practices in governing the technology. States include Kansas, Washington and New Jersey have already adapted and published state GenAI policies, which act as leading examples for the states that will soon follow. 

The future of GenAI’s utilization in government holds enormous promise, although state and local leaders must prioritize the foundational data elements to make GenAI a safe and strategic asset.