NACo AI Report Analysis

Artificial intelligence (AI) and its subset, generative AI (GenAI), are transforming how local governments operate. The National Association of Counties (NACo) has recognized this potential and developed the AI County Compass Toolkit, a comprehensive guide for local governance and the implementation of AI technologies (National Association of Counties, 2024).

What is the AI County Compass Toolkit?

The AI County Compass Toolkit is a detailed guide aimed at helping county officials understand and implement AI technologies responsibly. Developed by the NACo Artificial Intelligence Exploratory Committee, the toolkit provides a wealth of resources, from policy recommendations to ethical guidelines, tailored specifically for county governments.

The toolkit is divided into several sections:

• Executive Summary: An overview of the toolkit’s purpose and key findings.
• Guiding Principles: Seven principles to guide counties in their AI journey.
• Workgroup Recommendations: Detailed advice from four key workgroups.
• Roadmap and Next Steps: Practical steps for implementing GenAI solutions.

Key Findings and Metrics

The report provides valuable insights into the current state of AI in county governance. Here are some key findings and metrics:

• Widespread Use of AI: Over 75% of county officials and staff reported using GenAI tools both at work and in their personal lives. This high adoption rate underscores the relevance and growing dependence on AI technologies for improving efficiency and productivity in daily tasks (National Association of Counties, 2024).

• Top Tools: ChatGPT emerged as the most used GenAI tool among county officials, followed by Bing Chat, Grammarly AI, and Google Bard. The popularity of these tools highlights their effectiveness in providing solutions and aiding decision-making processes in county operations (National Association of Counties, 2024).
• Areas of Implementation: Information Technology (IT) was identified as the most likely area for GenAI use, followed by Human Resources, Finance, and Emergency Management. This shows that GenAI's impact is broad, affecting various critical functions within county operations and promising to streamline and enhance these areas significantly (National Association of Counties, 2024).

These findings highlight the growing acceptance and integration of AI in county operations, emphasizing the need for a structured approach to its implementation.

Recommendations

The toolkit offers a comprehensive set of recommendations divided among four key workgroups:

1. Policy Working Group: Promoting Policy Models

• Establish a Policy Framework: Developing a pyramid approach to implement GenAI policies is crucial. This method ensures that policies are layered and can address various levels of AI implementation, aligning with the National Institute of Standards and Technology (NIST) guidelines. A well-structured policy framework helps counties navigate the complexities of AI adoption, ensuring all aspects from security to ethical considerations are covered.
• Review Legal Considerations: Staying updated on federal and state legislative frameworks concerning AI is essential. As AI technology evolves, so do the regulations governing its use. Counties must be proactive in understanding these laws to avoid legal pitfalls and ensure compliance with the latest standards.
• Assess Procurement Policies: Robust procurement practices are vital to protect county interests when partnering with third-party AI service providers. Proper procurement policies ensure that AI tools and services acquired are secure, reliable, and meet the county’s specific needs without compromising data integrity or security.

2. Ethics Working Group: Establishing an Ethical Framework

• Fairness and Impartiality: Ensuring AI tools are fair and impartial is fundamental. AI models must be designed to minimize biases and provide equitable access to all community members, regardless of socio-economic background. This helps build trust and promotes fairness in AI-driven decisions.
• Transparency: Implementing transparency in AI use is critical. Counties should ensure that AI processes are explainable and that there are clear disclosures about when and how AI is used. This transparency helps in gaining public trust and ensures accountability.
• Privacy: Adhering to local, state, and federal privacy guidelines is non-negotiable. Protecting personal data and ensuring that AI applications comply with privacy laws prevent misuse and data breaches, safeguarding citizen information.
• Accountability: Maintaining accountability through impact assessments, risk remediation, and audit trails ensures that AI applications are used responsibly. Regular audits and assessments help identify potential risks and mitigate them promptly, maintaining the integrity of AI systems.

3. Applications Working Group: Enabling Responsible Applications

• Evaluate Use Cases: Regularly assessing potential AI use cases and their sensitivity is crucial. Counties should identify which AI applications provide the most benefit while considering the risks involved, ensuring that high-risk applications are handled with additional scrutiny.
• Familiarize with Federal Resources: Utilizing resources from NIST and other federal bodies to guide AI implementation provides counties with a solid foundation of best practices and standards. This ensures that AI systems are secure, efficient, and aligned with national guidelines.
• Data Governance: Practicing robust data governance ensures the quality and protection of data. Counties must implement policies that manage data collection, storage, and usage effectively to prevent misuse and maintain data integrity.
• Update Cybersecurity Measures: Strengthening cybersecurity to protect against breaches related to AI use is essential. As AI applications handle sensitive data, robust cybersecurity measures ensure this data is protected from unauthorized access and cyber threats.

4. Workforce Working Group: Preparing the Workforce

• Skills Development and Training: Focusing on training staff in AI technologies and providing ongoing learning opportunities is crucial. Continuous education ensures that county employees are well-equipped to handle AI tools effectively and adapt to new technologies as they emerge.
• Consider Skills Acquisition Options: Exploring various options for acquiring AI skills, including hiring specialized staff or collaborating with educational institutions, ensures that counties have access to the necessary expertise to implement AI successfully.
• Develop a Multi-Year Strategy: Creating a long-term strategy for workforce development addresses recruitment, retention, and career advancement. A strategic approach ensures that the workforce is prepared for the future of AI and can adapt to its evolving landscape.
• Engage with Unions and Workforce: Communicating the benefits and challenges of AI with organized labor groups and the broader workforce fosters collaboration and understanding. This engagement helps in addressing concerns and ensuring that AI implementation is smooth and beneficial for all parties involved.

Common Recommendations Require Solutions

This report, along with others, continues to identify common themes for public sector project planning, choosing an approach to AI, the need to train the workforce, and to create policies and procedures around confidentiality and data security. What all of these reports lack, so far, is to identify platforms that are capable of being introduced and scaled to meet the needs of the government workforce. That is not their fault though because there really isn't a good platform or product out there right now that is ideal to bring into the workplace right now.

I want to be clear, I am not talking about customer-facing AI solutions that help citizens access government information or resources. I think there are plenty of solutions out there now that show great potential. Polimorphic, frequently shown by Parth Shah is a great example of that type of product. I highly encourage you to check them out, and they are frequently highlighted in posts by Micah Gaudet .

What I am talking about here is exactly what you saw earlier in the graphic showing the products used for internal productivity in the GenAI usage chart.

• ChatGPT: Public sector workers are using the fire out of ChatGPT, but it is a very clunky platform to use for small to medium-sized local governments. They typically don't have large IT budgets to begin with, and an enterprise plan is not within the realm of reality for them. A teams plan can be set up, but OpenAI has not made this very user friendly so it will be an exercise in frustration that may cause many local government staff to throw in the towel up front. If you work with technology all the time this concept may sound foreign to you, but in my world, I see local government staff who struggle with the basics of technology and electronic files so I know ChatGPT teams would be a nightmare for them to navigate.
• Bing: Can I say enough about how horrible Bing is for anything? While it is powered by Chatgpt 4, it comes wrapped in an extra layer of Microsoft goodness that doesn't feel as good as real Chatgpt from OpenAI.
• Grammarly: I don't consider Grammarly as a Generative AI in the same way as I do ChatGPT, Perplexity, Claude, or Google Gemini. I view it more as a tool with AI. If Grammarly is going to be considered an AI, then why isn't Canva also an AI since it has an AI component to it? I've seen some policy guidance going around in recent months where government and policy officials seem to be confused as to what is or isn't AI and how it should be regulated and tracked for public record requirements.
• Google Bard (now Gemini): They are way behind, and in their race to catch up they are making mistakes. I won't be trusting Gemini anytime soon with advice like what is seen in the image below.

• Proprietary Solutions: I don't see many good proprietary solutions out there at the moment. The ones I do find are simply reselling you ChatGPT, Claude, Perplexity, etc. with their user interface on top of it, and not telling you about hidden API fees on the back end. The sheer amount of paperwork that government administrative staff could scan, upload, and process would be a budget destroyer if you tack on API fees to do anything. If anyone has a good solution, be sure to leave it in the comments.

So why is my summary of these solutions even important in the first place? It's because you cannot really train your employees in a coordinated way when it is impossible or extremely cumbersome and time-consuming to set up a collaborative environment that allows them to work and learn together. If you cannot train your employees, you don't adopt AI in the first place so small to medium-sized local governments have a real problem here until someone provides a solution.

Conclusion

The NACo AI County Compass Toolkit is an essential resource for county officials looking to harness the power of AI responsibly. By following the toolkit’s recommendations, counties can navigate the complexities of AI implementation, ensuring that it benefits both the workforce and the communities they serve.

The future of AI in local governance is bright, with endless possibilities for improving service delivery, enhancing productivity, and creating innovative solutions to local challenges. As counties embrace AI, it is crucial to prioritize ethical considerations, data governance, and workforce development to achieve sustainable and equitable outcomes.

If you haven't read my previous article on AI, please be sure to click the link!

References

National Association of Counties. (2024). AI County Compass: A Comprehensive Toolkit for Local Governance and Implementation of Artificial Intelligence. Retrieved from https://www.naco.org/resource/ai-county-compass-comprehensive-toolkit-local-governance-and-implementation-artificial.

Tom's Hardware. (2024, May 25). 17 cringe-worthy Google AI answers demonstrate the problem with training on the entire web. Retrieved from https://www.tomshardware.com/tech-industry/artificial-intelligence/cringe-worth-google-ai-overviews