States Have Big Plans for AI
A majority of states have established policy guardrails for adopting generative AI, and many CIOs are excited to scale up proofs of concept into production implementations, according to a recent report by the National Association of State Chief Information Officers and McKinsey.
Even as state officials are taking clear steps to adopt AI, progress is slow. At-scale use requires working through limited data quality, a lack of allocated budgets and privacy, security, and ethical concerns. CIOs also face talent challenges, including upskilling current staff and recruiting for new roles — such as ethics and privacy law experts — that aren’t usually within their purview.
READ MORE: Local governments are developing AI use cases.
Citizens Want Convenience When Interacting with Government
Many public sector agencies have already deployed AI in the form of digital assistants and chatbots, in part because they deliver a big bang for the buck. Launching an assistant or chatbot that can respond to constituents 24 hours a day, seven days a week, immediately improves customer experience. Only some people can quickly drop by government offices during business hours, but even the simplest chatbots can deliver around-the-clock responses to commonly asked questions or help people navigate complex websites.
In more sophisticated assistants, AI and machine learning refine responses and offer a more personalized experience. Other versions support conversations in multiple languages without relying on human translators. For an entirely spoken experience, agencies can connect their tools with voice assistants such as Amazon’s Alexa or Apple’s Siri.
These options can increase accessibility to agency information while freeing up human staff for more complex issues. Constituents get their questions answered more quickly and on a schedule that’s convenient for them.
LEARN MORE: State and local officials expect AI to benefit digital services.
Governments Must Mind the Details in AI Development
The phrase “garbage in, garbage out” takes on new importance with AI projects. High-quality and up-to-date training data is critical for developing AI systems for government use, and so are regular updates. After all, laws and regulations change constantly, and assistants and chatbots rely on what was last input. An inclusive development team with regularly scheduled reviews of training data can eliminate outdated, erroneous or potentially biased information.
Similarly, any system handling personal information should be designed to protect privacy. Agencies should minimize the amount of data collected and take special steps to disassociate information from identifiers that recognize an individual. Pairing these policies with technological best practices such as robust data encryption, multifactor authentication, and regular security audits and compliance checks can help preserve privacy even if data is compromised.
Despite challenges, government officials are forging ahead to take advantage of AI and customer experience enhancements, offering a promising starting point. By balancing innovation with ethical considerations and strong privacy protections, agencies can significantly improve accessibility, responsiveness and efficiency in their interactions and build a new foundation of trust with their constituents.
