Is AI Ready to Replace Human Policy Advisers?

A 50-state investigation in data journalism suggests the answer is, not yet. The AI agent was insightful on a number of fronts; but, while not descending into hallucinations, its mind strayed from instructions as the experiment went on. Government Technology Data Reporter Nikki Davidson tasked Google's AI tool Gemini (formerly Bard) to explore AI's perspective on government technology use. Davidson's innovative approach involved treating AI as a collaborative partner to generate insights on AI's potential applications in government. Despite Gemini's occasional inaccuracies and deviations from instructions, Davidson's project yielded diverse and unexpected use cases across different areas, such as mental health, opioid use and climate change.Gemini's recommendations extended to climate challenges and infrastructure needs, reflecting a surprisingly deep understanding of regional concerns. Gemini itself suggested it would be five to 10 years before AI is fully integrated in government operations, emphasizing that it is inevitable for government. Believing turnabout is fair play, Davidson asked Gemini for feedback on her work. It gave the article a strong numerical grade on a scale of 1 to 10 but did have suggestions on how to make it better. Human reviewers, including Benjamin Palacio, a senior IT analyst with Placer County, Calif., highlighted both the promise and challenges of AI applications, particularly in sensitive areas like mental health support. Davidson views AI as a valuable tool but underscores the necessity of human oversight and awareness of its limitations. SHOW NOTES Takeaways: AI can be used as a tool to explore the best uses of technology in government. Surprising use cases of AI in government include mental health and opioid abuse awareness. Ethical concerns arise when AI suggests analyzing sensitive data like social media and medical records. Human intervention is necessary to ensure the accuracy of AI. Chapters: 00:00 Introduction and Setting Up the Story 00:39 Using AI to Explore Government Use Cases 04:10 The Maps and the Impact of the Story 05:12 Surprising Use Cases of AI in Government 07:18 Ethical Concerns and Gray Areas 09:05 AI's Role in Infrastructure and Climate 10:26 Realistic Timelines for AI Implementation 11:22 The Relationship With AI and Hallucinations 13:17 AI's Opinion on the Worth and Accuracy of the Work 14:19 Responses from Human Reviewers 15:30 AI as a Tool With Limitations 16:23 Conclusion and Future Possibilities Related Links to items referenced in the episode: How Does AI Predict Governments Will Use AI? It Depends Where You Live Center for Public Sector AI, a division of e.Republic, the parent company of Government Technology and Governing. Our editors used ChatGPT 4.0 to summarize the episode in bullet form to help create the show notes. The main image for this story was created using DALL-E 3. Listen to this episode on the player below or subscribe for free on YouTube or the podcast app of your choice — Apple Podcasts, Spotify, Audacy and Audible.[Placeholder for audio player embed]