Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Artificial Intelligence topic
No spam. Unsubscribe anytime.
Highland Village staff outlines internal AI sandbox, policy and training plans
Summary
City staff presented the new AI policy, described current purchases (26 ChatGPT seats, BlueVoice for police, two Otter.ai licenses) and demonstrated an internal document hub that searches only city ordinances and budgets; staff emphasized training, citing one test where the hub answered five of six queries correctly but also produced one authoritative error.
Sunny, a city staff member leading the information-technology presentation, told the Highland Village City Council on June 23 that much of the software the city uses already includes AI features and outlined a controlled rollout of generative AI tools. She said the city has purchased 26 ChatGPT licenses at an annual cost of $9,600, the police department uses BlueVoice with 44 licenses ($6,500/yr), and the city holds two Otter.ai licenses ($385 total). Sunny said an internal group of "AI ambassadors" from multiple departments has been testing use cases such as drafting memos, minutes transcription, report-writing and ordinance lookups.
The presentation described a document-specific "hub" that limits AI searches to documents the city supplies — ordinances, three years of budget books and other city records — so that responses are drawn only from local sources rather than the open web. Sunny demonstrated a pilot where the hub answered five of six routine questions correctly; she said the single error occurred because the hub had not been fed the marketplace sign regulations and cautioned that AI "thinks it knows everything whenever it answers." She said that verifying source links and using the "cite sources" function will be taught in mandatory training.
Staff also told council they have prepared an AI policy that rolled out the week of the meeting and will introduce training modules in July as part of the state-mandated cybersecurity course. Sunny said future plans include adding department-specific hubs (for example, HR-only documents) and eventually exploring controlled access to outside legal sources if they can securely tokenize or constrain external lookups. She emphasized keeping public-facing AI tools internal until accuracy and verification controls are robust enough to avoid public misinformation.
Council members asked about open-records implications and about preventing repeated AI errors. Sunny said both prompts and AI-generated outputs may be subject to public-records requests and that training will instruct staff to request source citations on every response. The city manager and staff said they will continue phased testing and community transparency before any external deployment.
