There’s no denying that AI (artificial intelligence) is the buzzword du jour. Have you ever wondered how to classify records generated by AI on your retention schedule? Whether or not your government entity is actively using AI, it’s a question worth considering.
What is AI and Why Would My Government Entity Use it?
According to the University of Illinois Chicago, “Artificial intelligence represents a branch of computer science that aims to create machines capable of performing tasks that typically require human intelligence. These tasks include learning from experience (machine learning), understanding natural language, recognizing patterns, solving problems, and making decisions.” So, AI is a broad term that includes many types of computer programs. What the programs have in common is their ability to take in data points and output something useful that would normally require human reasoning. This output can be a report, an image, a map, text, and much more.
AI can be used in many ways to enhance government services. For example, because computers can process large amounts of data quickly, AI can be used to speed up processes and workflows. AI can also be used to look for patterns in data to help government employees make informed decisions.
Should My Government Entity Use AI and If So, How?
Weighing all the costs and benefits of AI is outside the scope of this article. However, it’s certainly worth pointing out that government entities should consider how to mitigate potential risks before using an AI program for government business. Things to watch out for include the potential for algorithmic bias, lack of transparency, data privacy or security risks, and over-confidence in AI tools and their outputs. Before implementing an AI solution, consider if your government has the appropriate expertise to use the tool wisely.
The Texas AI Advisory Council is currently studying uses of AI in state government. The Council will review each state agency’s use of AI systems and make a report to the Legislature with policy recommendations. So, stay tuned to the news for those exciting updates!
How to Classify AI-Generated Records on Your Retention Schedule
Generally speaking, AI-generated records should be classified the same as human-generated records. Now, this may seem like a bait-and-switch, but think about it—is a record drafted by an AI system really that different from one you created without AI’s help?
AI is an assistive technology and human staff are ultimately responsible for the records that are created on behalf of the government entity they work for. The final product of a record should be classified based on the administrative purpose and function the record serves for the government entity. So, records created with AI assistance should be classified the same on your retention schedule and kept for the same length of time as those created without AI assistance.
Hypothetical Examples
Let’s look at some hypothetical examples of records that could be generated by AI tools and how they would be classified. Citations are provided for the Local Government Retention Schedules, then the Texas State Records Retention Schedule (for state agencies and state universities).
Customer Service Chat Bot
Scenario | Classification |
---|---|
Members of the public who visit a city’s website have the option to instant message with an AI-powered chat bot. The chat bot responds to basic prompts and directs users to relevant resources on the website. For example, if a person asked about how to report a city code violation, the chat bot would send them the link to the code enforcement page and give a quick explanation of the reporting process. | General Correspondence (GR1000-26b, 1.1.008) |
Press Release Draft
Scenario | Classification |
---|---|
A government employee uses a generative AI tool to draft a press release for an upcoming event. The employee types in a prompt with the basic information they want to include. The AI tool responds with a draft of a press release, which the employee revises before publishing. | Public Relations Records (GR1000-33, 1.1.019) |
Staff Meeting Notes Transcription
Scenario | Classification |
---|---|
Staff meetings are held online through video conferencing software. An AI-powered note taking tool is used during the meetings. After the meeting, the AI tool provides detailed notes of what was discussed at the meeting. A staff member reviews and edits the notes. | Staff Meeting Minutes (GR1000-32, 1.1.063) |
Report Based on Data Analysis
Scenario | Classification |
---|---|
A government staff member uses an AI tool to find patterns in large sets of data. After reviewing the patterns identified in the data and comparing them to other relevant information, the staff member writes a report of the findings. | Reports and Studies, Non-Fiscal (GR1000-41, 1.1.067) |
Further Reading
- U.S. Federal Government AI Use Cases: https://ai.gov/ai-use-cases/
- National Archives and Records Administration (NARA) AI Inventory: https://www.archives.gov/data/ai-inventory