ADM-143 Use of Artificial Intelligence Policy
Last Content Update:
On December 19, 2023, the City of Lebanon adopted its first Artificial Intelligence (AI) policy, "ADM-143 Use of Artificial Intelligence." This policy reflects Lebanon's dedication to harnessing the power of AI while ensuring ethical, fair, and secure application. The policy is presented here in an open, accessible format and will be updated regularly by the Technology Review Committee as AI technology evolves. The official PDF version of this policy is available for download from our Open Records portal. We encourage you to submit your feedback about this policy and follow our AI Initiative page for more information and updates.
You can easily find this page again by visiting LebanonNH.gov/AIPolicy.
ADM-143 Use of Artificial Intelligence
Providing a framework and guidelines for the use of artificial intelligence (AI) in the City of Lebanon will help address the different risks associated with AI. There are a variety of benefits that result from using AI systems, but there must be guidelines in place for them to be used responsibly. The City of Lebanon is committed to safe and responsible uses of AI by its employees.
This policy applies to all departments and/or divisions of the City of Lebanon.
3.1 Algorithmic Discrimination occurs when automated systems contribute to unjustified different treatment or impacts disfavoring people based on their race, skin color, national or ethnic origin, cultural group, language, gender identity or expression, sexual orientation, mental or physical ability, age, religious or political opinion or activity, economic status, immigration status, or housing status.
3.2 Artificial Intelligence (AI) is a type of computer science which deals with computer systems that perform tasks which usually require human intelligence, such as reasoning, problem solving, perception, and language.
3.3 Bing AI is AI software that allows individuals to chat with an intelligent chatbot and generate various types of content. Bing AI is part of Microsoft’s search engine.
3.4 Chatbot is AI software that seeks to mimic human conversation through interactions via text or voice.
3.5 ChatGPT stands for Chat Generative Pre-trained Transformer. It is a complex machine learning model that is able to carry out natural language generation (NLG) tasks with such a high level of accuracy that the model can pass a Turing Test. It is based on the "GPT (Generative Pre-training Transformer) architecture, which is a type of neural network designed for natural language processing tasks.” ChatGPT is an AI chatbot that uses natural language processing to create humanlike conversational dialogue. The language model can respond to questions and compose various written content, including articles, social media posts, essays, code and emails. ChatGPT-3, ChatGPT-3.5, and ChatGPT-4 are the current models available to the public to use.
3.6 Generative AI uses both computer algorithms and large volumes of data to create new content, such as audio, code, images, and videos.
3.7 Sensitive Data or Personally Identifying Information (PII) (as defined in ADM-450 Securing Sensitive Information) is information that is private and must be protected.
3.8 Technology Review Committee (TRC) – This committee consists of the Cyber Services Director, Chief Innovation Officer, and Asset Manager. The committee works across all departments to review and audit any technology, including but not limited to software, technology, and AI.
This section is outlined in the subsections below.
City employees must be transparent and accountable when using AI technology. Following these guidelines are essential in ensuring the responsible and ethical use of AI systems within the city's operations. The following are important guidelines concerning transparency and accountability:
- Information Disclosure: City employees shall make efforts to disclose the use of AI systems in the workplace when and where appropriate. Employees must be open to sharing information, when appropriate, about the data being used, algorithms that are applied, and decision-making criteria. When disclosing the use of AI in the workplace, employees should utilize the guidance provided in this policy.
For example, if an employee uses ChatGPT or Bing to get ideas for a news article or gather helpful information, they would not need to disclose that they used AI to assist in their work, just like they would not disclose they used a spellchecker to check for grammatical errors.
However, a process that is fully automated without human intervention would need to provide appropriate disclosures. For example, if an AI was programmed to read an agenda and then summarize the contents to post on social media without any human oversight, the post would need to include that the text was AI-generated.
- Auditing: The City will regularly engage in monitoring, evaluating, and auditing its AI systems to ensure that the City is complying with ethical standards and legal standards. Reviews will be conducted by the Cyber Service Department in coordination with the Technology Review Committee.
- Responsible Decision-Making: City employees shall use AI as a tool to assist in the decision-making process, rather than relying completely on an automated output. However, there will be circumstances where decision-making will be completely automated, such as answers provided by a chatbot (see examples Section 4.1.a).
- Public Engagement and Feedback: City employees should make efforts to gather the questions, concerns, and other types of feedback from Lebanon residents regarding AI systems used by the City of Lebanon. By doing so, this can enhance fairness and inclusion in the responsible decision-making process.
In some cases, the AI systems that employees use have a direct impact on the security of the surrounding community. While there is significant value in AI technology, there are also potential risks involved, with some risks not being particularly obvious or understood. In order to secure and protect the digital rights of our community members, the following must be done:
- Data Protection: Protecting the personally identifying, confidential and financial information of the City of Lebanon’s citizens, employees, and other entities should be a priority for every City employee. The proper handling and storage of personal data is necessary, and there needs to be data protection policies and procedures in place to ensure there is no unauthorized access, use, or disclosure of sensitive information. City employees will handle sensitive data in accordance with applicable laws, policies, and procedures, as well as enforce strong safeguards to protect data from being accessed by unauthorized users.
- Education and Awareness: Management and supervisory staff are encouraged to help City employees who use AI by educating and raising awareness about the risks associated with using these types of systems.
Using AI should not in any way discriminate against any individual’s race, skin color, national or ethnic origin, cultural group, language, gender identity or expression, sexual orientation, mental or physical ability, age, religious or political opinion or activity, economic status, immigration status, or housing status. In accordance with City Code 185 Welcoming Lebanon, every action carried out by City of Lebanon employees is a reflection upon the City, and thus it is imperative to act respectfully and responsibly. For this to be achieved, the following must be done:
- Avoiding Bias: Any biases that are identified must be addressed and corrective actions should be taken. Bias can come in many forms, such as, biased training data, sampling bias, and stereotyping/societal biases. The City of Lebanon will continuously strive to act ethically when using AI technology and will make improvements as needed.
- Preventing Algorithmic Discrimination: AI systems should be used in an equitable way and should avoid algorithmic discrimination. Any instance of discrimination against a protected class, as delineated in City Code 185, shall be reported in accordance with ADM-126.1.
- Promoting Inclusion: City employees should use AI systems fairly and shall not discriminate against any individual. Including underrepresented and disadvantaged groups within datasets can help prevent biases and promote inclusion.
The appropriate amount of human oversight should be used when AI systems fail to meet their expected duties. Transparent, effective, and timely human intervention shall be applied when it is necessary to do so.
- Human Intervention: City employees should be able to override AI systems when appropriate and find corrective solutions and remedies to problems that arise. Human intervention shall be effective, accessible, fair, and shall not present an unreasonable burden on those intervening with the AI technology. The public deserves a back-up human intervention system where human review is initiated in case of an AI system failure or a threat caused by the system itself.
- Transparent Human Alternatives: Transparency should be held as a high priority by City employees, and this will help individuals understand the decision-making processes behind AI systems. Transparency allows humans to make reasonable judgements and take responsibility for the results.
- Effective and Timely Human Alternatives: Human intervention should be effective and timely in order to adequately address any AI challenges. This will allow AI system failures and threats to be resolved quickly and efficiently. An example of human alternative to AI is demonstrated in the City’s chatbot where a person looking for answers is also offered human alternatives for communication, such as phone numbers and emails of staff members, or Ask LebNH, a service that connects residents with the right person.
Cyber Services in coordination with the TRC will provide training opportunities for City employees to provide them with the necessary knowledge to use AI technology safely and appropriately. Employees who utilize AI in their work shall be trained on those systems.
- Training Resources and Tools: The City of Lebanon will provide employees with access to a variety of training resources to promote AI literacy, and will assist in providing the necessary knowledge in order to safely, efficiently, and responsibly use AI technology in the workplace. The TRC shall serve in a role to provide guidance to city staff and the use of AI.
- Policy Enforcement: Non-compliance with these guidelines may result in disciplinary action or restriction of access, and possibly even termination of employment. It is important to act safely, responsibly, and to use a degree of caution when using AI technology.
The Technology Review Committee is tasked with reviewing this policy and proposing revisions to the City Manager. The TRC will closely monitor the evolving use of AI and issues related to the use of AI in the City as well as the broader cyber environment.
This section is still currently under development and review.
- City of Boston, Interim Guidelines for Using Generative AI, (boston.gov)
- City of Lebanon, Cyber Policy ADM-450 Securing Sensitive Information
- City of Lebanon, City Code Chapter 185 Welcoming Lebanon
- City of Lebanon, City Policy ADM-126.1 Complaints and Investigations
- City of Seattle, Use of Generative Artificial Intelligence in City of Seattle, Seattle IT Memo (nlc.org)
- City of Tempe, Ethical Artificial Intelligence (AI) Policy
- White House, Blueprint for an AI Bill of Rights | OSTP | The White House