By Dan Tienter, Municipal Advisor

Best Practices for Using Artificial Intelligence Systems

In recent years, Artificial Intelligence (AI) emerged as a potentially transformative force in our personal and professional lives, perhaps forever changing how we live, work, and interact with technology. Yet, as with any powerful tool, AI comes with its own set of advantages and disadvantages. When properly deployed, it offers enhanced efficiency, data-driven insights, improved decision-making, and more personalized end-user experiences. However, it also presents several challenges, including potential job displacement, questions about bias and fairness, privacy concerns, cybersecurity risks, trust issues, and even basic inaccuracies. Perhaps the chief executive of a large Minnesota county put it best: AI technologies represent “sort of a mixture of promise and terror” for all organizations, including local governments.

The benefits of AI-supported systems, much like information technology in general, focuses on its potential as force multiplier and its capability to improve service delivery.

1. Enhanced Efficiency. AI systems can automate repetitive tasks, leading to increased productivity, effectiveness, and efficiency.
2. Data Review & Analysis. AI systems can process and analyze large, sometime incomprehensible, datasets quickly, extracting valuable insights and patterns.
3. Decision Making. In certain situations, AI systems can make data-driven decisions faster and more accurately than humans or other computer systems.
4. Availability. AI systems can operate uninterrupted, allowing organizations to provide certain services and information at any time of the day.
5. Personalization. AI systems enable personalized experiences for users, such as recommendations and customer service interactions.
6. Safety. In certain environments, organizations may use AI technologies to monitor risks and respond more quickly to threats and other concerns.

However, the disadvantages of AI technologies often focus on its impact on individuals, its reliance on sometimes flawed data, and the risks to data or even national security.

1. Employment Changes. Automation as a result of AI technologies may lead to job displacement, changes in the workforce, and potentially overreliance on AI systems themselves.
2. Bias & Fairness. AI systems often possess the same biases found in the people who created them, which may lead to unfair or even discriminatory outcomes.
3. Privacy Concerns. AI systems often rely on large amounts of data, raising privacy and concerns about data collection and usage.
4. Ethical Considerations. AI systems raise ethical questions around issues like algorithmic accountability, transparency, copyright infringement, and plagiarism.
5. Security Risks. Like all networked systems, AI technologies are vulnerable to security threats, such as data breaches and similar cyberattacks.
6. Complexity & Understanding. Developing and maintaining AI systems can be complex and resource-intensive, requiring specialized skills and expertise.

To navigate these emerging, and sometimes competing, AI technologies, local governments should regularly review and adhere to established best practices related to AI technologies. In 2023, the National Institute of Standards and Technology (“NIST”) of the United State Department of Commerce released an Artificial Intelligence Risk Management Framework to help organizations responsibly design, develop, and deploy AI systems. This framework is based on seven key trustworthiness tenets:

1. Valid & Reliable. AI systems should regularly produce consistent and accurate results and do so within a reasonable time, which helps ensure the systems function as intended and allow for rapid intervention to correct errors or issues.
2. Safe. Al systems should not endanger life, health, or property. More specifically, the AI system should conform with the safety standards of the specific industry using the software. Generally, this includes robust testing, real-time monitoring, and the application of various risk mitigation strategies.
3. Secure & Resilient. AI systems must be able to withstand and recover from unexpected events. They should maintain confidentiality, integrity, and availability by preventing unauthorized access or use. And like any other technology, the protection of data must remain a paramount concern.
4. Transparent & Accountable. AI systems should allow users to clearly understand the purpose of the software and know exactly when they are interacting with an AI-supported tool or technology. They should also keep accountability measures in place commensurate with the sensitivity of the activity or data employed by the system.
5. Explainable & Interpretable. Al system should provide users with insights into the functionalities of the software. It should also possess the capability to clearly demonstrate how and why the AI system generated its outputs.
6. Privacy-Enhanced. AI system design, development and use should place value of anonymity, confidentiality, and control to prevent unnecessary intrusion or observation of individuals and/or groups.
7. Fair. To avoid and manage bias, AI system should recognize the importance of fairness as a concept. They should also test for systemic (i.e., uses of AI systems), computational and statistical (i.e., use of representative data), and human (i.e., individual or group perceptions) biases.

In keeping with the specific trustworthiness requirements, any local government seeking to use AI systems should develop (and regularly review and update) an AI Usage Policy to incorporate current best practices, access and use conditions, current and future technological advancements, regulatory requirements, and system supervision standards. Through such a policy, local governments may better deliver on the promise of AI technologies while avoiding the potential terror associated with these powerful new tools.

Lastly, perhaps to make the potential and concerns of AI more readily apparent, it should be noted that a particular AI tool helped draft portions of this article. While the authors and editors certainly applied their individual expertise, style, and context, AI generated meaningful portions of the actual content. As a result, it made us, the authors, more efficient, insightful, ultimately useful to the you, the readers. But it also raises questions of authenticity, priority, and utility. Of course, this article will not completely solve for these questions, nor will it be the last example of an effective and thoughtful use of an AI technology. However, it should serve as something of an ironic demonstration of the need for meaningful policy development regarding the use of AI.

Important Disclosures:

Ehlers is the joint marketing name of the following affiliated businesses (collectively, the “Affiliates”): Ehlers & Associates, Inc. (“EA”), a municipal advisor registered with the Municipal Securities Rulemaking Board (“MSRB”) and the Securities and Exchange Commission (“SEC”); Ehlers Investment Partners, LLC (“EIP”), an SEC registered investment adviser; and Bond Trust Services Corporation (“BTS”), holder of a limited banking charter issued by the State of Minnesota.

Where an activity requires registration as a municipal advisor pursuant to Section 15B of the Exchange Act of 1934 (Financial Management Planning and Debt Issuance & Management), such activity is or will be performed by EA; where an activity requires registration as an investment adviser pursuant to the Investment Advisers Act of 1940 (Investments and Treasury Management), such activity is or will be performed by EIP; and where an activity requires licensing as a bank pursuant to applicable state law (paying agent services shown under Debt Issuance & Management), such activity is or will be performed by BTS. Activities not requiring registration may be performed by any Affiliate.

This communication does not constitute an offer or solicitation for the purchase or sale of any investment (including without limitation, any municipal financial product, municipal security, or other security) or agreement with respect to any investment strategy or program. This communication is offered without charge to clients, friends, and prospective clients of the Affiliates as a source of general information about the services Ehlers provides. This communication is neither advice nor a recommendation by any Affiliate to any person with respect to any municipal financial product, municipal security, or other security, as such terms are defined pursuant to Section 15B of the Exchange Act of 1934 and rules of the MSRB. This communication does not constitute investment advice by any Affiliate that purports to meet the objectives or needs of any person pursuant to the Investment Advisers Act of 1940 or applicable state law.