Navigating AI in the Public Sector: An OCM Perspective

Everywhere you turn, artificial intelligence (AI) is the hot topic in nearly every industry, with its potential seeming endless. However, for the public sector, the applications and potential for bias with AI introduce significant concerns. With layers of oversight, funding requirements, public records requests and heavy audit requirements, the concerns with adopting AI are very real and complex. AI is coming, however, and the public sector can, and should, prepare, manage, and reinforce this transformative change with thoughtful and strategic approaches.

At Treinen, we understand these complexities. Our team, comprised of Prosci® certified change practitioners and consultants with Massachusetts Institute of Technology (MIT) AI certifications, specializes in guiding public sector organizations through thoughtful AI transformation by applying proven Organizational Change Management (OCM) principles.

Preparing for AI Transformation

The first phase of the transformation is preparing for the change. Begin by determining why you want to adopt AI and, perhaps more importantly, how you want to use it. To determine how your organization will use AI, identify your objectives and key results you want to achieve with AI. Don’t adopt AI simply because it’s cutting edge and could save time. Identifying “what’s in it for me” is foundational for effective OCM.

Start developing policies now, if you haven’t already, to set up guardrails and allow opportunities for the use of AI in your organization. Being proactive is crucial, as you may already have early adopters in your organization who are using generative AI chatbots without any known guardrails, creating risk for your organization. As you develop your policies and processes, focus on the human benefits and robust human involvement with AI.

Next, anticipate resistance before you begin introducing the change. According to a Prosci® study of change management professionals, the top concerns with using AI are a fear/lack of understanding (29%) and governance/compliance (22%). Fear and lack of understanding can be met with education around AI as well as showing people the benefits of using AI in their work, and how to use it appropriately. Developing thoughtful policies with detailed governance structures can address the second concern.

Managing the AI Transformation

The power of AI to support public service is compelling. Incrementally adopt AI in ways that make the most sense for your organization. Regularly reflect on what is working and what is not working and adjust as needed.

There are several use cases for AI to assist overburdened public servants with their daily work:

·       Taking meeting notes so staff can focus on engaging during a meeting

·       Analyzing staff skills and identifying training/resource needs

·       Developing charts from complex data

·       Keeping tabs on industry news and insights

·       Developing visually appealing presentations

·       Reviewing plans/presentations and providing recommendations for improvement

·       Practicing video presentations and providing recommendations

Keep the people in mind

With this potential comes great responsibility, and at Treinen we have begun exploring these use cases in our internal work. We are focused on making our work better, thoughtfully adopting tools and processes that incorporate AI rather than outsourcing our human thinking.

One of the greatest concerns with AI is the potential to replace people by doing their jobs. Directly address this concern by establishing clear expectations for AI use - specifically when staff are unable to complete a task, when AI can enhance efficiency or effectiveness, or when it enables employees to focus on higher-value work for the organization. Make it clear to staff that AI should augment human work, not replace their work.

A critical tenet of appropriate AI usage is keeping “humans in the loop.” This means that humans are involved at each step of the work AI is doing for you. Humans should be validating work, checking for bias, and editing outputs to meet the needs of the organization. Nothing created or edited by AI should be final or shared until a human has reviewed it for accuracy and bias. For example, this article was written by a human, then reviewed by AI for improvements. Humans reviewed the AI suggestions and accepted some and rejected others.

Approach incrementally and evaluate

Ensure your AI implementation is incremental. Select one or two use cases, and a small group of early adopters, and evaluate. Adjust as needed and add more uses as it makes sense. Provide training opportunities for the appropriate ways to use AI. Consider setting up a Community of Practice with early adopters who are excited and concerned about AI – to gather honest feedback on needed policies, procedures and best practices. Empower staff with choices along the way to increase buy-in, as well as the ability to pause initiatives if issues arise.

Collect feedback and listen to concerns to address them before going further. Talking with other public service organizations like yours can also provide valuable insights to help you adopt the right policies and procedures for using AI, as well as glean lessons learned from their experiences.

Implementing AI in the public sector necessitates rigorous data governance, not only sensitive data, but any identifiable data about the public you serve. Begin by applying AI to datasets not attached to people in any way. Consider AI tools (and carefully evaluate their terms of service) that securely store data in a server separate from all other organizations using the tool. Keep in mind your organization’s auditing, reporting and public records requirements to ensure you remain in compliance with all laws and rules as you use AI.

Plan for and navigate resistance

Planning for and navigating resistance through change is a core part of effective organizational change management. Several potential sources of resistance were mentioned above. Keep track of concerns you are hearing, and work to address them.  Observe how mitigations work (or do not) and adjust.  Maintain open, ongoing communication about progress. Go beyond listening and follow up with them about their concerns and what is being done. Showing people that you are acting on their concerns goes a long way to earning trust and helping them move forward with change.

Once you have iterated through some trials of AI use and have your policies adopted, consider establishing a champion network to expand AI across your organization. Have your early adopters serve as champions, leading their peers through the change, even if (or maybe especially if) those champions still have reservations themselves. Encourage the champions to set up their own feedback loops with the newest AI adopters and continue with your resistance management process to address concerns and mitigate risks.

Reinforcing Sustainable AI Adoption

Adopting AI at your organization will not be a single project with an endpoint, but an ongoing transformation in how your organization operates. Continue exploring what AI can do to help your teams, adding use cases and evaluating best practices for your organization. Continue reviewing and communicating policies, processes and tools. Provide clarity to staff about expectations, best practices and appropriate use cases for AI.

Taking a thoughtful, step-by-step approach to adopting AI can transform how you work while ensuring everyone is part of the journey toward achieving your desired outcomes.