InnovateUS has concluded a multi-month engagement process to gather input on the development of a new philanthropically-funded, national online training program on responsible use of artificial intelligence (AI) for public sector workers. This post summarizes what we learned from leaders in federal and state governments and from industry, academia, and civic technology organizations.

Government leaders and AI experts universally agreed that training on the responsible use of AI to better serve residents is important and urgent and should be widely available, providing a clear framework of AI terms, use cases, and organizational considerations, with hands-on exercises connected to the learner’s day-to-day work. Informed by these insights, development of InnovateUS’s new free, scalable, and asynchronous online course — Responsible AI for Public Professionals — is underway, with the course anticipated to be available by early summer 2024.

Series of Planning Convenings Concluded

With our partners, we hosted a series of planning workshops to seek broad input into the development of a new curriculum to support public professionals in understanding and using AI (including Generative AI).

The first convening was hosted on December 4 with representatives of 11 state governments (hosted by InnovateUS and Beeck Center for Social Impact + Innovation at Georgetown University), the second on January 11 with representatives from 19 federal departments, agencies, and offices (hosted by InnovateUS and the Partnership for Public Service), and the third on January 30, with participation from a cross-sector group of experts from 16 industry, academic, and civic technology organizations. We also sought input from our InnovateUS State Partners, as well as from the State Chief Data Officers Network and the Chief Digital Service Officer Network, both part of the Digital Service Network organized by the Beeck Center.

What We Learned

During these conversations, we explored the following topics and heard the following input:

Roles and Functions to Prioritize for AI Training

In response to which types of public-sector occupations most need AI training in the near term, we heard that there is a need to design baseline AI training for a general audience of public professionals and for AI decision makers. These core audiences include:

AI DECISION MAKERS who are already or will soon be involved in policy development, program integration plans, and decisions related to whether to build AI capacity or buy AI services and tools;
A GENERAL AUDIENCE OF PUBLIC PROFESSIONALS, including frontline staff who deliver services, engage with the public, work with data, and perform core day-to-day agency business activities also urgently need to learn about AI.

In addition, we heard encouragement to consider future specialized trainings for:

Procurement and acquisition professionals who will design solicitations and review proposals from vendors and providers of AI tools and services
Technical and information technology (IT) staff at all levels
Roles that are focused on the production, management, and use of data, including statistical offices, privacy officials, and custodians of records.
Customer service and digital transformation teams
Human resources professionals and hiring managers charged with building AI-ready teams
Legal professionals engaged in evaluating AI risks and risk mitigation (planning for an AI course for government legal teams is in the works here at InnovateUS, and we anticipate having the program online by September 2024)

Looking outside of federal and state executive branches, we also heard the importance of ensuring training on the responsible use of AI is available to local government workforces and to legislators, mayors, city and county councils, and other elected officials.

“Getting [training to] chief decision makers who are likely going to be on the very near term front lines of making important decisions would be quite useful.”

- former federal official

“I'm hearing from tech professionals that it is the business leaders who need to go [through training] first because they will be the ones who are procuring tools, and if they don't fully understand risks and benefits, how a pilot might work, how a sandbox might work, they might be talked into something that either isn't useful or doesn't get them where they need to be.”

- current state training program manager

“It becomes really difficult, I think, to separate what traditionally has been considered AI versus generative AI, and those capabilities, and educating people who are deciding procurement contracts, or who are figuring out what solution answers what problem.”

- state technology expert

“We're focusing squarely on the ‘nontechnical line level employee’ that is about to get access to the most unprecedented advancement in computing technology they've ever gotten. If there were an opportunity rating scale, it would be closely correlated to the terror rating scale on this one.”

- current federal leadership development executive

“I think you can break it down into the categories of the willing, the somewhat curious, and then the resistant. And we're focusing on the willing and a lot on non-technologists–the more lay folks who want to lean into this–and how they can use it in their jobs and then getting feedback from them on how it's been useful, what's not been useful, and going from there.”

- current federal technology policy leader

Key Learning Concepts to Promote Responsible Use of AI

In response to what topics are essential to explore through the curriculum for public professionals on the responsible use of AI for service and mission delivery, we heard demand for hands-on training that provides a framework of basic terms and definitions of both generative and traditional AI, as well as a typology of AI use cases relevant to government work and exercises that advance understanding of the roles of leaders and staff and the importance of data quality, prompt engineering, assessment, and evaluation in ensuring responsible AI use. Specific concepts prioritized for inclusion in training were:

FRAMEWORK OF BASIC TERMS AND DEFINITIONS: The need for a basic overview of types of AI with a plain language explanation of how these tools work and how they are trained, as well as definitions and terminology that can support shared language and understanding within and across agencies;
TYPOLOGY OF USE CASES AND HANDS-ON EXERCISES: Enthusiasm for hands-on exercises that help learners identify and explore the relevance of generative AI to their day-to-day workflows–informed by a simple typology of generative AI use cases as well as real-world examples of AI applications in the public sector;
FOCUS ON DATA: Encouragement to infuse course modules with information about the role of data management, quality, and processing in improving outcomes of AI applications, including considerations of privacy, security, and safety;
ACCOUNTABILITY FOR DESIGNING RESPONSIBLE AI APPLICATIONS AND MITIGATING RISK: The importance of assessing risks and risk mitigation strategies during training, including how to avoid or mitigate unintended consequences and legal risk; identify potential bad actors/applications; protect civil rights; mitigate bias; and infuse values and ethics into AI application development, training of AI models, and prompt engineering;
CHANGE MANAGEMENT: Connections between AI training and broader executive-level efforts on AI change management and organizational readiness, such as how agencies can build AI-ready teams; and
BOTH GENERATIVE AND TRADITIONAL AI: The need for other advanced training content for AI decision makers and leaders on use cases for traditional AI/machine learning, assessment of AI solutions, evaluation of AI implementation, and effective practices related to governance and accountability.

“One of the most important things really is that people learn to use the tools in ways that make sense to them and are true to who they are and the work that they are trying to do in the world.”

- civic technology leader

“I think many of our [AI] use cases depend on folks understanding the implications for how they collect and manage the data they want to feed into these systems."

- state data leader

“There are a lot of things that we do [with AI] that could be rights- or safety-impacting because, for example, it has an effect on employment choices, so doing something on that up front to provide examples of use cases that need to be thought through, in the context of civil rights and safety."

- federal agency official

“I find that it's important to help people see that they can impart their own values, essentially, into what's going on with the AI.”

- graduate business school university professor

Effective Formats for AI Learning

In response to how this training should be delivered and which formats would be most helpful to learners, we heard enthusiasm for a modular, learn-at-your-own-pace asynchronous online course, and for hands-on exercises that give learners an opportunity to try one or more of the main Generative AI tools freely available online.

Participants encouraged us to help ensure that learners use data in those hands-on exercises that are either already in the public domain or synthetic sample data, and that learners are able to complete the exercises within their agency’s IT environment or on their personal devices. Participants also expressed interest in including quizzes or other forms of integrated assessments so that learners can test their progress as they go.

“I think the practical experience will make the outcomes [of training] richer…even if you have to do it in a sandbox of sorts. Fingers on keyboard–I think it is the difference between success and not being successful. It is basically essential that they actually put fingers to keys. And so figure out ways either inside of the [training] modules to allow them to do that, or alongside.”

- industry expert

"I'm focused on marrying the idea of training that is consumable and easily integrated, with the idea of having learners contextualize what they are learning and see how it connects back to their role."

- federal agency training manager

Access to Learning Opportunities

In response to how state and federal government agencies might best access training for their workforces and whether there were any considerations for how to encourage uptake across public-sector workforces, we heard interest in having access to a single free or low-cost foundational course available to and common to all staff as a baseline. In addition, we heard interest in:

ABILITY TO SUPPLEMENT WITH CUSTOM CONTENT: flexibility to add custom modules or more advanced content as needed based on policy or role/function;
UNION SUPPORT WOULD BE USEFUL TO UPTAKE: Observations that support from employee unions, management associations, and other labor groups and associations for training could encourage uptake of training within workforces;
INFORMATION FOR MANAGERS AND EXECUTIVES: Suggestions to provide information for managers and executives on the role, goals, and expected outcomes of the training course, and to connect course modules with AI skills competency frameworks developed by agencies/administrations;
PLAN FOR AS-NEEDED CONTENT UPDATES: Encouragement to plan for potentially frequent as-needed updates to course modules as technology capabilities evolve; and
ACCESS TO COURSE COMPLETION DATA: Interest in being able to receive data about employees’ course completion, and possibly linking course completion to certification, badges, or employee performance plans.

“[Unions] should be advocating, encouraging, using this as a lever, not just with helping their members increase their skill sets to get higher wages. This also has a component of how [their members] are going to be influenced and impacted...A partnership with labor will help to accelerate acceptance.”

industry expert

"I, as an actual employee, only have X number of hours of professional development. So how do I fit this into all the other stuff I'm required to take for compliance or given what my supervisor has said."

public interest technology leader

"[Our agency] created a competency framework in which they articulated all the relevant skills that are rated to data across various domains. Consider crosswalking the [training] modules, and how they relate to the [competency framework's] language. [Our agency is] now going to...chose the specific role, the specific skill levels that we need to help [our workforce]. …so that's an integration point there."

federal data leader

"Anytime something new is introduced, it's going to be challenging. And so the communication around it, the consistency of communication, the debunking of myths that will continue to pop up about it...ongoing communication from a change management perspective, or else a training is just a training, and then it gets left behind. How does it best fit into that organizational change that's underway?"

former federal executive

“Really making sure that this whole [course] is designed with change in mind because this field is evolving really fast."

professor of computer science and civic technology expert

What’s Next and How to Stay Engaged

Responding to the feedback and input we have gathered, curriculum development is well underway. We are planning interactive learning modules for general audiences and for AI decision makers and business leaders.

Follow along here at innovate-us.org for upcoming updates about this new course on the responsible use of AI by public professionals. Our next update will provide an overview of our course content and modules, prior to the course’s launch.

Looking to get started now with AI programming? Ahead of our course launch, InnovateUS is offering various live learning workshops on AI. Check out InnovateUS's free AI resources here! In addition, the Partnership for Public Service offers a no-cost AI Federal Leadership Program.

About InnovateUS

InnovateUS is a nonprofit, non-partisan multi-state initiative to upskill the public sector workforce in digital, data, and community engagement skills. All our programming is free and designed to upskill public professionals to use technology responsibly to serve the public, meet missions and solve problems.

Want to be a part of our community of innovators? We’d love to keep in touch! Sign up for our mailing list or follow us on LinkedIn.

Cristin Dorgelo is a Visiting Fellow with InnovateUS and the Burnes Center for Social Change.

Blake Valenta is a Senior Fellow with InnovateUS and the Burnes Center for Social Change.

Something went wrong

Want to be a part of our community of innovators?

Subscribe for updates