Commentary: An Insider’s Road Map for AI in Government

The emergence of artificial intelligence has brought about many concerns and theories about its effects on the future. In the midst of the national dialog, the fact is we are all consumers (users) of products and services that contribute to the machine learning that is driving AI’s rapid evolution.

In seamless and nuanced ways, AI has reduced cognitive load and simplified decision-making in our daily lives, from media streaming choices, voice-activated personal assistants, social media feeds, and automotive navigation systems to environmentally responsible infrastructure that enables thermostats and irrigation systems to preserve energy.

Many individuals may make the mistake of disassociating these products and services from the underlying AI that makes those things so useful. But understanding AI, even conceptually, allows us to filter mis-, dis-, and mal-information, manage our personal privacy, and remain compliant with policies our employers have implemented to protect us and the constituents we serve.

By now, most have probably realized that generative AI (GenAI) is a different beast even in its infancy stages. In the era of modern economics, the Industrial Revolution is the only other thing that comes to mind that has the potential to rapidly disrupt so many vertical markets and economic systems the way AI may do.

The aggressive market penetration and rapid user adoption of AI have been fascinating to watch. I came from a generation that grew up watching movies depicting futuristic fictional AI like SkyNet, R2-D2, HAL, iRobot, etc. GenAI’s replication of human cognition coupled with a healthy imagination from these great sci-fi movies might be the driver behind many of the salacious perceptions of AI we are seeing today.

But just as Henry Ford’s 1940s prediction of flying cars has not yet materialized, even after almost 100 years, we realize that while the possibilities are infinite, consumer demand and market relevance often move at a slower pace than our imaginations.

AI’s current market noise and anxiety amongst consumers, labor advocates, legislators and industry leaders strongly align with the historical events of industrialization. The Industrial Revolution profoundly changed the traditional workforce and economy, reshaping social structures, labor practices, and domestic and international economic systems. While the Industrial Revolution is historically reflected as an era of progress and innovation, it also brought challenges such as exploitation, inequality, and environmental degradation.

What role did government play during times of major transformation? The evolution of government began as a result of its struggle with industrialization. With poor labor laws leading to exploitative working conditions and having little to no health and safety regulations in factories, problems such as widespread worker illnesses, injuries, and unsafe working conditions ran rampant. Environmental neglect also resulted in pollution, deforestation, and damage to our natural resources.

Unionization and degrading social welfare conditions began to push government to reset and create labor laws to protect children and improve workplace safety. Government also had to learn how to formulate policies to capitalize on industrialization and facilitate international trade.

This led to new taxation systems, the formation of trade agreements, tariffs, and other policies to support domestic industries and rules associated with imports and exports. Our legislators and government leaders immersed themselves in understanding employment law, factory production systems, macro/microeconomics, and social welfare systems.

However, the Industrial Revolution and government’s response to the economic and socioeconomic transformation did not happen overnight. It is generally considered to have started in the late 18th century, approximately during the 1760s or 1770s, and lasted until the early 1900s. The exact duration of industrialization can vary depending on the specific context and the region.

Therefore, much of the consumer speculation and anxiety about “autonomous AI-driven systems and services taking jobs and disrupting markets” lacks a general understanding of how, and at what pace, large complex systems actually transform. Revisiting Henry Ford’s prediction on flying cars, we realize that he could not have anticipated the innovation that redirected efforts toward alternative-powered and autonomous vehicles.

As we continue to see refinements of road-based automotive technologies, we see a tangible example of Friedrich Nietzsche’s quote, “He (she) who would learn to fly one day must first learn to walk and run and climb and dance; one cannot fly into flying.” When predicting AI’s evolution and its impact on our lives, imagination must be counterbalanced with education and a realistic contextual understanding of industrial and labor systems transformation.

As AI begins to play a bigger role in our daily lives, government will need to evolve along with emerging technologies. Our legislators, elected officials, and administrators must understand, at least conceptually, AI and the underlying technology that powers it. The key is to be educated, and not to fake expertise — don’t attempt to govern what you don’t understand. Creating policy and/or law before you first understand the technology will only lead to counterproductive bureaucracy suppressing technological evolution and the associated value proposition of AI.

Legislation should not be overly prescriptive and shouldn’t reference specific technologies, but rather it should describe the desired outcomes. Beyond the buzzwords learned in a news article, talking points drafted by staffers, or catchy quotes for media coverage, government leaders will need to educate themselves on how AI can both positively and negatively impact constituencies, and government processes that facilitate critical services.

Government leaders should respect the academic discipline of computer science and information technology. They should listen to technology managers the same way they listen to doctors, accountants, engineers and lawyers. If you want maximum results from technologies such as AI, entrust your technologist to sit at the table with other senior government executives.

AI needs a healthy foundation to thrive. Many government organizations have unintentionally underinvested in technology. If your organizations lack data governance and literacy or have an immature enterprise architecture, then you are not likely to find value in AI beyond elementary use cases. If you view IT budget investments as building blocks, AI would be one of the top blocks. Be intentional with your IT investments and understand that AI needs a healthy IT ecosystem to grow and thrive.

In government, technology is a social science. We must understand that while AI can help solve immense issues related to homelessness, child protection, social justice, or mental health, acknowledging inherent bias will be critical in designing inclusive and equitable AI services.

It is important to understand that technology bias did not begin with AI, nor will it end there. Data can perpetuate racism; government has unintentionally and unknowingly designed and implemented biased technologies for decades. Predictive policing systems, facial recognition technologies, and hiring algorithms for years have resulted in unfair and/or biased outcomes. There are many biased data sets coursing through the veins of our government systems. Untrained or poorly designed AI will lack the discretion to filter or flag those anomalies and may recommend inequitable actions.

Data practitioners and AI designers will need to ensure that data sets are diverse and representative of the population they serve and interact with. This process is not simple, especially when historical data sets may be riddled with biased data. If designed correctly, AI can be trained to identify biased data sets and help improve the fairness and accuracy of decision-making.

Your AI design must be human-centered. As government embarks on this AI journey, including labor unions, front-line staff and the public in the planning and design process will be critical to success. Understanding your stakeholder community is the most critical aspect to designing good AI-driven processes. This will also allow you to educate your stakeholder community and dispel some of the misperceptions about AI. More importantly, a human-centered design process will allow your AI developers to evolve AI designs to best meet the needs of your stakeholders.

Your workforce may be skeptical and potentially resist using AI; move slow. Introduce small, simple use cases that increase efficiency and reduce workload for front-line processes. Introduce AI in simple ways to allow staff to dispel notions of ’80s sci-fi — recommending slight adjustments on calendars, or drafting routine or salutatory emails. Chatbots can draft recommend responses to questions before they are sent by human client support staff.

Allow your lawyers to use AI to help with eDiscovery and crawl digital case notes and legal journals to help identify case law, precedent, and other legal trends. Allow your appointment systems to recommend bookings to optimize resources and respond to service demands. Use AI for language translation during emergencies, when firefighters and law enforcement personnel may not have access to human translators.

Allow humans to be the bookends to your AI-driven processes. Verify your AI’s accuracy. Allow your data scientist, practitioners, lawyers, and financial analyst to validate your AI processes. Humans must continue to ensure that the data, machine learning, and overall designs beneath AI are producing quality and accurate results. In the words of my quantitative methods professor, “Show your work.” When flaws or inaccuracies are identified, clean your data, adjust your designs, and calibrate learning models. When adjustments are made, be transparent and communicate changes to stakeholders; transparency builds trust.

As your organization matures, GenAI use cases can be considered using content generation that allows for a more immersive and conversational experience between government and their constituents. In addition, your organizations will need to access the vast amount of enterprise data between your collective government agencies in a safe and secure manner to allow for summarized analysis that provides value to your residents.

Artificial intelligence and human capabilities are not in a zero-sum relationship. While AI can automate tasks and improve efficiency, the more important value proposition is the potential to complement human skills and enhance productivity. The symbiosis between humans and AI has very distinct roles, allowing humans to be creative, emotionally intelligent, empathetic and allowing AI to do rudimentary and repetitive tasks.

As government leaders enter a potentially turbulent budget climate, traditional approaches to governing and managing safety-net services are no longer feasible. The public will depend on us to architect a new way forward, and progressive ways to deliver more with less.

AI will help us to fill those gaps.

Jeramy Gray, a frequent speaker at government and industry forums, was featured in an Industry Insider — California One-on-One interview in October 2022.