As artificial intelligence (AI) has moved out of the realm of science fiction and into our daily reality, organizations across a broad swath of industries have increasingly begun to explore how best to deploy AI in the service of their missions. In the public sector, one category of AI seems particularly capable of helping government agencies deliver better, fairer, and faster service provision to constituents: machine learning (ML).

Machine learning mimics human decision-making without human intervention to improve existing processes and learn new ones. It has vast potential to help local governments analyze existing conditions, predict how those conditions might change, and respond in kind. However, the appeal of ML is not uncategorical. There are several inherent challenges to its adoption by government agencies, especially at the local level.

Here’s how the future of local government operations might be tied to machine learning.

Big Data is Bigger in Government

The sheer volume of data collected by government agencies is staggering — everything from parking fees and payroll concerns to citizen complaints, road analyses, social service, and security data. Warehousing, preserving, and retrieving this so-called “big data” efficiently and effectively across many disparate agencies has long been a bureaucratic stumbling block.

At the federal level, the Cloud Smart computing strategy articulated in 2019 has in some ways served as a mandate to find intelligent solutions to these ever-growing and evolving information needs. Cloud Smart has served as a model for local and state government entities, emphasizing the evaluation of existing and future skills requirements and improving worker training and hiring protocols to address skill gaps. This forward- and inward-looking approach is a terrific model for other government entities — both state and local — as they evaluate their current and projected personnel needs.

Additionally, the federal government has provided a model for AI and ML to solve current challenges and predict future ones, which local and state governments are increasingly investigating for their own needs. They are finding that emerging technologies, like ML, are helping government agencies and vendors improve processes for employees and constituents alike.

Understanding Machine Learning is a Predictive Powerhouse

Machine learning is an umbrella term that encompasses deep learning and tiny machine learning. What all machine learning has in common is the ability to process massive volumes of data and deliver human-like — or better than human — operations based on that data.

For instance, we’ve already discussed the fantastic complexity behind self-driving cars. Similarly, deep learning is a subset of machine learning that replicates human decision-making by modeling the neural pathways in a human brain. Deep learning relies on neural networks to complete advanced analytical operations, such as image and speech recognition.

Machine Learning is a Predictive Powerhouse

Machine learning enables computers to process data as a human would, at exponentially faster speeds and with exponentially greater accuracy. But, increasingly, ML is making it possible to predict future outcomes based on existing data, at a level of complexity beyond the scope of most humans.

Within the government sector, the predictive analytic capabilities of ML show particular promise in several spheres. These include:

Threat Prediction

Potential ML-driven capabilities for image and voice recognition, and real-time language translation, could help intelligence and law-enforcement agencies anticipate rising threats of fraud, terrorism, or cybersecurity attacks, like those that plagued several large school systems during the past year. Machine learning could also help health agencies to identify future public health threats and pandemics before they happen, allowing governments to adjust operations to prevent these threats. ML can help social service agencies track constituents, identify anomalies, and intervene in situations of abuse.

Predictive Maintenance

Machine learning has the potential to help governments analyze infrastructure issues — everything from road and bridge conditions to waterworks and sewage — and predict future maintenance needs before they become crises.

Dynamic Data Provision

Local governments might use ML to guide a constituent to online resources and predict what other resources might be of use to that constituent, based on various queries (not just one particular phrase or keyword).

Special Considerations for ML in a Government Setting

As powerful as ML is for local governments, it’s not without challenges. Several factors need to be considered when weighing an investment in machine learning solutions. These include:

Legal Issues

The novelty of machine learning technology means that the law around its use, especially around security, surveillance, and government intelligence, can be pretty gray. The threat of litigation, especially regarding the invasion of privacy, can be daunting to some government users. As ML becomes more pervasive, however, we can expect more clarity regarding the law around it.

Qualified Practitioners

As with all emerging technologies, finding employees who are adept with machine learning applications can feel like a herculean challenge. However, industry training programs are increasingly incorporating AI and ML skills into their curricula. A solution-agnostic consultant can help local government agencies identify strategies and methods for incorporating ML into their operations.

Evolving Tools

Navigating the complexities of machine learning is made more complex by the ever-changing nature of ML tools themselves. Machine learning is still a new field, constantly reinvented by new research and new applications. Getting a handle on ML and how it might serve your agency is certainly not easy in this shifting landscape. A management consultant trained in AI solutions for government clients can help you navigate this terrain.

“Black Box”

AI and ML applications are often inscrutable — carried out in the “black box” of the computer without explanation or rationale. It may be hard for human users to trust the analyses and predictions made by these applications. For a government user who relies on verifiable data to make informed decisions that affect constituents’ lives, this lack of transparency can be unnerving and even unacceptable. However, one of the recent advances in the field is the development (led by MIT researchers) of neural network training that provides accurate predictions and rationales.

Get Better, Fairer, and More Efficient Outcomes with ML

One of the most compelling reasons to incorporate machine learning into government programming is that ML has the power to reduce bias in reporting, analysis, and program recommendations. ML can actually improve algorithm accuracy over time and does so without the biases inherent in human decision-making. If your agency is committed to getting the right services to the right users, ML may help you do that more fairly and expeditiously than any person might.

Want to know more about what machine learning holds for the future of local government? Contact Momentum for further insights.