When Using Digital Twins, Start With the Problem and the Data
Every agency will be able to make a strong case for an application of digital twins.
A government agency with strong traffic data, for example, might start by modeling the flow of vehicles on roadways. But officials still must define what they want to understand. Is the goal to examine congestion or crashes? Does the agency want to see how weather affects the movement of traffic across roads and highways?
Those questions determine exactly what data the model needs. If officials want to model the effects of weather but lack reliable historical weather data, the results will suffer.
The use case should therefore start with a specific problem and a realistic assessment of the information available to model it. This may require interagency partnerships and correlation of data that’s traditionally held in silos.
READ MORE: Here is a government IT guide for AI governance.
Digital Twins Require an Accurate Picture of Current Conditions
An agency might want to know how traffic will change during a storm or where flooding could occur. Before officials can anticipate what might happen, however, a digital twin first must accurately represent current conditions.
Think of that as the descriptive stage. Government personnel gather enough reliable data to establish a baseline of the environment they want to model. Once that baseline is sound, they can examine why conditions changed or begin predicting what could happen under different circumstances.
More advanced models can move into prescriptive uses, recommending an action or potentially automating one. And while that may sound exciting, agencies cannot jump directly to that stage. If the baseline is inaccurate, the predictions built on it will be inaccurate too.
LEARN MORE: Local AI agents can expand tools for government agencies.
Data Silos Still Matter Despite Advanced Digital Twin Technology
Building the baseline often requires information held by different government systems.
Imagine a transportation agency that wants to determine the impact of precipitation on traffic flow. A different agency may have the rain and snow data in its own data stores. The transportation agency would need that data before it could build a functional twin.
That creates a familiar problem: Some government departments do not share data easily.
Agencies should examine their data architecture and data strategy early. They need to know whether they can access the required information, whether their infrastructure can support it and whether governance or compliance rules affect how the data is shared.
The people working with that information matter too. Collecting data is only the first step. Agencies need staff or partners who can determine which data belongs in the model and how it should be used.
DIVE DEEPER: Data dashboards support citizen services through data analysis.
Start Small and Build Out When Using an AI Digital Twin
As with other AI tools, the quality of the result depends heavily on the quality of the information provided to the system and how the model is built and maintained.
As digital twins become easier to explore, there can be a temptation to start with an ambitious project, such as modeling an entire city.
We recommend choosing a narrower path forward.
Pick a subsystem, process or operational problem that officials understand well; ideally, one backed by a strong and accurate data set. Build the model and see whether it reflects what is actually happening. Once leaders can trust the results, they can add more sophisticated capabilities or expand the scope.
Start small and build out.
