Take the guessing out of government with AI

By Esri

Combining geography and artificial intelligence creates a powerful new tool for decision makers.

Car accidents can be the tragic byproducts of any number of factors, each more unpredictable than the next: a sudden thunderstorm, a busy intersection, or even perhaps, gravel on the road.

These are just a few examples of the great deal of uncertainty government officials must deal with every day. Even the colours used in a heatmap can sway decisions, says Mansour Raad, a subject matter expert on Advanced Analytics, BigData and AI at Esri, the leading GIS company.

However, policies must always be made on accurate facts rather than human subjectivity. We look at how sensors, artificial intelligence, and geospatial technologies can help officials dig out underlying causes, identify patterns, and ultimately, make informed decisions.

Understanding the environment

Government officials are now able to gather an extensive amount of data from the environment and infrastructures using ubiquitous sensors. For instance, sensors on cars can detect slippage on roads, and alert other drivers to be cautious and drive slow, Raad explains.

Some governments who do not yet have a sensor network are using consented citizens as sensors, with the data submitted typically through their smartphones and apps. “If you are the sensor and you are willing to push information out, that is more reflective of what you are doing or what is happening around you,” Raad adds.

However, the sensors and the data they generate cannot provide answers on their own. A government must analyse it to distill patterns and causes that help answer their questions. Accuracy and clarity of decisions are crucial at this stage.

Making decisions with data

Geography is a crucial part of understanding and making decisions with such data because it relates objects to others in its proximity and puts things in perspective. “Everything is related to everything else, but nearer objects are more strongly related than distant ones,” Raad says. This is the first law of geography.

A method commonly used by governments is to create “heat maps,” with different colours denoting the density of crimes or traffic accidents, for example. These are meant to draw attention to the most critical areas but can be misleading to policymakers as they rely on human subjectivity.

In the above figure, the maps are generated from the same data, but each is classified and coloured differently. “If you colour-code a map in a certain way, you can sway the person deciding to go one way or another,” Raad points out.

Instead, governments should use “hotspots” which are built on proven statistical reasoning. Decision makers can be confident that the hot (or cold) spot is not a random event, but an emerging pattern they must pay attention to.

Asking the right questions

IoT, Big Data, and Artificial Intelligence are now going a long way to take guessing out of the game and make decisions more accurate. Governments can “start asking questions that could never be asked before” with the “marriage of geospatial and AI” - or what Raad calls Geo.AI. Implementations using machine learning can “plow through billions of records,” he adds, and “pick up those needles in the haystack” without explicit programming.

Esri is working with the Public Authority for Civil Information in Kuwait to enhance an app that uses AI for better traffic management.

The key to the analysis is that machine learning programs can understand the concept of traffic conditions and patterns without explicit instructions from engineers or programmers. “Instead of spending hours playing the programmer’s version of Whac-A-Mole, we decided to let the computer do the learning,” Raad wrote in a blogpost.

Esri is now working with other governments to look at how this combination of geography and AI - or GeoAI - can be used to tackle other knotty problems, like food security, epidemics, and crime.

However, even when technology supports decision-making, policymakers’ expertise remains crucial. A common pitfall is to assume that if two factors are related, then one causes the other. “Correlation is not causation,” Raad warns, adding that “this is where human intuition comes in.” Governments will require policymakers and business leaders who understand citizens to ask the most sensible questions from the system.

While artificial intelligence brings extensive data-mining capabilities to government officials, it is the combination with geography and expert intuition that brings about greatest success and accuracy.