Artificial intelligence systems can – if properly used – help make government more effective and responsive , improving the lives of citizens.
Artificial intelligence systems can – if properly used – help make government more effective and responsive , improving the lives of citizens. Improperly used, however, the dystopian visions of George Orwell’s “ 1984 ” become more realistic.
On their own and urged by a new presidential executive order , governments across the U.S., including state and federal agencies, are exploring ways to use technologies .
As an researcher for more than 40 years, who has been a consultant or participant in many government projects, I believe it’s worth noting that sometimes they’ve done it well – and other times not quite so well. The potential harms and benefits are significant.
An early success
In 2015, the U.S. Department of Homeland Security developed an system called “ Emma ,” a chatbot that can answer questions posed to it in regular English, without needing to know what “her” introductory website calls “ government speak ” – all the official terms and acronyms used in agency documents.
By late 2016, DHS reported that Emma was already helping to answer nearly a half-million questions per month , allowing DHS to handle many more inquiries than it had previously, and letting human employees spend more time helping people with more complicated queries that are beyond Emma’s abilities. This sort of conversation-automating has now been used by other government agencies , in cities and countries around the world.
A more complicated example of how governments could aptly apply can be seen in Flint, Michigan. As the local and state governments struggled to combat lead contamination in the city’s drinking water, it became clear that they would need to replace the city’s remaining lead water pipes. However, the city’s records were incomplete, and it was going to be extremely expensive to dig up all the city’s pipes to see if they were lead or copper.
Instead, computer scientists and government employees collaborated to analyze a wide range of data about each of 55,000 properties in the city, including how old the home was, to calculate the likelihood it was served by lead pipes. Before the system was used, 80% of the pipes dug up needed to be replaced, which meant 20% of the time, money and effort was being wasted on pipes that didn’t need replacing.
The system helped engineers focus on high-risk properties, identifying a set of properties most likely to need pipe replacements. When city inspectors visited to verify the situation, the algorithm was right 70% of the time. That promised to save enormous amounts of money and speed up the pipe replacement process.[…]