Government Service Delivery: Beyond Al and Machine Learning

Recently, it seems the whole world has started using Artificial Intelligence (AI) for everything from making purchasing recommendations online and setting the temperature on your thermometer, to controlling machines in factories and driving trucks. 

But the inability for these systems to explain “why” the decisions and recommendations where made can cause confusion, uncertainty, and loss of trust. Prominent folks such as Bill Gates, Elon Musk, and Stephen Hawking have sounded the warning on relying too much on these opaque systems. Although many things need to be done to ensure the safety and increase transparency of AI, a new area of research called Explainable AI (XAI) may be part of the solution.

XAI aims to build AI that can describe why a certain recommendation was made, leading to systems that are easier to understand, trust, and ultimately to work with. There are several large research efforts being undertaken by organizations such as the Defence Advanced Research Projects Agency (DARPA) of the US Department of Defence, as well as the Palo Alto Research Center (PARC). Aki Ohashi will describe why XAI may be needed, the current activities within this nascent field, and the potential impacts and implications on government and industry.