With the availability of intelligent virtual assistants like Siri, Cortana and Google Now fundamentally changing the service expectations of citizens, federal government agencies are increasingly looking to cognitive computing and natural language processing to improve digital service delivery.
It’s a long way from Microsoft’s (often loathed) office assistant Clippy in the late 1990s. Today’s sophisticated virtual assistants now understand colloquial language, provide intelligent answers to questions and assist users to navigate websites.
For government, this means the possibility to personalise interactions with citizens, in line with the Coalition’s policy for e-government and the digital economy, and simplify complex and abundant information on agency websites.
The Australian Taxation Office was the first agency to implement a virtual assistant — “Alex” — on its website in August 2015. It responds to queries about personal tax, and importantly, understands conversational language and learns from questions to improve answers.
However, despite learning from every interaction, Alex doesn’t rely on cognitive computing software to automate interactions. She does, however, “have other capabilities often considered as elements of cognitive computing such as natural language processing and reasoning”, according to an ATO spokesperson.
“Alex learns how clients ask her questions, assessing syntax and grammar, to deliver the correct information, and refining when clients indicate she got it wrong,” the spokesperson said.
Alex’s software is provided by Nuance Communications, a company that has made a name for itself through its speech recognition and imaging software. Nuance Communications’ FreeSpeech voice authentication solution has been used across the ATO since September 2014 to simplify the taxpayer authentication process at call centres and on the mobile application.
IP Australia also leverages Alex, developed in collaboration with Datacom and Nuance Communications. Now the Department of Human Services is looking to go one better, planning the introduction of an intelligent virtual assistant.
Chief technology officer Charles McHardie told the 2016 CeBit Conference in early May 2016 that a virtual assistant or avatar would be a cornerstone of DHS’ next-generation splash page:
“We’re doing huge amounts of work around virtual assistance, cognitive computing and machine learning with IBM at the moment, and we’ll have a big release around that soon.”
McHardie said that unlike the ATO’s virtual assistant, which uses a logic tree and looks for key words as the basis for responses, DHS intends to use true machine learning for its virtual assistant. The agency will work closely with Microsoft and Auckland University, and continue its ongoing collaboration with IBM on Watson.
More details will likely become available when DHS publishes its 2016-20 technology plan, expected to be released shortly.
Just how the development of numerous virtual assistants fits in with the Digital Transformation Office’s plan to phase out the use of individual government websites to deliver information and services to citizens remains to be seen. As information services, virtual assistants and smart answer forms fall under the scope of the DTO’s Digital Service Standard, and are required to meet 14 criteria to ensure all newly designed government services are simple, clear and fast.
Intermedium’s half-day conference — Digital Transformation: How Far? How Fast? How Much? — will benefit anyone involved in improving services to citizens, or improving the efficiency and effectiveness of their agency.
