The wealth of data available to government today can be overwhelming. Organisation data, program data, external agency data and data created by the Internet of Things (IoT) are among the sources of information accessible to government. But is it useful? And if so, how do you generate useful analytic insights to support decision making?
Sifting through the data to identify potential patterns relating to important outcomes is not a simple task. With a wealth of data comes noise – irrelevant information that provides no value to the decision-making process.
Artificial intelligence, or AI, is a tool that can assist in this complex process. No longer a concept in the realm of science fiction, AI helps to sift through data and identify patterns that could improve decision-making, providing critical insights that may previously have been invisible.
Ray Greenwood, AI and Machine Learning Domain Expert with SAS Australia and New Zealand, is among the experts assisting government in making the transition to an AI-supported environment. As he explained, it is the next phase in a continuing evolution of helping public service agencies make better use of data for stronger decisions – and a new market standard for organisational analytics.
“It is the latest tool and the best available,” Greenwood said. “It’s the latest generation of techniques that can help us achieve the exact same goal we have always had for better decisions and governance.”
How can AI support government?
AI is already being used globally to support a range of programs.
In Africa, the WildTrack program is using AI to match crowd sourced photographs of animal footprints in the wild to a known footprint database. It allows wildlife experts to monitor and track populations of endangered species and identify impacts of environment, population spread and poaching on wildlife populations on a scale not feasible using only human resources. With the geographic footprint animals can cover, this is a critical tool for the protection of animals.
“AI can help out in really unexpected spaces,” Greenwood said. “The key thing to remember about AI is that it’s not like we’ve invented a new technique to solve completely new problems. The problems are still old problems – but we’ve now got better tools to address them.”
For Australian government agencies, a common requirement of advanced analytics is in supporting workforce planning.
“Being able to identify what set of skills might be required to support particular programs, and understand that in advance, requires some very strong analytic capability.” Greenwood said. But AI, he believes, is best when it tackles highly structured and repetitive processes that supports decision making – tasks that consume a “fair amount of human time”.
“That would be a great place to start,” he said. “Machines can do a good job of automating these processes, and if you can free up human capital to be applied within another part of the organisation, that will really start to multiply the effects of AI rather than simply answering business questions with numbers.”
Free breakfast seminar
ATO Deputy Commissioner of Smarter Data, Marek Rucinski joins our expert panel who will discuss new leadership strategies using system thinking, AI and the deployment of advanced analytics.
What does government need to benefit from AI?
An important point Greenwood stresses is that data doesn’t always have to be large volume or perfectly structured to enable government to benefit from AI.
“From an AI perspective, you can still get big insights from small data,” Greenwood explained. “There is definite value regardless of the volume of data. An agency that is more policy focused, that doesn’t necessarily have a direct contact relationship with citizenry may have sparse data. But I suggest advanced analytics and AI would still assist in supporting better decision making.”
There is no conversation Greenwood has had with a government agency where the quality and structure of the data would completely prevent progressing an AI orientated project.
“It may require effort, but in any analytics project 80 per cent of the effort goes into data management,” he said. “That has been true since the day dot of analytics and continues to be true today.”
Focusing on getting data “perfect” however could mean over-investing in data storage, extraction and management systems – not necessarily critical with AI.
“The return on the investment only comes through analysis,” Greenwood said. “As soon as they start the analysis it will, inevitably, highlight potential improvements and changes that could be made to how data is managed.”
But to support an AI rich organisation, all staff will need to have a broad understanding of data analysis to ask the right questions. And a culture to embrace innovation.
The barriers to AI
Being aware of the barriers today is important in building decision-making models that can utilise AI effectively. But the barriers are not technical. They are social and cultural.
A key barrier to AI, Greenwood believes, is transparency. Without it, there is lack of confidence in the intelligence process and the outcomes it produces. It also creates a risk that sources of data from the public may dry up.
“We need to get data – and we’ll get as much as we want as long as those providing the data trust us to do so,” Greenwood said. “If you look at legal frameworks like European General Data Protection Regulation, a conservative use of data is being promoted making it more incumbent on decision makers to make it very evident what they are doing with this data — make it transparent and help people to see how data collection and analysis is beneficial to them.”
With AI commonly using personal data to generate insights, government agencies can be concerned about the level of information they can release without risking data breaches. But the methodology behind AI processes and the benefits it creates for stakeholders — including the government and public — need to be clearly communicated to build trust.
Several recent innovations in the application of AI technology have been focused on methods to embed transparency to create greater confidence in the processes used to generate intelligence. Combined with effective and open communications from government, the public and other key stakeholders will have a strong buy-in and become owners of the intelligence process.
AI does not bring all the answers — but it helps to shape them
AI is the next step in advanced analytics and will become more commonplace in government as the data sources used to support it become more accessible – including sensors on mobile phones that have the ability to for example generate insights on the location of potholes to improve road services. AI will be embedded in a range of government processes that will enable citizens to have smoother interactions with all government agencies that touch their lives. It will help traffic lights operate smoothly and support the delivery of faster payments with less paperwork.
“The future will be more and more data delivered at a faster rate,” Greenwood said. “Some of this is driven by the capacity to store all of this information, and some by the desire to analyse it. The more interesting part will come from the joining together of the IoT world – devices, pocket sensors and more to collect the data we are generating en masse.”
Despite AIs ability to sift through data for better insights, Greenwood warned it was important to not get too caught up in the hype that surrounds AI or set excessively high expectations of what it will achieve.
“It is easy to overstate it and think that AI is the tool you install and it solves all of your problems,” Greenwood said. “A human element needs to be overlaid to bring any of this analysis to life. AI simplifies the task of identifying patterns. It helps shape the data. But machine learning does not provide all the answers. There are always exceptions and it is people augmented by AI that will support better government decision-making.”