Building a data strategy for AI-powered Government

By Wing Leong Ho

Citizens shouldn’t have to figure out the intricacies of government to get their fundamental needs met. Could there be a form of anticipatory government which does this for them?

It would never occur to someone who has just lost a loved one in Singapore that they must go to the National Environment Agency to take the next steps.

That is how it must be done right now, but this is set to change. Singapore, along with many other countries, are redesigning services around the biggest milestones in life: getting a job, starting a family, opening up a business, or losing a loved one.

The best form of public service is when citizens can get it done quickly and effortlessly. Citizens shouldn’t have to figure out the intricacies of government to get their fundamental needs met. Could there be a form of anticipatory government which does this for them?
 

AI in government


Three key principles are guiding how governments think about this: be predictive and proactive; serve citizens on whatever platform suits them best; and adapt the experiences to meet their unique circumstances.

A number of countries have already begun this journey - Singapore’s Moments of Life app and New Zealand’s SmartStart and End of Life services are examples. Elsewhere, Estonia has taken this a step further with a vision for AI-enabled government. Their strategy is to allow citizens to be served on any device using virtual assistants.

Voice is the most intuitive way for most people to communicate, Estonia’s new AI strategy says. AI assistants would remind you when it’s time to renew your passport or talk you through welfare schemes as you start a family. “The whole complexity and rigidity of the public sector can be moved to the background then,” it says.

Taiwan’s Digital Minister has led the use of AI to tackle complex challenges like social cohesion and climate change. It uses algorithms to crowdsource opinions and ideas, and collectively shape policies on critical issues. It is trialling machine learning to ensure utilities are not disrupted by natural disasters, and has become the first jurisdiction to legalise experiments with autonomous vehicles that can drive, fly and swim.

In Britain, the next phase of GOV.UK services will also take this approach. It will move away from thinking of itself as simply a website and instead, inform users of what they need to do and guide them through complex life events, writes the head of Gov.UK.
 

Powered by data


A strong data strategy is crucial to make this vision of government work. Behind the scenes, platforms must allow agencies to share and consolidate information without breaking security and privacy rules. They must be able to do so at scale and in split seconds, to provide information to citizens whenever and wherever they need it. This is something that Cloudera specialises in and has worked with agencies globally on, from climate modelling to uncovering fraud.

The strategy for this can be broken into three building blocks. First, what happens when governments receive new data? It must be moved in the most efficient way possible from the device - which could be a smartphone, sensor, website or telephone call - to the next stage. Every piece of information has policies attached to it on how it can be used, how long it can be stored, and who can access it. Cloudera’s technologies allow agencies to ensure they always respect these boundaries while moving information.

Next, it should be taken to a large repository - a data lake - from where it can be accessed and analysed. A data lake is able to receive and process petabytes of data in split seconds. It continues to ensure that privacy and security policies are maintained as agencies dip in and out. If someone without the authority to view personal details accesses a file, they will only see masked data.

Governments have a number of options today for how they choose to store this data. More sensitive information is kept on government-owned property, while others can be stored in the cloud. It’s likely to use a hybrid approach with a combination of both and a number of cloud providers. Cloudera’s data platforms are vendor agnostic and ensure that data can be seamlessly moved from one place to another, and its security and governance rules follow wherever it goes.

The final stage is to apply machine learning on the data. This requires huge amounts of computing power that governments may not have themselves. They can bump this up by adding cloud resources, process the data on the cloud and then move it back to agency servers if needed. The algorithms created through this process can then be reused on new data as it comes so the model is continuously learning.

These three steps will allow governments to join up agencies and build powerful tech that delivers services proactively, works with any form of engagement and can adapt rapidly.

Wing Leong Ho is ASEAN Technical Lead at Cloudera.