March of the machines: How governments can benefit from AI
Rapid recent advances mean that intelligent machines will soon be able to replace human workersShare article
AI needs large data sets and while they’re very high cost to create, they’re zero to replicateShare article
Government has yet to fully exploit the power of AI but this is a dam likely to be breachedShare article
I recently flew from Boston to Miami. The journey, thankfully, was uneventful but there's no doubt I was more at risk in the taxi rides to and from the airports than when 36,000 feet up. The autopilot and advanced airline technology offered me the assurance of safety, whereas on the ground I was exposed to the vagaries of drivers and others on the road.
This is just one example of how we all benefit from the power of computers. But what we are accustomed to today will be very different in the future. Artificial intelligence (AI) is poised to transform how we live, work and play. This isn't a new phenomenon - the idea of developing computers that could think like humans stem from 1950s programs - but rapid recent advances mean that intelligent machines will soon be able to replace human workers in all sectors of the economy. Now that's a public impact.
But what does this mean for governments? And how can they respond?
The pursuit of AI has not been without some bumps in the road. Until quite recently, the general approach sought to understand how humans reason and then replicate that in code. However, new techniques were developed in the late 1980s and 1990s for approaching those same questions statistically rather than cognitively. The aim was simply to use statistical methods to identify regularities in the answers that humans come up with. And the most important of these methods is convolutional neural networks.
These networks - which simulate lots of densely interconnected brain cells inside a computer - are also rooted in the late 1980s and 1990s, but they didn't achieve their full potential at the time because they had insufficient data. However, the explosion of big data at low cost, combined with ever-increasing computer power, has led to some extraordinary breakthroughs. For example, computers can now tag images with contents of that image more accurately than humans can - a previously impossible task.
Governments hold a vast amount of data about healthcare outcomes, social security and taxation but, before even considering how they might deploy AI to improve public services, it is important to remember that there are massive economies of scale in the data itself. AI needs very large data sets and while they're very high cost to create, they're zero cost to replicate.
The process of actually taking this data and designing convolutional neural networks to interpret it is something that a team of two or three people can do in a couple of weeks. The most effective way to actually get the problem solved is to have multiple self-organising, autonomous teams all take a crack at solving it.
Government's main advantage is being able to collect, and have visibility across, vast swathes of data about its citizens and the services they depend on. As repositories of these extremely rich data sets, governments can now code this information and release it into the world for interpretation, which should - in theory at least - lead to better services and better citizen outcomes.
Admittedly, this can be problematic. After all, governments have their own business models which are vulnerable to disruption - no government-backed post office will welcome its database being made available to private sector competitors, for example. This dilemma is characteristic of disruptive technologies in general and is one of the reasons government has yet to fully exploit the power of AI. However, the pace of these changes - and their potential to revolutionise services for the better - means this is a dam likely to be breached sooner rather than later.
AI: What's next?
So much for what is possible today - where will this journey lead us in the future? One avenue could be the development of an algorithm that could, for example, help jobseekers back into work by developing individually tailored programmes based on evidence about what has worked in the past.
Such a system could “learn” by introducing small random variations to explore what changes are most likely to improve outcomes. This would involve taking two randomised subsets of the population and making some small variation with one but not with the other, and then comparing the results. By doing this you can very quickly identify whether A is better than B, in which case you might change your procedures, thereby converting it into a continuous learning process. It is a quantum leap from where government systems are now offering the possibility of significant improvements within a matter of days or even hours compared to the months or years offered by today's more cumbersome, formal studies.
This is just one example of how the power of AI and robotics can transform the public impact of governments. Change is coming and governments can get on board or be left behind. Humanity will thank them if they opt for the former.
The Centre for Public Impact is investigating the way in which artificial intelligence (AI) can improve outcomes for citizens.
Are you working in government and interested in how AI applies to your practice? Or are you are an AI practitioner who thinks your tools can have an application in government? If so, please get in touch.
- Transforming technology, transforming government. Rare is the policymaker who doesn't see digital as a doorway for strengthening public services. But as Miguel Carrasco explains, the pace of the digital evolution means there is always more to do
- Power to the people. Few countries have embraced the digital era as successfully as New Zealand. We talk to one of its government's key digital transformation leaders, Richard Foy, about how they've done it.
- Computer says yes. Governments are increasingly reliant on digital technology to deliver public services - and Australia's myGov service is a potential game-changer, says Gary Sterrenberg
- Online, on track? Miguel Carrasco looks at how policymakers can improve the delivery of digital services
- Digital dawn. It may not be obvious, but US policymakers have had an important role to play in the creation of today's digital era. But sometimes it involves stepping back rather than stepping up, suggests David Dean