Automation, AI and machine learning in the public sector

In this guest article, public sector lead at Contino, Alistair Smith, explains why automation and AI can contribute, positively, to governmental reform, why the debate around them is so polarised, and what can be done

“GDS, which became a model for other governments to follow, including in the USA and Australia, is becoming sidelined and underpowered. The powerful and revolutionary idea of ‘government as a platform’ is dead.” Those words were spoken by Francis Maude during his lecture on the Future of the Civil Service, in September 2017.

Disillusioned with the current state of the ‘innovation-hostile’ Civil Service, Francis Maude’s speech, quoted above, outlined a series of ideas for reform, spearheaded by cultural change and innovation. According to Maude, the Civil Service has a “bias to inertia” that needs to be replaced by a “bias to action”.

As citizens increasingly consume government services via digital channels, any civil service action or reform must have digital innovation at its core. And here we run into a problem: digital services will be sidelined in governmental practice for as long as people feel threatened that they will be replaced by automation brought about by innovations in artificial intelligence and machine learning.

Digital reform necessitates a level of readiness which departments might not originally have been in. In driving the transformation, GDS has been forced into a position in which it is biting the hand that feeds it, which will block the most meaningful route forward for Civil Service reform: organisational adoption of digital service thinking – changing the way citizen services are delivered and the role the Civil Service has in that delivery.


The need for automation

Government is a process-focused beast, by nature. This makes it resource-intensive as people and data pass through a multitude of process checkpoints, each of which has to be properly manned and attended to. Efficiency reduces the resource-intensity of these processes, allowing for a less bureaucratic, less wasteful government that can deliver better services, quicker.

Automation is a key driver of efficiency, which means that the government can makes best use of the skills and people it has at its disposal. The output will improve, which is the point of government in the first place.

Done correctly, automation is a gateway to higher-quality government services at lower cost. There will be attrition, but unless the efficiency of automation is embraced, the government will stagnate as the efficiency of the rest of the world increases around it.

But how much attrition? Are the fears justified? Let’s have a look at the debates around automation to see why there’s so much resistance.


The great automation debate

To start with, topics such as automation, artificial intelligence and machine learning are all the rage, but the debates around them are extremely polarised. On the one hand, proponents point to the coming AI utopia, while critics warn of the potential for mass layoffs and the creation of a useless underclass.

But it’s not as black and white as this.

Even for the blue-sky thinkers, there will always be a large degree of work that cannot be automated. Coupled with the fact that progress will be slow, particularly in government, both for the sake of due diligence, and so as not to overwhelm civil servants with incessant, rapid change. And the fact that creating change in government is a long, difficult process (partly for the excellent reason that we shouldn’t be able to change how we run the country overnight because massive instability would result).

For those who fear unemployment, automation doesn’t just directly correlate to lost jobs. Parts of a job are automated, not entire roles. Nor is it just about removing low-skill, manual work. It’s more intelligent than that. It removes a lot of bureaucracy from government (including people in middle management), focusing on those mundane, highly-repetitive tasks that add the least value.

Taking a wider look at the history of IT in government, it is much more justifiable to make noise about the vast sums of taxpayer money that has been routinely funnelled into legacy system integrators over the past few decades, which has resulted in failed project after failed project. Yes, these projects create jobs, but these are jobs that make government services as a whole worse, which is surely not the aim.

Government services will always require people who can make judgements based on the human values that underpin government policy. But, given that citizens increasingly consume government services digitally, automation is a necessary step in maintaining those services at acceptable levels, given the nature of digital ecosystems (i.e. highly dependent on repeatable tasks that it would be ludicrous to attempt to carry out manually).


The solution? Reskilling

Where we once required thousands of civil servants to perform the repeatable, manual tasks while simultaneously handing over wads of cash to system integrators that thrive on creating long-term dependencies, that’s no longer the case. Just as the internet changed how we can interact with citizens and changed the civil service workforce, so too will automation that is being driven by AI and machine learning.

Resistance from public sector workers can be expected, but arming them with the facts – AI and automation won’t steal their jobs – and providing them with a path to grow their skills, can allay fears and build a stronger, happier and more engaged worker. Pandering to the most vocal who are resistant to change simply reinforces the “bias to inertia” that Francis Maude warns us about.

Governmental skills should be reoriented to match the society-wide trend towards digitalisation. Departments need to consider how they can best deliver digital services, facilitate pan-departmental data exchange and so on. How can civil servants add value to digital policies and differentiate their implementation?

By withdrawing from toxic vendor relationships with large SIs, increasing efficiency through automation and upskilling as many of the surplus human resources, important steps would be taken towards government-as-a-service.

Related reading