AI will help LGAs do more with less — provided they have support to get started

When floods swept through the Northern Rivers area of NSW in 2021, distressed, stranded and strapped residents turned to their local council for help.

The same occurred in Victoria during the Black Summer bushfires, and with Western Australia’s Shire of Northampton and surrounding councils, in the wake of Cyclone Seroja.

And when COVID first sent shock waves across the economy, it was local councils who commissioned small business advice and community support, sponsored pop-up stores and enabled kerbside trading.

In fact, whether it is natural disasters, health emergencies or just a gap in the services that everyday Australians believe are important to how they live their lives, local governments are the natural place people turn.

Ask Australians what they think of local government, and they say it is the best level of government to be making decisions about the area where they live.

But continuing to deliver and grow these services is an increasing challenge.

They raise just 3.8 per cent of the total tax take in Australia, but manage 33 per cent of public non-financial assets. They look after 75 per cent of the country’s roads.

At the same time, nine in 10 local governments are experiencing a skills shortage, up from six in 10 four years ago.

And nearly one in three council workers in Australia is either nearing — or has already passed — retirement age.

What could artificial intelligence do to help solve this problem?

It’s the question we were called to answer at the recent WA Local Government Association (WALGA) State Conference, to consider how generative AI might help local governments do more — but also, do so with less.

The conference, with the theme Local Futures, had heard from several speakers about the recognised importance of local government, and the ongoing pressure to deliver a broader range of services.

So, let’s start with what ‘more’ looks like for local government.

What if local governments could use AI to handle routine enquiries, freeing up humans to focus on complex or emotionally sensitive tasks?

In a crisis, what if that AI could automatically sort and prioritise incoming requests, so they could address urgent needs faster?

When it comes to drafting policy, what if generative AI could do first drafts based on a set of guidelines, reducing the administrative burden?

How about better engagement: What if AI could analyse community sentiment in real-time, enabling governments to adapt communication on the fly?

What if it could personalise public services for each resident, in their preferred language, with information and community program recommendations?

And what if AI could assist in revenue generation for LGAs by identifying the most valuable and ethical ways to utilise community data?

All these are possible, right now.

While this sounds like a long wish list, much of this is doable — today — through generative AI tools such as ChatGPT … provided local governments know how to use the system safely.

We can see this in other jurisdictions, such as the City of Boston, which has encouraged everyone in their teams to experiment with generative AI, to learn how to use the tools and to get to understand where the frontiers lie in what it can and can’t do.

Some of the simple use-cases they have explored include using it to write a memo or job description, summarise citizen feedback, translate material into multiple languages, rewriting documentation to be more effective, and redrafting legalistic content in Plain Language to make it easier to understand.

At the same time, Boston’s guardrails have also been freely made available.

These include:

• Don’t include confidential material in the prompt
• Don’t use AI to respond to sensitive issues (such as a press release on a shooting)
• Don’t ignore inclusivity, accuracy, and real-world testing of outputs
• Don’t take shortcuts (such as making decisions based on a summary)
• And do require transparency about use (such as ‘this text was summarised by Google Bard’).

LGAs (Local Government Areas) will need support to take the next step.

WA local governments are not quite at this stage yet, but the mood at the WALGA conference underscored the high level of interest in investigating AI further.

Some of the questions they will need to consider will go to the heart of how responsibly they can introduce use of the tool.

This means making a call on things such as the transparency they will require (from their people and their suppliers), and how they will reduce the risk of bias or misrepresentation, and avoid breaches of privacy or exposure of commercially sensitive material.

They will also need to ensure decisions made by or with AI are able to be justified and explained, not just left in the hands of a black box.

Inevitably, there will be internal change, both as people train up in delivering additional projects and as roles that were purely human are ceded to the indefatigable AI.

Managing that change and ensuring humans are freed to do the things they do best will be a critical piece of work for every LGA.

The opportunities for local governments to use AI to scale and extend their services — as well as reduce the time spent on business as usual — are enormous.

It’s now time to make sure local government teams are equipped with the skills they need for an AI future.