Data Story
Introduction
People love being vocal and opinionated about government services. And not just the services, but the way their responsibilities are delivered within the context of the individual.
- "Why can't the government fix all these potholes in our roads?"
- "How is it that the government can spend $5 billion on domestic violence prevention programs and I can still hear my neighbours at each other every night?"
- "That developer is cutting down all those beautiful trees. That can't be legal, can it?"
- "My kids want to join the local soccer team but I can't afford the club fees with the cost of food these days. Doesn't the government fund sports clubs?"
Most of the time though these complaints just boil down to: This is not acceptable!
And you know what? Most of the time; it isn't.
The problem on the other side (the government side), it that they know all these types of complaints are either unfounded, misdirected, or so hard to provide a non-specific (static) answer to - or even suggest a responsible party for - that it all just becomes overwhelming to those tasked with managing government advice. Even though websites are able to handle a large number of visitors and provide an enormous volume of information, given the depth and breadth of government services, the one thing the complainant needs becomes hidden amongst less relevant pages, much like the proverbial needle in a haystack.
When searching on either Google or the site itself, the results may indeed find many results that appear to be the needle, however it is often a service description or indexed content page with no actionable items (buttons to click). "Googling" also assumes some knowledge of how to do an advanced Google search; not much use to a voice only user with a learning disability. To compound the problem every government department is directed to navigate the requirements for search engine optimisation (SEO) algorithms, as are hundreds if not thousands of other government services all over the world doing the same. It is understandable that this leads to a lot of frustration for the general public.
There is however someone - or rather something - that has effectively read the entirety of the published internet, and has been trained to arrange words that appear to answer many of those questions. It just needs a little help and a gentle prompt to set it in the right direction. The next problem however is that we are left with the same conundrum as before: how does a general member of the public craft a prompt that will elicit the right response from a Large Language Model?
Proposed Solution
TINA is designed to leverage generative language models (AI) to improve the accessibility of Australian Government services at all levels, and in responding to civic enquiries, better craft the communication of public servants back to constituents in a manner that respects their privacy, demographics and circumstances while observing the Australian Government Style Manual.
By delivering more accessible and inclusive communication mediums, TINA aims to enhance the quality of government services, foster trust between the government and its constituents, and contribute to a more inclusive and equitable civic society with greater participation and engagement in democratic processes.