5 uses cases for ChatGPT in local government
Dall-E: Happy Council workers using artificial intelligence to make community service more efficient

5 uses cases for ChatGPT in local government

G'day everyone!

Despite being back in the private sector for nearly two years, local government remains a passion of mine which I'm able to fulfil in my role at sensen.ai servicing local government clients and accounts.

Over the last week or so I had the chance to talk both on the phone and face-to-face with ten or so local government people from a few jurisdictions around Australia.

The topic: the use of ChatGPT.

Firstly, if you have not yet come across ChatGPT you are reading the wrong article - why don't you try the excellent Introducing ChatGPT blog post and give the tool a try?

With that out of the way, on with the results of these discussions - which as you can expect, were varied.

Some local government folk had not come across ChatGPT at all and were astounded by a brief demonstration or explanation and wanted me to hang up or leave so they could give it a try on their own.

Some local government folk were using it for idea and content generation as I was.

And, some local government folk helped expand my knowledge of what ChatGPT can do giving me some insight into what are, for me, advanced uses of it.

As briefly as possible, I will share what people were generous enough to share with me:

1. Making it quicker to write a business case

Spending community funds requires rigor and good record keeping. Writing business cases and recommendations takes time and this time can get in the way of making the right investment at the right time.

One local government associate described using ChatGPT to write the core of a business case considering the use of particular technology in a specific branch of the Council.

ChatGPT listed advantages and disadvantages ranging from officer safety and transparency through to privacy and cost.

It was even able to compare two leading products at a technical level showing their strengths and weaknesses against each other.

I was able to easily reproduce this experience using ChatGPT and can see how much time this can save.

2. Grading a position description

Creating a new position and position description carries an inherent level of risk. In particular, getting the grading incorrect can lead to budget overruns and/or unhappy employees once the role is filled.

One associate reported that their HR leader was using ChatGPT to check position description grading.

Being a very regular user of ChatGPT I was curious about how this might work and I spent some time trying this out. It turns out I was able to achieve this and it was fairly useful.

Its a bit of a three step process, firstly I pasted in the position description.

Then I pasted in some grading guidelines for the appropriate award.

Lastly, I asked ChatGPT to grade the position description and it produced some prose explaining where the position fit in the grading scale. A snippet of the text was "The description also highlights the need for leadership, innovation, and strategic thinking, which are consistent with the expectations for this level".

3. Answering a question from a member the public

When I worked at one Council my landline number was one keypress different from the switchboard number. This means I fielded my fair share of calls from members of the public asking questions such as when the basketball courts were open.

I could handle these easily without transferring the caller by just using Google to search for their question and making sure I used information from the Council website.

In 2023 ChatGPT and the surrounding and supporting technology can automate this.

One of the folk I spoke to created an experimental email bot they could forward website enquiries to. This would use a modified version of the OpenAI engine (which ChatGPT uses) that favoured content from the Council's website.

They were successfully able to write email replies to enquires from members of the public that made sense.

I received access to this bot and was able to send many test questions to it.

I found most of the replies were accurate but needed some rewriting and rejigging before I would have considered sending them to a customer.

This person is still experimenting with this email bot and considering what it would take to get it ready for production.

4. Quickly getting started on something

In a more traditional use case, most former colleagues and customers reported using ChatGPT to just quickly get started on something.

Some quick snippets that I can recall from my discussions include:

A policy for employee reimbursement. A SWMS for installing furniture into a park. Some guidelines for updating an internal policy. A draft internal policy. An email to the team reminding them that customer experience is important despite the fact they are not front-line customer facing officers.

In my role I find using ChatGPT like this is the main use case and this sort of "get me up and going with a draft" the biggest timesaver.

5. Advising the Council

One contact was using ChatGPT to help them provide written advice to Council about the pros and cons of community engagement. In this Council a decision was held up, as the Council believed there needed to be more community engagement and were seeking the staff's input on this.

The officer was looking for very clear language to explain to Council the pros and cons of performing repeated rounds of community engagement.

To replicate this, I tried a few different questions to ChatGPT about the use of community engagement to inform local government decisions.

It was able to regurgitate reams and reams of information which seemed quite technical and academic and not a very useful way to communicate to governance bodies such as a Council.

Asking for a briefer version of this information ended up with what, to me anyway, a decent list of pros and cons for conducting a second round of community engagement to guide a decision.

Considerations

Overall folk using ChatGPT were very pleased to have a tool that saved their time and let them get things done more quickly, or reduce their risk.

Although better covered elsewhere some considerations sprang to mind as I spoke to people and drafted this article.

Firstly, ChatGPT is largely unattributable. Asking for a reference for text it just produced is met with the familiar I'm sorry, as an AI language model, I don't have access to current references or information. In an evidence-based environment such as local government, officers will need to in many cases seek supporting information if they are using ChatGPT to draft assertions.

Secondly, ChatGPT could get it wrong or have biases. Using it to compare products seems likely to be biased toward which product has better or clearer or more discoverable marketing material online which has been included in ChatGPT.

Thirdly, ChatGPT can't make me an expert. If I am not a subject-matter expert in, say, community engagement, I should find one before I present information from ChatGPT as expert advice about community engagement.

This leads to the overall consideration, accountability. With possible built-in biases and lack of clear references, its unclear who is accountable for decisions based on content created by ChatGPT.

Tools like ChatGPT are another step in the path to avoid repetitive monotonous work that wastes human endeavours, and seeing the uptake in my local government network is very exciting.

Lastly the obvious question - did ChatGPT write this article? Sadly, No, although not for lack of trying.

I tried "top down", that is, feeding in points for each of the five use cases and found the resulting text missed the point.

Then I tried "bottom up", that is, feeding in hastily written fully fledged text and found the re-written results either missed the point or didn't sound like me at all!

Unwilling to lose my voice to ChatGPT just yet, I present this handwritten and edited article - all mistakes mine!

SenSen

sensen.ai is an Australian-founded global data fusion enterprise working in the sensor AI space supporting compliance and citizen well-being in many every-day situations as diverse as public beaches, airport terminals, parking, waste and casinos.

Please reach out to any of our amazing people if you would like to learn more how SenSen can help you make what seems impossible possible.

Prof. Subhash Challa

Positively Transform People's Lives with AI

11 个月

great article

回复
Diana Schreiber, MPA

Clerk and administrator in local government.

1 年

Are there suggestions as to what AI applications clerks can use for transcription in minute writing?

Claudia Feiloivao

Business Improvement ?? | Facilitator and Coach of Corporate Innovation ?? | Intrapreneur??| Bible Teacher ?? | Artist ????

1 年

Fascinating! I must create an account and give ChatGPT a crack! I enjoyed reading your article Nathan and especially your writing style, very personable and informative. It’s interesting to see how local governments are using ChatGPT, and I think it’ll be helpful personally, especially with ideas to just ‘get started’ with business cases, emails and the like.

Jennifer Moody

Advising stakeholder relations and community engagement for major infrastructure projects

1 年

Love this Nathan! Interestionly, my husband was asked to marry friends of ours. With no idea where to start, he used ChatGPT to plan out a script. I also played around and used it to pull together a business case from scratch. Really looking forward to seeing what else it can do, if nothing else - it’s fun!

Adam Vine

Principal ERM at Metropolis Advisory

1 年

Well done Nathan Rogers , superb article. Balanced and the right mix of tactics and detail. Thankyou!

要查看或添加评论,请登录

社区洞察

其他会员也浏览了