ChatGPT’s Coming for You!

So embrace it today.

Recently, some of my colleagues in various government policy divisions have been asking me how they should be responding to this whole AI wave that’s been seizing the world:

• Have you heard? We are coming up with Pair. Soon our jobs of writing will be obsolete! (see link)
• Does that mean our jobs of policy analysis and horizon scanning will soon be replaced by AI?
• You mean policy reviews can be done by AI? Wait, you mean even writing minutes and functional emails can be done by AI?

Presumably, they think that just because I am in GovTech, I should have all the answers. Well, I don’t.

I do have a perspective, however.

Understand Your True Role

Let’s look at the same issue in the tech sector. Should software engineers be worried that ChatGPT will replace them eventually in writing consistent code?

Well, the reality is that ChatGPT will definitely be better in writing consistent code — how’s a human to compete with a machine in consistency and standards? Machines were designed for consistency and reliability, not so much humans with all their caprices.

But you see, that’s not the true role of a software engineer. (If you were such a software engineer archetype i.e. one who simply writes code, well, then perhaps you should be afraid.)

Fundamentally, a software engineer is a problem solver. In the context of his work, his primary value-add is to translate business requirements into solutions, and then into software requirements (i.e. what we call code). That he happens to know how to write code and understands system architecture is just a convenient matter of the situation.

In fact, I’ve spoken to many of my software engineers, and here’s what they have to say about their job:

• “Yong Kiat, the best solution is always a no-code solution. Because a code-free solution is bug-free! So that’s why we don’t just keep thinking of what new features to code in order to solve a problem. ”
• “I don’t know really know how to describe it, but tech-driven harmonisation never works, you know? The best way to solve organisational efficiency issues… is just to solve them upstream first before loading all these burden onto the system.”
• “Hey, you and I, we’re not that different. We’re interested in solving the problem. Just that you solve it conceptually by moving people, dynamics and monies as a policy officer, and well, I try to solve the problem using a software.”
• “Yong Kiat, if I could choose to not write code to solve the problem, I would definitely do that.”

I put it to you that policy officers are (ought to be) the same problem solvers that software engineers are (again, ought to be). In fact, all that we do as policy officers, be it writing that set of minutes or getting that budget paper approved, is to simply to move resources (e.g. people, resources, plans, systems) to solve a common problem (in tech, we say we are swarming the problem — in policy, we say we are working through the system and aligning incentives, but it’s really not that different).

Once we view ourselves in that light, ChatGPT becomes an enabler, not a disrupter.

Dare to Embrace the Tech

ChatGPT really is just another piece of tech, like any other, and we ought to be using it to enable our work. Again, let’s turn to the tech sector to see how they are (ought to be) embracing ChatGPT.

For those who are unitiated to the latest in Generative AI developments, ChatGPT can indeed code. In fact, there’s a service offering known as GitHub Copilot, where OpenAI’s GPT is built into your code editor. This allows you to drastically cut down the amount of coding time and effort because the AI is now able to generate code based on meaningfully-written comments, code structure and even function names (no wonder they all say one of the two most important things in the software world is naming things).

I totally dig this as a casual coder — imagine all the visualisations and simulations I could now perform for my students!

Are software engineers shunning this new and shiny tool? You bet they aren’t — many are already thinking about how to include GitHub Copilot into their routine workflows. In fact, AI allows the software engineer to focus more on how to translate those pesky business requirements and broad business problem statements into actual solutions, by saving them the time in further translating them into software requirements.

(Now that I’ve articulated this, Business Analysts should be wary that software engineers may well be coming after their jobs — fear not ChatGPT, but the problem solvers!)

And so it is the case for policy officers. Or anyone else who sees their role as solving meaningful problems. Need to draft that coherent set of minutes in order to rally the sector? Require that consistency and accountability in procurement and tender specifications in order to obtain the right service for our citizens? AI isn’t going to tell you what problems are worth solving — you tell AI what problems you want to solve, and use it to cut down the work you shouldn’t have needed to do if it was an ideal world.

And………

……

…

I’m sorry, I don't know how to discuss this topic. 
You can try learning more about it on bing.com.
How may I assist you today?

(for those who don’t understand the running joke at the end of this article: https://www.nytimes.com/2023/02/16/technology/bing-chatbot-microsoft-chatgpt.html)