How an economist uses AI to work more efficiently
Originally published on Substack by Rashida Tayabali
This piece was first written by Rashida Tayabali, who interviewed me for her Substack earlier last month. I’m sharing it here because she captured so well my reflections on how to best use AI in strategy. I wanted to bring it onto my blog so more of you can access this perspective and, I hope, take away practical insights for your own context.
How an economist uses AI to work more efficiently
Founder of Omni Strategic Consultancy, Ariadne (Ari) Magalhaes FGIA GAICD guides ASX boards, NFPs and purpose‑driven organisations through complexity, strategy and execution. I reached out to Ari because I was interested to learn how she uses AI to deliver complex, strategic projects for her clients.
Tell me about yourself and the work you do
I'm a Non-Executive Director and Economist. I'm an advisor for boards, CEOs and executives. I come in when the organisation is going through some sort of pressure, like financial funding pressure, or regulatory changes. I support them in turning things around or redesigning their model or their strategy to make sure they are sustainable.
What were your initial thoughts on AI and some early experiences?
I was really impressed with the potential. It was hard to believe that things that would take me hours or days were ready in seconds.
At first, I was really unsure about the privacy matters - around what you share, how it's used, and the challenges in that space. I was keen, but I was very careful around how I used it and continue to be.
I love Google’s NotebookLM. It’s free and I use it with public documents. When I want to do a deep dive on a topic, I just upload all the information that's publicly available, for example, around the New Age Care Act, and try to understand the implications by asking it questions.
I'll ask about the context. For example, in context X, Y, Z, how could this be applied? Something that used to take me two days is done in 20 or 30 minutes.
I have a teams subscription to ChatGPT because at enterprise level they don't use the data to train the model. You have a higher level of privacy, unlike the free and Plus versions of ChatGPT.
I’ve created custom GPTs based on their purposes. I have a strategist, marketer, business analyst, board strategist GPTs that I use daily.
Before AI, I spent a lot of my time researching, reading and collecting information. Now I can do it much faster. And if there's specific information that I need from a document, I can ask ChatGPT to tell me where it is, and I'll go straight to it.
What type of guardrails have you put in place so that your expertise doesn't get eroded?
I see AI as a tool to augment my work, not replace my expertise. It's there to enhance outputs and outcomes. Before I use any tool, I decide on a few things: the output I want, the outcome, and what my thoughts are around it.
I also read through the documents. I have some ideas because I’ve had meetings, discussions and from previous experience. From this, I create an initial draft.
Then I ask ChatGPT (or sometimes not at all), to provide additional input or restructure and enhance what I have. I'll go back to it, but I'll never start from a blank page and ask it to do it for me.
I have an approach, a framework that I developed based on best practices, after having worked across eighteen industries. So I adhere to my approach and framework first and use Gen AI to improve the output so I can deliver with the same time and investment from the client. And the client gets more valuable outcomes
What are your top three tips for strategists who are just getting started with AI?
Treat AI as a thinking partner, not replacing thinking
See it as an additional input, not something that will do the work for you, but something that will enhance what you're doing. It's important not to hand over formal advice or recommendations or potential directions to Artificial Intelligence. That’s something you should be developing. It's your name on the project, you need to have ownership.
Build before you borrow
Develop your hypothesis and structure first.
Ask yourself, “First, what do I want? What’s the outcome?” You’ll get better results and better outputs. Recently, I attended a conference, and one of the AI specialists said you really have to be a specialist in the subject to use AI, because you have to be able to critically assess what it's being produced.
Sometimes I’ll produce something that sounds really nice, but it makes absolutely no sense. I’ve used ChatGPT and other tools to create things that sound great, but they are not logical. It’s not something that a human would say or would do. You have to be an expert in the area to use it well.Focus on the outcome, not the output.
Ask yourself: What am I trying to do here? What's the value that I'm trying to generate? And how can I get AI to enhance it?
What challenges are you seeing in organisations when it comes to AI implementation?
The main challenge is not thinking about objectives first. So they’re thinking, ‘AI is great. I want to use AI. What tool is out there?’ But that's not best practice.
What you really should think first is what are my strategic objectives, and can I use technology in general to support and drive delivery? And within the theme of technology, how can I use AI to do that? So think about the objectives first, and not the technology first, or tools first.
If you're competing with someone that's using AI and therefore his or her intelligence is multiplied by 50, then they'll eventually win. So it is important to use it. You’ll fall behind if you don’t. But you have to do it the right way and be very mindful of the risks as well.
What's the future for strategists with AI in the mix? Do you think it's going to take your job?
I don't think it’ll take my job but strategists, consultants and advisors who don't use AI will eventually become redundant. It’ll become more challenging to enter the consulting world as a junior, because a lot of it can now be done by AI. But you still need that reasoning and thinking, and the real human view of considering different variables and subjective variables.
There are always other things at play that we strategists don't write down, so ChatGPT wouldn't pick it up. We also have to be mindful of the bias because generative AI is built on our biases. We cannot rely on AI and use it as a single source of truth because it has biases in it.
Do you formally disclose to clients that you use AI in your work?
AI is part of my toolkit.
I have a clause in my contract that mentions I'll use AI but it’s just another tool for me, like accessing IBIS reports. I'm not replacing myself and my expertise. I'm enhancing the outcomes.
What's the best way that a strategist can get started with AI?
Start using it every day in different things. Test different tools. I've probably tested around 300 tools. Not even exaggerating!
I’ve signed up for various newsletters where I look at what the latest tool is. If you use it every day, you get better at it right? The theory of 10,000 hours! Get started, but at first, don’t share confidential information. Do it for mundane, simpler tasks, and learn from it. Create an account on any of the platforms and just use it.
