Opinion
Retirement planning is undergoing a powerful revolution driven by AI, and not just in the office of financial advisers, who are adopting it to speed up their businesses and improve their margins. It’s happening around the kitchen table, on the couch late at night, and in bed too.
More than 65 per cent of retirees in my community admitted this week to using it to ask questions, and 38 per cent admit they’ve used it to ask questions and run calculations.
Those two numbers tell us all something important: that people are hungry for accessible, affordable and helpful answers, and they’re going to find ways to get them with or without the industry’s blessing.
The financial services industry has not entirely welcomed this shift. If you’ve spent any time in adviser circles, you’ll sense why: for the first time, the knowledge gap that made professional advice feel essential for people with relatively straightforward situations is starting to close.
AI is, in my opinion, the best thing to happen to retirement planning in 20 years. It’s the only tool I’ve ever seen that has managed to engage everyday people to explore their money keenly, in an easy-to-use, curious manner.
So I’m going to say what many won’t – and that is every Australian heading toward retirement should have AI open on their phone right now, asking it the questions that the system has made it too expensive, too scary and too difficult to answer.
AI is not a replacement for learning how things work from valuable factually correct sources that are trustworthy.
You should ask it how stuff works. Ask it to explain compound interest and inflation until it finally makes sense to you. You should get it to explain the different ways of calculating how much you need and how well your money will last.
And you should ask the questions you’ve been too embarrassed to ask your super fund about whether you’re likely to be eligible for the age pension, and how that might work with your super.
And, when you reach a decision point, throw it this curly one – ask it about who has a vested interest in you making the decisions in front of you, and what’s in it for them, what’s in it for you and whether your interests are aligned, before you buy.
Ask it at 11pm on a Sunday night in your pyjamas, when you don’t look forward to going to work any more. Or ask it on a Thursday, at lunchtime when you’re fed up with meetings. It won’t judge you. It won’t send you a multi-thousand dollar bill, and it will not make you feel like you should already understand how stuff works.
But use it carefully and know where it sits in your toolkit. Because AI is not a replacement for learning how things work from valuable factually correct sources that are trustworthy. It is also not a place you should consider a “secure” place to upload your own personal data. And it should not be the overriding source you take advice from unchecked.
In a recent speech, MIT Sloan finance professor Andrew Lo, who has been researching AI and retirement planning since ChatGPT launched, says AI is genuinely good at explaining trade-offs, exploring scenarios, and behavioural coaching, but remains weak at precise tax optimisation, maths, and regulatory compliance. He says that AI reasons by probability and narrative, rather than algorithmic logic which means its maths can be unreliable. Take heed.
I’ve tested both ChatGPT and Claude extensively. I’ve run retirement projections, modelled drawdown scenarios, built calculators for the sweet spot, and asked AI to calculate how long a super balance would last under different conditions.
More often than not, unless very specifically guided, it comes back with age pension thresholds that are out of date, ASFA benchmarks from a previous year, inflation either missing entirely from the calculation or sitting at a figure that bears little resemblance to what Australians have actually lived through recently.
It looks right. And it’s presented with absolute confidence. So if you don’t already know what those numbers should look like, you would never know to question it.
But, on the upside – if you do know how things work, and you push back, question the inputs it used and work through some back-and-forth to check the numbers against real sources, rather than taking it on faith. It can deliver truly useful information. The problem isn’t the tool – it’s knowing enough to use it well or know when it’s feeding you false information.
Think about it like this. If you ask your AI “will I have enough to retire?” and you don’t know the questions to follow that with, about inflation, the age pension, mandatory drawdown rates or how sequencing risk might affect your retirement if the markets fall in your early years, you might come away with a bold new level of reassurance, but your simple question might have just got an overly simplistic answer.
But if you understand the age pension, superannuation’s mandatory drawdown limits, bucket strategies, sequencing risk, and how safe spending works, you can direct AI and steer the conversation, provide it with up-to-date inputs and push back when something doesn’t add up. You can ask what it left out of the calculations, and fact-check the inputs against the real sources manually.
And in that case, AI can help you understand your options and make better decisions. That’s because unlike anyone in the financial services industry – a financial adviser or super fund adviser – it can debate the options with you, with no product to sell and no ongoing fee-for-service arrangement that might colour its thinking for the rest of your retirement.
With AI in at your fingertips, and plenty of questions on your mind, I want you to use it carefully, as a collaborative partner, not an oracle. Start a project you use for your requirement planning questions so it remembers the work you do together. And think about it like this.
Tell it your scenario parameters like you would share a case study – impersonally. Never enter your real personal details, like your name, date of birth, tax file number, super balance or member number or account details. Use round numbers, non-personal data that’s close enough to be useful but nothing that could expose you if that data was ever compromised.
Then, think about what you could use it for:
- Ask it to teach you the basics of retirement finances and to recommend books, websites, and people you can follow to support your learning journey. Check them out.
- Use it to develop your options, multiple options. Then, ask it what it’s left out of each one and why.
- Ask it to verify that the information it’s using is correct and timely, and provide you the source links so you can sight and review the primary sources of inputs.
- Ask it to tell you why it might be wrong, and to debate your thinking so you can find weaknesses in the scenarios you’ve built or reasons they might not work as planned.
- Ask it to build you a financial model, in Excel and show you the calculations it’s basing its conclusions on. Download the Excel and improve it, mine it, use it to learn.
- Ask it to state the uncertainties and the things you should worry about so you can interrogate them further.
And then, ask it the questions you need to answer and who the most trustworthy source of answers is, whether it’s your super fund, a financial adviser or a government website – and then follow the breadcrumbs to test what you find.
Just don’t be afraid to ask. It’s not the asking that will hurt you, it’s acting on the answers blindly.
Bec Wilson is author of the bestseller How to Have an Epic Retirement and the newly released Prime Time: 27 Lessons for the New Midlife. She writes a weekly newsletter at epicretirement.net and hosts the Prime Time podcast.
- Advice given in this article is general in nature and is not intended to influence readers’ decisions about investing or financial products. They should always seek their own professional advice that takes into account their own personal circumstances before making any financial decisions.
Expert tips on how to save, invest and make the most of your money delivered to your inbox every Sunday. Sign up for our Real Money newsletter.