This week, a client came to me with a detailed loan structure recommendation - courtesy of ChatGPT.
It was actually quite impressive.
Well researched, detailed, and incorporated information he’d given it around some very aggressive short term goals.
The problem was that it was totally wrong for him.
The AI didn’t know a range of things about him that he wouldn’t have known were relevant details.
He had a lower deposit, which meant he was already going to be paying a higher margin on his rate.
It didn’t know to look across a range of lender products, and also didn’t have any detail of bank policies that aren’t publicly available.
It overestimated his ability to make repayments, based on a rigid budget which didn’t include a lot of common things.
And it unilaterally recommended a large revolving credit facility - a structure that wasn’t only unavailable in his situation, but would have actually cost him significantly more in interest over time, not less.
We had a chat, I explained how we could make some changes, and what popped out was a plan that would meet his goals but would also allow for some flexibility and additional payments to be made, just with a lower interest impact.
I hear quite a lot about how “AI will take over our roles as advisers” and while I never say never, I think there’s a lot of context to be mindful of here.
To be clear, I’m not saying we shouldn’t use it, just that we should use it wisely, and with a genuine understanding of what might not be taken into consideration.
So where does AI actually help?
I use AI tools myself, and they’re genuinely useful for:
Running calculations and working out cash flow surpluses - much faster than my wee brain!
Building out a budget and identifying expenses you might have missed (with the correct prompts of course)
Sense-checking your thinking (sometimes their ideas are great)
Generating questions to ask your adviser (lots of us actually really enjoy the challenge to our thinking, too!)
Understanding what different loan structures are and what situations they are useful in (floating vs fixed is a common one)
Many of the online AI tools are awesome.
They provide really useful information, quickly, and they often ask good questions.
But, save for a very select few, they don’t have the same level of context that a mortgage adviser has.
Nuances around goals, property type, income type, loan structures, policy, and product are numerous, and they make a huge difference to the lender that might be the right fit in each situation.
Overlay that with constant changes to interest rates and cashbacks, offerings that are often not publicly advertised, and we’ve got a melting pot of context that could be missing.
Bank policy is still fairly tightly held
Every lender in New Zealand has a credit policy that guides their decision making, and that policy is not publicly available (for numerous reasons of course!)
Bank staff and accredited mortgage advisers have access to and knowledge of that policy.
So when AI tools give lending advice, there is generally no credit policy context, and in most situations that has a significant impact.
On top of that, policy frequently changes, and one that is quite common is the test servicing rate. Online calculators can also miss this, and give people “pre-approvals” that are no longer connected to lending policy requirements.
And, the difference between an “okay” loan structure and one that actually saves you money can often be due to changes in policy or bank appetite for your income type, your deposit tier, what flexibility they’ll allow on your loan structure and the type of product that suits you best.
Your goals are hard for AI to genuinely understand
I have conversations with clients a lot who say ‘oh I know this goal isn’t a great one because I might lose money but…’
This goes for anything non-financial, taking a pay cut to get flexibility in your role, starting a business despite being on a great salary, or one parent staying at home to look after children.
These are all lifestyle decisions and are yours to make.
Often it’s hard for an AI to balance conflicting goals. For example; something non-financial like dropping income with something financial, like paying off your mortgage as quickly as possible.
Both of these could still exist together, just that the latter would likely take a bit longer or there may be a redirection of expenses.
And while I’m not an advocate for budget leaking (spending more than you’re planning to on certain things, and not accounting for others), the reality is that it does happen, and that’s also hard to account for.
Most mortgage advisers have experienced similar things to their clients - balancing goals, working to achieve them, and often making mistakes, too.
That knowledge and context is also valuable when giving advice.
Context is everything, even in mortgage advice
I’m sure there will be people who feed every detail of their lives into AI and get some genuinely helpful outputs.
That’s great, and anything that moves you forward financially is awesome.
But the thing about mortgage lending and aligned advice to your goals is that you don’t know what you don’t know.
You might not realise which details matter, and which are less relevant, or what questions to ask or prompts to give.
Remember that AI is only as good as its context and if some of that is missing, we all know what happens - it makes things up!
Think of an adviser as a co-pilot. Not to replace your thinking, but to overlay it with context that isn’t available anywhere else. To bounce ideas off.
So - use AI as a starting point, absolutely.
Use an adviser to get the big decisions right.
