This photo, in New York, Thursday, May 18, 2023, shows the ChatGPT app on an iPhone. The software can offer directions from building an emergency fund and paying down high-interest debt to creating a budget for a down payment to buy a house. THE CANADIAN PRESS/AP-Richard Drew
TORONTO – As an experiment, Kelley Keehn asked ChatGPT to suggest a financial plan for a small Canadian family.
She prompted the artificial intelligence software to craft a plan for a couple in their 30s who have two young children and want to save for a house. They’re dealing withhefty debt and have zero savings but they’re a high-income household.
“What should they do first?” Keehn, a personal financeeducator,asked ChatGPT 3.
The software offered more than a dozenrecommendations the family could follow – from building an emergency fund to paying down high-interest debt to creating a budget to build up a down payment for a house.
But Keehn said the AI-generated advice lacked specifics for the family to achieve financial stability.
Financial experts say artificial intelligence may be an important tool to democratize financial literacy among Canadians but the software is still not sophisticated enough to interpret the nuances of life and tailor advice to individual financial situations.
“AI is a great resource for the average Canadian seeking financial advice but just not quite with that human element yet,” Keehn said, adding that a financial adviser would’ve had a different take on the case.
“What I think a human would say, is, ‘Look, you’ve got young children, you’ve got debt, so insurance probably should be number one,'” said Keehn.
“Because if that family is left with high debt, there could be a huge gap in looking after those children,” she said.
“If they have a high income, a human (adviser) might recommend a strategy to invest in an RRSP, take that tax refund and use that to pay down high debts.”
A recent RBC poll, conducted by Ipsos, found younger Canadians were much more likely than older demographics to turn to AI to help manage their finances.
David Lewis, president of behavioural science consultancy BEworks Research Institute,says people are more likely to be honest about their financial situation or admit their lack of knowledge to a machine.
“No one likes to turn to another human and say, ‘I’m an idiot and I have no idea how a mutual fund works,'” he said.
“We’re more willing to be honest with the computer because we don’t view a computer as judging us, but we view humans as a judge,” he said.
It is also less expensive than consulting a human financial adviser, Lewis said.
“For those starting out or in lower socio-economic classes and don’t have as much in assets to manage, AI advice can help have better financial outcomes,” he said.
Keehn said AI is “brilliant as an educational tool to help you understand your situation and where you need to brush up on your (knowledge) before working with a human adviser.”
She said the software can be a great tool for preparing to see a financial adviser and to ask the right questions.
Jessica Moorhouse, a personal finance blogger with an Accredited Financial Counsellor Canada designation, agrees that AI can serve an educative purpose but suggests cross-checking information learned on an AI app with a human adviser to identify potential inaccuracies.
“Don’t rely on it 100 per cent,” she said. “Just as I tell people to not rely 100 per cent on hiring a professional to make a financial plan.”
ChatGPT has a disclaimer that the results may produce incorrect information about people, places or facts.
Lewis said AI-generated financial advice may be misleading and could result in inferior or suboptimal outcomes.
“One of the reasons is that you have no idea about the quality of advice you’re getting from AI or an online influencer,” he said.
“The challenge with AI and advice is that you really need the advice, (but) you’re also in a position where you have less capacity to understand the quality of advice.
“That’s a paradox – the more you need it, the less you know whether it’s good or bad.”
Lewis suggests a hybrid model, where the computer processes information and numbers while humans focus on guidance and long-term planning, as a perfect marriage of human and machine skills.
“Your computers are not good at empathy,” he said. “They’re great at collecting facts and organizing information.”
He said if humans have to spend less time on asset allocation and account administration, it will free up more time for things they’re good at – empathy and coaching.
This report by The Canadian Press was first published Oct. 26, 2023.
