Property investors placing their trust in unreliable AI systems for investment advice face huge financial risk, a new report reveals.
A national study by financial services group MCG Quantity Surveyors, in which it tasked ChatGPT with listing suburbs matching certain key investment criteria revealed AI remained woefully inept at simple property-related information gathering tasks, generating recommendations that were statistically dubious.
The report put AI through its paces, from beginner to advanced mode, with the results mixed, and concerningly, worse when performed in “Deep Research” mode.
AI is supposed to make our lives easier. But this report shows it can cause some major headaches. Pic: Supplied
The report tasked ChatGPT, in basic mode, to list – with supportive evidence – its top three suburbs in the wider Adelaide region that would suit a budget of $1m with rental yields of 4 per cent or higher and have the best growth potential over the next five years.
AI suggested a two to three-bedroom unit in Mawson Lakes, a two-bedroom unit in Port Adelaide, and a two-to-three-bedroom unit in Oaklands Park.
While MCG Quality Surveyors’ report praised the insights it provided into growth drivers, its proximity to transport and the accuracy of matching realestate.com.au data, it criticised its listing of only units, despite there being plenty of houses available.
Mike Mortlock of MCG Quality Surveyors.
Report author and MCG Quality Surveyors Mike Mortlock warned against buyers agents and other investment groups spruiking AI models to pick investment locations.
“There are a lot of agencies saying they use AI,” he said.
“Seeing the results, we realised there should be a word of caution around making huge financial decisions around it.”
Mr Mortlock said the suburbs and housing AI picked as good investments were problematic as many were high-rise units markets where conditions did not favour investors – in some cases, because there was an oversupply of certain types of apartments.
“AI selected units in areas that don’t outperform and at price points that didn’t match the prompts,” he said.
“It would be hard to argue that these are good picks.”
AI for investment advice? Yeah, we’re not quite there yet …
The information provided to ChatGPT in “Deep Research” mode was far more detailed and the output requested more specific.
In this mode AI recommended Brompton/Bowden units, Salisbury houses and Christies Beach/Noarlunga Downs houses – earning praise for its supplied insights into growth drivers and the fact that it requested houses in two of three, but criticism for failing to match data provided and an unacceptable level of accuracy.
“When we gave it more context, provided 10 years of trend, and lots of data points, it worked worse,” he said.
“It was quite a concern, it was hallucinating. The data accuracy wasn’t there.”
Cameron Galloway can’t imagine relying on AI for investment advice. Picture: Jonathan Ng
Property investor Cameron Galloway was 35 when bought his first property in Modbury North through a buyer’s agency and said turning to AI to help in the property search would be concerning for him.
“I think AI is not at that stage yet and sometimes it gives you the answer that you want rather than that you need to hear,” he said.
“It’s a little bit self-pleasing, so it’s a bit worrying.”
– with Aidan Devine
English (US) · 