Artificial intelligence (AI) and machine learning provide the possibility of enabling SMSF software and systems to autonomously and automatically pick out funds that have issues, lessening the burden of compliance on practitioners, according to BGL Corporate Solutions managing director Ron Lesh.
“Machine learning is a subset of AI,” Lesh told the BGL Regtech17 conference in Sydney yesterday.
“The term was first coined by computer scientist Arthur Samuel in 1959 and [he defined it as] a field of study that gives computers the ability to learn without being explicitly programmed.
“So the question is: what can AI and machine learning do for you?”
Predictive compliance was a key upshot of machine learning specifically for the accounting and SMSF sectors, he highlighted.
“What would happen if the software told you when your funds were non-compliant through alerts?” he said.
“We’re already starting to do that with alerts for pensions when you haven’t paid enough or paid too much or getting to a point where you’re going to pay too much, and also with contributions going over limits – that’s just a bit of [predictive compliance].
“But how about the monitoring of transactions and determining that this fund is breaching one part of the Superannuation Industry (Supervision) Act or maybe the tax act? Does it have an in-house asset? Is the loan-to-value ratio against the property valuation in breach of the loan documents?
“All of this smart stuff we can do around the data that’s in the software.”
Another benefit machine learning offered was in the area of asset and return assessments, he added.
“It can look at the particular returns on a set of assets and find that the fund has five investments in listed property syndicates or real estate investment trusts, for example, and they’re performing well below the market average, so should you be investing in those?” he said.
“More automated statements of advice are another outcome, as well as smart transaction matching where the machine learns the rules you’ve created around data and auto processes it next time.”