Person
David Hardoon
This page collects what David Hardoon said on the Analyse Asia podcast with Bernard Leong, in an episode on enabling AI at scale and using governance as a competitive advantage. The talk covered AI in banking, from regulation and fairness to fraud monitoring and cyber attacks.
David Hardoon said that while at the Monetary Authority of Singapore he took banks' compliance officers to lunch to learn what held back AI adoption. He also said current transaction monitoring systems produce about 97% false positive alerts, and that AI overlays have cut them by up to 50%. Each fact below links to the moment it was said.
1 episodes · 17 quoted claims
While at the Monetary Authority of Singapore, David Hardoon took local banks' compliance officers to lunch to learn what was holding back AI adoption.
“I took the compliance officers from the various banks locally to lunch. and they were very shaken. They were like, wait, wait, wait, why is MAS taking us to lunch?”
Analyse Asia with Bernard Leong · 0:10:10
According to David Hardoon, the FEAT principles state objectives to be met rather than prescribing how to meet them.
“In the FEE principles, it's not prescriptive. It's providing the objective that is trying to be met.”
Analyse Asia with Bernard Leong · 0:11:51
On the use of sensitive attributes in AI models, the FEAT principles required firms to justify the use rather than allowing or banning it.
“We didn't say you can or you can't. What we said is you need to justify it.”
Analyse Asia with Bernard Leong · 0:11:51
David Hardoon cited the Apple Card controversy at Goldman Sachs as an example of debate over use of gender information in credit models.
“let me give you a famous reference of what happened with the Apple card in Goldman Sachs where there was a big hoo-ha.”
Analyse Asia with Bernard Leong · 0:16:30
David Hardoon said that excluding gender from a sophisticated model does not mean its influence is absent, so the attribute may be needed to detect discrimination.
“And what you find is that when you start using the more sophisticated type of techniques, actually using explicitly within a model doesn't mean it's not there.”
Analyse Asia with Bernard Leong · 0:16:30
David Hardoon said fully autonomous AI agents performing cash reconciliation are in production use at banks, correcting errors that people previously handled.
“you have literally fully autonomous agents in terms of cash reconciliation. Almost think of it as self-healing.”
Analyse Asia with Bernard Leong · 0:34:01
David Hardoon said current transaction monitoring systems produce about 97% false positive alerts.
“Today, our solutions, and I'm not throwing mud because it is what it is, 97 % false positive.”
Analyse Asia with Bernard Leong · 0:35:33
David Hardoon said he has seen AI overlays reduce transaction monitoring false positives by up to 50%.
“I've seen overlay of AI that has been able to drastically reduce that by up to 50%.”
Analyse Asia with Bernard Leong · 0:35:33
David Hardoon said that in zero-day cyber attacks, intruders typically sit inside a system about 130 to 180 days before the actual attack.
“when they've already infiltrated the system usually happens on average about 130 to 180 days before the actual attack.”
Analyse Asia with Bernard Leong · 0:36:05
David Hardoon described one organization that chose not to use AI in a certain area because it conflicted with its own conduct approach.
“They decided we don't want to because of our approach, our conduct, blah, blah, blah, whatever it may be, we are actively choosing not to leverage on these capabilities”
Analyse Asia with Bernard Leong · 0:17:01
David Hardoon said that for teams whose purpose is innovation, revenue and quantifiable benefit are downstream effects and direct measurement is not appropriate.
“Because innovation, your revenue, your potential quantifiable benefit, it's a downstream impact.”
Analyse Asia with Bernard Leong · 0:46:35
David Hardoon advocates measuring AI adoption and implementation with quantitative financial metrics.
“I do believe in enforcing a quantitative and for simplicity financial monitoring.”
Analyse Asia with Bernard Leong · 0:47:13
David Hardoon gave an example of an operations initiative that reduced collection calls from five to two per dollar collected.
“if I previously had to make five calls before to collect $1, now I only have to do two calls to collect that $1.”
Analyse Asia with Bernard Leong · 0:49:29
David Hardoon said he uses Grok for grammar and recommendations on his own writing rather than having it write for him.
“I use Grok extensively. My preferred approach of using it is I don't have Grok write for me, but I write”
Analyse Asia with Bernard Leong · 0:52:39
David Hardoon recommends ending AI prompts with an instruction asking the AI to pose questions back to the user.
“literally end with, ask me questions as part of this conversation.”
Analyse Asia with Bernard Leong · 0:53:06
David Hardoon likes to treat the AI process as an assembly line to give consistency and replicability in development and strategy.
“I try and look at things in a way which results ultimately in an assembly line.”
Analyse Asia with Bernard Leong · 0:30:21
David Hardoon's academic thesis focused on semantic models, which capture knowledge so that the learning mechanism becomes secondary.
“I was fascinated about the world of semantic models. So the premise behind semantic models is capturing information, capturing knowledge.”
Analyse Asia with Bernard Leong · 0:03:53