Latest / The Money Puzzle / How to optimise AI as an active investor

Transcript

How to optimise AI as an active investor · The Money Puzzle

• 0:10
  Speaker 1: Hello and welcome to the Australian's Money Puzzle podcast. I'm
• 0:13
  Speaker 1: James Kirby.
• 0:13
  Speaker 2: Welcome aboard.
• 0:14
  Speaker 1: How far can we go with AI? How far will
• 0:19
  Speaker 1: AI go with us? Can we harness it, you know,
• 0:22
  Speaker 1: to become better investors?
• 0:24
  Speaker 2: That's what I want to.
• 0:24
  Speaker 1: Look at today. AI, we're across AI. We all know
• 0:28
  Speaker 1: about the basics of AI. What about using it and
• 0:30
  Speaker 1: using it as an investor, right? To explore its possibilities,
• 0:34
  Speaker 1: but also its risks. And there are real risks there
• 0:37
  Speaker 1: as we're starting to realise as the weeks go by.
• 0:40
  Speaker 1: and searching for someone as a guest for you who
• 0:44
  Speaker 1: could really be across this issue, I came upon Daniel Gara.
• 0:48
  Speaker 1: Daniel started his career as a financial advisor, and he
• 0:52
  Speaker 1: later left. He co-founded technology company Advisor Logic that was
• 0:56
  Speaker 1: purchased by Morningstar. He's a founder again now of a
• 0:59
  Speaker 1: group called Vectatech. So basically, Daniel's career has been a
• 1:05
  Speaker 1: dress rehearsal for this podcast, I'd like to think. How
• 1:07
  Speaker 1: are you, Daniel?
• 1:09
  Speaker 2: I'm really well, James. It's great to be here. Thanks
• 1:11
  Speaker 2: for having me.
• 1:12
  Speaker 1: I'm delighted to have you. Actually, delighted to find you
• 1:14
  Speaker 1: in that we haven't talked much, but we have dealt
• 1:16
  Speaker 1: with each other, of course, in recent weeks when we
• 1:19
  Speaker 1: tried to capture up on the issues that were sparked,
• 1:21
  Speaker 1: I suppose, to some degree by OpenAI's, what would you say,
• 1:24
  Speaker 1: invasion of Medicare, which became global news, but sort of
• 1:29
  Speaker 1: captured people's imaginations, too, about what on earth is possible
• 1:33
  Speaker 1: and not possible with AI, just to bring it to
• 1:37
  Speaker 1: basics for everyone listening. And on that whole area of
• 1:42
  Speaker 1: AI in the mass market, where do you think we
• 1:45
  Speaker 1: stand with AI in terms of how we, being Australian investors,
• 1:50
  Speaker 1: use AI compared to other regimes or overseas jurisdictions? Do
• 1:54
  Speaker 1: you think we're more or less up to date? Would
• 1:56
  Speaker 1: you say we're ahead, behind or on par?
• 2:00
  Speaker 2: I think we're more or less aligned with everyone else.
• 2:03
  Speaker 2: It's all still pretty new for most people. And the
• 2:05
  Speaker 2: reality is most people aren't using it yet. So if
• 2:08
  Speaker 2: you are, you're probably ahead of the curve. Yes.
• 2:11
  Speaker 1: I saw that you had someone saying that people actually
• 2:14
  Speaker 1: are not using it half as much for all the
• 2:15
  Speaker 1: talk and all the interest. The people aren't actually using it.
• 2:19
  Speaker 1: So what's your observation on that? Do you think that's
• 2:21
  Speaker 1: the case?
• 2:23
  Speaker 2: It is, but it's improving every day. And People will
• 2:27
  Speaker 2: use it overtly. They'll sort of think about it and
• 2:29
  Speaker 2: maybe log into ChatGPT and use it. But more and more,
• 2:33
  Speaker 2: it's just becoming woven into the other products, other software
• 2:36
  Speaker 2: products that we're using. So whether you think about it
• 2:39
  Speaker 2: or not, you're still using AI in one way or another.
• 2:42
  Speaker 1: Right, yes. So it's becoming meshed, if you like, into
• 2:45
  Speaker 1: the technology that we use. But tell us about how
• 2:49
  Speaker 1: we could harness it. When we were working together on
• 2:52
  Speaker 1: a piece I was doing for The Australian, you were
• 2:54
  Speaker 1: talking about how there's so much more people could do
• 2:58
  Speaker 1: as investors with AI. Could you just give us some
• 3:00
  Speaker 1: idea of the possibilities that are there and to what
• 3:04
  Speaker 1: extent they're free and to what extent you'd have to
• 3:05
  Speaker 1: pay for them?
• 3:07
  Speaker 2: It's endless. There is pretty much whatever you can think of,
• 3:11
  Speaker 2: you can do with AI or do better with AI.
• 3:14
  Speaker 2: There are certainly free options. All the main providers will
• 3:17
  Speaker 2: give you a little bit for free. If you become
• 3:19
  Speaker 2: a very heavy user, then they'll want you to pay
• 3:21
  Speaker 2: whatever it is, $ 20, $ 30 a month. But Probably for
• 3:24
  Speaker 2: your listeners, that would be worth paying if they're getting
• 3:28
  Speaker 2: some use out of it. And I think the opportunity
• 3:31
  Speaker 2: is for an investor, you've got to be really quite
• 3:34
  Speaker 2: careful about this because AI is tremendously powerful, but it
• 3:38
  Speaker 2: is also imperfect. And there's a bunch of different things
• 3:41
  Speaker 2: to consider. And so for instance, it's only as good
• 3:45
  Speaker 2: as the question you ask it. So if you don't know,
• 3:48
  Speaker 2: if you're not quite sure what it is that you're
• 3:51
  Speaker 2: trying to find out, You can ask it a question
• 3:54
  Speaker 2: and it can give you an answer to that question,
• 3:57
  Speaker 2: but that doesn't necessarily make it, you know, that may
• 4:00
  Speaker 2: not be the thing that you need to know to
• 4:01
  Speaker 2: do what you need to do. So you sort of
• 4:03
  Speaker 2: need to be a little bit careful with that.
• 4:05
  Speaker 1: Is it the case, I've read, that they are compelled,
• 4:09
  Speaker 1: if you like, to give you an answer? Because sometimes
• 4:11
  Speaker 1: they would be better off not giving an answer, but
• 4:14
  Speaker 1: they must give an answer every time.
• 4:15
  Speaker 2: Is that true? That is true. And I think that's
• 4:18
  Speaker 2: what can really catch people off guard. Because the AI
• 4:22
  Speaker 2: might give you an answer which sounds, you don't need
• 4:25
  Speaker 2: to be afraid of the answer that sounds ridiculous, because
• 4:27
  Speaker 2: you'll know that's ridiculous and you'll push back. It's the
• 4:30
  Speaker 2: answer that sounds totally plausible, that sounds highly intelligent and
• 4:35
  Speaker 2: very well thought out, that you take as truth, that
• 4:39
  Speaker 2: may be a total hallucination, or at least not quite true.
• 4:43
  Speaker 2: It may be a little bit out of date, or
• 4:45
  Speaker 2: it may be for a different jurisdiction. That's the one
• 4:47
  Speaker 2: that can get you into trouble if you rely on it.
• 4:49
  Speaker 1: What do you think investors should be doing? I mean,
• 4:52
  Speaker 1: I'm thinking of what's possible. Do you think investors should
• 4:56
  Speaker 1: be looking at asset allocation and portfolio construction, or do
• 5:00
  Speaker 1: you think they should be looking at long-range computations about
• 5:04
  Speaker 1: how much I need in super or how much I
• 5:06
  Speaker 1: should spend a year?
• 5:07
  Speaker 2: What's the sort of.
• 5:07
  Speaker 1: Things you think would be elementary usage of AI for
• 5:11
  Speaker 1: a listener?
• 5:12
  Speaker 2: I think the narrower the question, the better AI will
• 5:16
  Speaker 2: be in this case. And so... If you want to
• 5:21
  Speaker 2: know something, you're talking about a diversified portfolio, you're trying
• 5:25
  Speaker 2: to build one, something like that, AI can probably help
• 5:28
  Speaker 2: you with that. What I tend to do, I think
• 5:31
  Speaker 2: this is key, rather than ask the AI what I
• 5:36
  Speaker 2: should do, I tell it what I'm thinking and then
• 5:38
  Speaker 2: I ask it to tear that to shreds. Tell me
• 5:40
  Speaker 2: all the reasons I'm wrong. Okay, give me an example.
• 5:43
  Speaker 1: It's very interesting.
• 5:45
  Speaker 2: Like what?
• 5:45
  Speaker 1: Just give me anything at all.
• 5:47
  Speaker 2: Really anything at all. If I'm thinking of, you might say, well,
• 5:52
  Speaker 2: it's something totally different. I'm thinking of going on this
• 5:55
  Speaker 2: holiday in June. If you said, is that a good idea?
• 5:59
  Speaker 2: It'll tell you all the reasons that a June holiday
• 6:01
  Speaker 2: to that location is a good idea. But if you say,
• 6:05
  Speaker 2: tear that idea to pieces, it'll say, oh, well, that's
• 6:09
  Speaker 2: monsoon season. So You might end up in the rain
• 6:13
  Speaker 2: every day, and that's not what you're after.
• 6:15
  Speaker 1: So if I said, I'm thinking of putting 10% in gold,
• 6:18
  Speaker 1: is that a bad idea? Then it would interrogate that notion,
• 6:23
  Speaker 1: would it?
• 6:25
  Speaker 2: Well, it will. I would suggest, though, that kind of
• 6:29
  Speaker 2: a question is much deeper because it's missing all the
• 6:33
  Speaker 2: context about who you are, what your goals are. What
• 6:38
  Speaker 2: other things, you know, what the other sort of facets
• 6:41
  Speaker 2: of your financial planning are that it won't be aware of.
• 6:44
  Speaker 2: So again, it could give you an answer, but is
• 6:47
  Speaker 2: that answer going to be all that useful to you?
• 6:49
  Speaker 2: What's right for one person could be totally wrong for another.
• 6:52
  Speaker 1: How would you frame it then if you had a
• 6:54
  Speaker 1: diversified portfolio and you were a pretty good investor, but
• 6:57
  Speaker 1: you never touched gold or you never touched bonds or
• 7:00
  Speaker 1: you never touched whatever emerging markets. And you said, how
• 7:03
  Speaker 1: would you phrase it to optimize the usefulness of AI?
• 7:07
  Speaker 1: to get an answer?
• 7:09
  Speaker 2: Yeah. So James, let's not look for a definitive answer
• 7:13
  Speaker 2: from an AI. That is, I think that's how you
• 7:15
  Speaker 2: can get into trouble. But you might say, this is
• 7:19
  Speaker 2: what I've got. This is my portfolio. I'm thinking of
• 7:22
  Speaker 2: adding gold to it. I'm not quite sure how much, 5%, 10%, 20%.
• 7:28
  Speaker 2: Give me the pros and cons. And then that could
• 7:31
  Speaker 2: be the starting point So you evolve in your thinking
• 7:34
  Speaker 2: on the topic rather than you just getting a straight
• 7:37
  Speaker 2: out answer that you then rely on.
• 7:38
  Speaker 1: So the questions kind of evolve, at least the answers evolve.
• 7:42
  Speaker 2: Correct. Think of it as a really smart buddy that
• 7:46
  Speaker 2: you can bounce ideas off rather than a mentor that
• 7:51
  Speaker 2: will give you definitive answers.
• 7:53
  Speaker 1: Definitive answers, yeah, which probably is what most people will do.
• 7:57
  Speaker 1: They'll say, oh, okay, so sort this out for me. Okay,
• 8:00
  Speaker 1: so it actually leads you along and the questions evolve
• 8:03
  Speaker 1: so the answers get better and closer to what you're
• 8:05
  Speaker 1: looking for and you can kind of refine it. That's
• 8:08
  Speaker 1: sort of looking at some possibilities of AI, folks. On
• 8:11
  Speaker 1: terms of risks, the original, when I talked to Daniel,
• 8:15
  Speaker 1: it was about the open AI, as I say, invading
• 8:18
  Speaker 1: the Medicare records without permission. And it turns out that
• 8:22
  Speaker 1: they have done this, not just in the Australian government,
• 8:24
  Speaker 1: but several governments.
• 8:26
  Speaker 2: Around the world.
• 8:27
  Speaker 1: And I think it was probably a landmark where people realize, oh, wow,
• 8:30
  Speaker 1: you know, AI, anything is possible because they can go
• 8:35
  Speaker 1: beyond what anyone realized. Could you explain in really simple terms, Daniel,
• 8:40
  Speaker 1: what happened with that open AI breach? I've been reading
• 8:42
  Speaker 1: about how AI agents as such have come to the
• 8:46
  Speaker 1: point that they can train AI agents. And to that extent,
• 8:49
  Speaker 1: it propels itself. And does it ever reach a level
• 8:52
  Speaker 1: where the parent group don't know what the AI agent
• 8:57
  Speaker 1: is doing.
• 8:57
  Speaker 2: Is that possible? If they are willfully ignorant and choose
• 9:02
  Speaker 2: not to know what's going on, then that's possible. But otherwise, no,
• 9:07
  Speaker 2: I doubt that's possible. What is happening there? Let's take
• 9:12
  Speaker 2: a step back. What we just discussed before is that
• 9:16
  Speaker 2: an AI can give you an answer which sounds plausible,
• 9:19
  Speaker 2: which sounds knowledgeable, but is incorrect. That's the first level
• 9:24
  Speaker 2: of risk that we have with AI. The next level
• 9:27
  Speaker 2: of risk is agentic AI, which is not just in
• 9:32
  Speaker 2: a chat where you are talking to the AI and
• 9:33
  Speaker 2: it's giving you an answer. This is where you give
• 9:35
  Speaker 2: it agency to go out and perform tasks on your behalf.
• 9:39
  Speaker 2: What happened with Medicare was agentic AI basically went out
• 9:44
  Speaker 2: and did something that it was not asked to do.
• 9:48
  Speaker 2: That's how that happened. There's the risk of asking it
• 9:52
  Speaker 2: the wrong thing and getting the wrong outcome. And then
• 9:55
  Speaker 2: there's the risk of asking it to do A and
• 9:59
  Speaker 2: it just decides. for whatever reason to go out and
• 10:01
  Speaker 2: do B. It decides. Yep.
• 10:06
  Speaker 1: Is that a danger? When you say it decides, what's
• 10:08
  Speaker 1: really happening there?
• 10:11
  Speaker 2: Well, it's got an objective and the person setting that
• 10:14
  Speaker 2: objective is unaware or doesn't consider what actions are available
• 10:19
  Speaker 2: to that AI. And that it might go out of
• 10:22
  Speaker 2: what you expect it to do to try and achieve
• 10:25
  Speaker 2: that objective. Right.
• 10:26
  Speaker 1: It's ethically agnostic, obviously, unless it's been coded as such.
• 10:31
  Speaker 2: Well, I'm not sure that's right because it's all got
• 10:37
  Speaker 2: a bias, right? Because someone's programmed it. So there's a
• 10:41
  Speaker 2: bias of the company and the people in that company
• 10:43
  Speaker 2: that are embedded into that tool. And that may come
• 10:46
  Speaker 2: out in ways that aren't so obvious. There must be a.
• 10:49
  Speaker 1: Real danger that, you know, people, fraudulent activity will be spectacular.
• 10:55
  Speaker 1: That we will have spectacular incidents of AI agentic-driven fraudulent activity.
• 11:01
  Speaker 1: I mean, we must be really close to it.
• 11:03
  Speaker 2: That's always been the case. As computers and technology have evolved,
• 11:07
  Speaker 2: that sort of fraud has, the potential for it has grown.
• 11:11
  Speaker 2: But what's also grown is the strength of those fighting
• 11:14
  Speaker 2: it and the ability to push back. So, yeah, sure.
• 11:18
  Speaker 2: New models such as Mythos from Anthropic are very powerful
• 11:23
  Speaker 2: and have been able to expose bugs in software and vulnerabilities.
• 11:27
  Speaker 2: But also the ability to protect and to patch those
• 11:30
  Speaker 2: gaps is now increasing dramatically. So as those vulnerabilities are
• 11:35
  Speaker 2: found by the good guys, so to speak, the ability
• 11:37
  Speaker 2: to patch them and defend very quickly has evolved dramatically.
• 11:41
  Speaker 2: So I'm not too pessimistic on that front.
• 11:45
  Speaker 1: Okay, that's interesting that you're not because you're right at
• 11:47
  Speaker 1: the heart of it. Okay, just bringing it back to
• 11:49
  Speaker 1: basics for the listener in terms of their own risk.
• 11:53
  Speaker 1: There is a specific risk, isn't there, that you're, if
• 11:55
  Speaker 1: I go in and I put into AI, you know,
• 11:58
  Speaker 1: I want to retire at a certain age or I
• 11:59
  Speaker 1: want this amount of money and I want to reverse
• 12:01
  Speaker 1: engineer it. So what do I do from today to
• 12:03
  Speaker 1: get there or whatever? That all that information linked with
• 12:06
  Speaker 1: me is out there and it's in the system and
• 12:10
  Speaker 1: it can be reused by any AI company any way
• 12:15
  Speaker 1: they want, really. And I have a way to stop them,
• 12:18
  Speaker 1: but most people don't realize I can stop them. And
• 12:21
  Speaker 1: it's in the settings. Could you explain to the listeners
• 12:23
  Speaker 1: how they can stop AI tools that they use using
• 12:27
  Speaker 1: their information in the future? You gave me a marvelous
• 12:30
  Speaker 1: example of OpenAI, which had this splendidly ambiguous button that
• 12:33
  Speaker 1: nobody would guess is actually the security button. But how
• 12:36
  Speaker 1: would people, when they're using, let's just use chat GPT,
• 12:40
  Speaker 1: which is probably the most common one. How would someone
• 12:42
  Speaker 1: turn off that, AI tool's ability to reuse their information?
• 12:49
  Speaker 2: Good question. So because the default setting in ChatGPT says
• 12:56
  Speaker 2: that they can take what you've put in there and
• 13:00
  Speaker 2: use it to train their model, which basically means that
• 13:04
  Speaker 2: whatever you put in there, they get to hold onto
• 13:06
  Speaker 2: that information. That's probably not what most people want, but
• 13:10
  Speaker 2: that's the default setting. So in order to stop that,
• 13:14
  Speaker 2: You need to go into settings in ChatGPT, and there's
• 13:17
  Speaker 2: a little section called data controls. And then you need
• 13:21
  Speaker 2: to turn off the beautifully named, improve the models for everyone.
• 13:26
  Speaker 2: So rather than being the good guy and improving the
• 13:29
  Speaker 2: model for everyone, you need to instead turn that off,
• 13:34
  Speaker 2: which means that data that you put in won't be
• 13:38
  Speaker 2: used to build the model. you should be aware that
• 13:41
  Speaker 2: they can still hold onto that for a variety of reasons.
• 13:45
  Speaker 2: And they will for normally a month is what you
• 13:48
  Speaker 2: see in the TNC. But it certainly should reduce your risk.
• 13:53
  Speaker 2: I still wouldn't be putting in very personal information that
• 13:56
  Speaker 2: can be linked back to you. I wouldn't put a
• 13:58
  Speaker 2: tax file number in, for instance, and anything you can
• 14:01
  Speaker 2: avoid putting in, I'd avoid it. Okay.
• 14:04
  Speaker 1: It's fascinating that. So folks, if you're using ChatGPT, The
• 14:08
  Speaker 1: default setting is that they can reuse your information. If
• 14:11
  Speaker 1: you don't want it used, you go in.
• 14:13
  Speaker 2: There's a button.
• 14:14
  Speaker 1: It literally says improve the product for everyone. And that
• 14:16
  Speaker 1: button is on. You turn it off. And it goes
• 14:19
  Speaker 1: against your good nature, right? You want to improve the
• 14:21
  Speaker 1: product for everyone. They're so clever, these people. But that's
• 14:23
  Speaker 1: basically how you stop them reusing your information. In the
• 14:27
  Speaker 1: next segment, I'm going to take a look at the
• 14:29
  Speaker 1: most more advanced operations that we can use as investors
• 14:33
  Speaker 1: on AI.
• 14:33
  Speaker 2: Stick with us.
• 14:50
  Speaker 1: Hello, welcome back to the Australian's Money Puzzle podcast. James
• 14:53
  Speaker 1: Kirby here with Daniel Gara of VectorTech. Daniel is an
• 14:57
  Speaker 1: AI technology expert. He is a former financial advisor, which
• 15:02
  Speaker 1: is marvellous because he really understands what we're talking about
• 15:05
  Speaker 1: here on the show. And at the moment, he's actually
• 15:08
  Speaker 1: building something, which we'll talk about in a short while,
• 15:10
  Speaker 1: what he's trying to do in this area. For most listeners,
• 15:14
  Speaker 1: it's worth knowing, Daniel, I think that financial advice has
• 15:16
  Speaker 1: been pretty quick To use AI, I mean, I'm in
• 15:20
  Speaker 1: some financial advisor chat groups and they were really in
• 15:24
  Speaker 1: really early. I mean, two years ago, they were saying,
• 15:26
  Speaker 1: what have we got? What can we do? Because they
• 15:28
  Speaker 1: had this notorious load, if you like, of low, what would.
• 15:33
  Speaker 2: You call it?
• 15:34
  Speaker 1: Regulatory red tape that most advisors will complain they spent
• 15:38
  Speaker 1: most of their time not advising, but ticking boxes and
• 15:42
  Speaker 1: that this could be done by AI. And it does
• 15:44
  Speaker 1: seem to be the case that there is.
• 15:47
  Speaker 2: A lot happening.
• 15:48
  Speaker 1: Now, the big arch rival to OpenAI and ChatGPT is
• 15:53
  Speaker 1: the Anthropic Group, and they have CLAUDE, C-L-A-U-D-E, which is
• 15:58
  Speaker 1: a worldwide attempt, isn't it, to basically nail the financial
• 16:02
  Speaker 1: advice market's AI demands. Tell us about CLAUDE and what
• 16:07
  Speaker 1: Anthropic have planned. What can it do that couldn't be
• 16:09
  Speaker 1: done previously, and how might it change the world for investors?
• 16:14
  Speaker 2: Daniel? Right, so this is... So Claude's their standard product
• 16:18
  Speaker 2: and what they've launched is Claude for Advisors. And Claude
• 16:22
  Speaker 2: for Advisors is intended to sit between all the online
• 16:27
  Speaker 2: tools that advisors use to make them more efficient in
• 16:30
  Speaker 2: their day-to-day work. So whereas previously an advisor might have
• 16:34
  Speaker 2: a client coming in and they want to update their
• 16:39
  Speaker 2: records on what that client is holding, what insurances, what investments,
• 16:43
  Speaker 2: the idea is that Claude for Advisors would just go
• 16:46
  Speaker 2: in to all their systems, grab the data, synthesize it
• 16:50
  Speaker 2: into a nice report that they could review themselves to
• 16:54
  Speaker 2: get up to date on where that client's at and
• 16:56
  Speaker 2: maybe show it to the client as well.
• 16:58
  Speaker 1: Right.
• 16:59
  Speaker 2: Yeah.
• 17:00
  Speaker 1: And it's the first time in a formal fashion that
• 17:02
  Speaker 1: one of the big AI outfits has zoned in on
• 17:06
  Speaker 1: this profession of financial advice. Will it change financial advice,
• 17:10
  Speaker 1: do you think?
• 17:13
  Speaker 2: It will over time. It's one of the things that will.
• 17:15
  Speaker 2: It's not the only... thing but it's definitely a step
• 17:18
  Speaker 2: in the right direction it's intended to remove the tedious
• 17:23
  Speaker 2: tasks that are not particularly value add but just need
• 17:27
  Speaker 2: to be done and i think you know historically over
• 17:30
  Speaker 2: over centuries we humans have been pretty good at picking
• 17:33
  Speaker 2: those up and getting value.
• 17:35
  Speaker 1: From them yeah it is a particular it would seem
• 17:38
  Speaker 1: to be a genuine complaint of the financial advisory sector
• 17:41
  Speaker 1: that the more problems they had the more regulations were
• 17:44
  Speaker 1: laid upon them not that Those regulations seem to make
• 17:48
  Speaker 1: much difference. We have an endless line of, a conga
• 17:50
  Speaker 1: line of fraudsters in financial advice, and they never seem
• 17:53
  Speaker 1: to go away, regardless of what regulations are put upon them.
• 17:57
  Speaker 1: I want to ask you now about your own, what
• 18:00
  Speaker 1: you're trying to do in this area in Australia. Tell
• 18:04
  Speaker 1: us a little bit about VectorTech, what you're trying to build.
• 18:07
  Speaker 2: Yeah, thanks, James. So the idea of Vector is we're
• 18:12
  Speaker 2: looking to help the segment of the market that doesn't
• 18:14
  Speaker 2: currently get advice. So if you have a relatively sophisticated
• 18:19
  Speaker 2: situation and you have enough money that you can justify
• 18:22
  Speaker 2: spending the thousands of dollars on an advisor, then you
• 18:27
  Speaker 2: already have a solution. And that's great. I'm a huge
• 18:29
  Speaker 2: proponent of financial advice. But then there are people who have,
• 18:34
  Speaker 2: let's say, maybe $ 100, 000 in super and a mortgage, and
• 18:40
  Speaker 2: they need advice. They want advice. But they can't really
• 18:43
  Speaker 2: justify or afford the price of an advisor. And so
• 18:48
  Speaker 2: using AI, what we're looking to do is reduce that
• 18:52
  Speaker 2: price point so that investors can use Vector for the
• 18:58
  Speaker 2: initial part of the financial planning process, which are things
• 19:01
  Speaker 2: like exploring your goals, understanding your current financial position, your
• 19:06
  Speaker 2: propensity to take risks, and a few other areas, and
• 19:11
  Speaker 2: wrapping that up and bringing it to an advisor who
• 19:13
  Speaker 2: can then provide advice. And the idea is that will
• 19:18
  Speaker 2: dramatically reduce the cost to serve for an advisor, which
• 19:22
  Speaker 2: means that they can reduce their prices for that kind
• 19:26
  Speaker 2: of interaction, still profitable for them, and bring it to
• 19:30
  Speaker 2: a price point where more Australians will be able to
• 19:33
  Speaker 2: look forward to see an advisor.
• 19:35
  Speaker 1: So they can concentrate on the value add that the
• 19:38
  Speaker 1: advisor brings to the table, as opposed to their ability
• 19:41
  Speaker 1: to tick boxes on really broad elemental stuff like your goals,
• 19:45
  Speaker 1: et cetera. I see. Very interesting.
• 19:47
  Speaker 2: When is this out?
• 19:48
  Speaker 1: When is it released?
• 19:48
  Speaker 2: Are you there yet? We're doing pilot programs as we speak. Okay.
• 19:53
  Speaker 1: Very good. What I want to do now is we'll
• 19:55
  Speaker 1: take a break, but I've collected some questions for you,
• 19:57
  Speaker 1: which is a collection of listener questions that have come in.
• 20:01
  Speaker 1: And we're going to talk about what random questions, but
• 20:04
  Speaker 1: useful for you in terms of AI. Could you build
• 20:08
  Speaker 1: a portfolio on AI? What sort of a portfolio could
• 20:11
  Speaker 1: you expect to build on AI? You know, what's the
• 20:13
  Speaker 1: best thing we've seen around? What are the worst things
• 20:16
  Speaker 1: we've seen?
• 20:16
  Speaker 2: Back in a moment.
• 20:28
  Speaker 1: Hello, welcome back to the Australian's Money Puzzle podcast. James
• 20:31
  Speaker 1: Kirby here talking to Daniel Garra, Okay, Daniel, so let's
• 20:35
  Speaker 1: really test it. Let's say I sit down with a
• 20:37
  Speaker 1: couple of AI tools, just the usuals, you know, ChatGPT,
• 20:40
  Speaker 1: Google Gemini, whatever. I mean, how far can I go?
• 20:43
  Speaker 1: Can I assemble a reasonably diversified, reasonably dependable portfolio just
• 20:52
  Speaker 1: with those tools, do you think?
• 20:55
  Speaker 2: You can. You can, you know, create an infinitely diversified portfolio.
• 21:00
  Speaker 2: I think that the trick is, though, that Diversification is
• 21:03
  Speaker 2: not the objective. It's just the tool that you need
• 21:07
  Speaker 2: when creating a portfolio. Again, you need to be careful
• 21:10
  Speaker 2: here what questions you're asking of it. When you're speaking
• 21:14
  Speaker 2: to the AI, what are you trying to do? What
• 21:17
  Speaker 2: sort of diversification are you looking for? What's the objective
• 21:20
  Speaker 2: of the portfolio? You need to know what it is
• 21:25
  Speaker 2: that you want when you start a conversation with an AI.
• 21:30
  Speaker 2: There's no point in saying, hey, give me a great
• 21:32
  Speaker 2: diversified portfolio. Are we looking for something that you could
• 21:37
  Speaker 2: cash out within the year if you needed to or
• 21:39
  Speaker 2: something that is for 20 years? It also won't know
• 21:43
  Speaker 2: what other assets you have and what sort of volatility
• 21:46
  Speaker 2: and risk exists there.
• 21:48
  Speaker 1: Okay. Tell me if I was, if I plowed ahead
• 21:51
  Speaker 1: and it did one for me, would it? I'm guessing,
• 21:54
  Speaker 1: but I imagine that they have a bias towards listed investments.
• 21:57
  Speaker 1: I expect it would all be ETFs. I doubt they
• 21:59
  Speaker 1: would say to me, buy property and buy number 22
• 22:02
  Speaker 1: railway cuttings in Western Sydney.
• 22:08
  Speaker 2: It depends which AI you're using. If it's an AI
• 22:10
  Speaker 2: from a company that sells those products, then yes, I
• 22:12
  Speaker 2: think it's highly unlikely that an ETF company would provide
• 22:15
  Speaker 2: an AI that tells you to go out and buy
• 22:17
  Speaker 2: an investment property. But if it's one of the generic
• 22:21
  Speaker 2: ones like DeClaude or Gemini, chat GPT, it probably doesn't
• 22:26
  Speaker 2: have too much of a bias. But again, it's also
• 22:30
  Speaker 2: not really going to have a proper understanding of your background.
• 22:33
  Speaker 2: So I still see that relationship with a good financial
• 22:39
  Speaker 2: advisor is the gold standard. What this might help you
• 22:42
  Speaker 2: do is come to that advisor with better questions or
• 22:47
  Speaker 2: a suggestion that you can unpack with that advisor rather
• 22:50
  Speaker 2: than just being a blank slate. Right. Yeah.
• 22:56
  Speaker 1: How far do you think you can go alone without
• 22:59
  Speaker 1: ever going to an advisor? I mean, can you go
• 23:01
  Speaker 1: further now than you used to be able to with
• 23:03
  Speaker 1: the arrival of AI tools?
• 23:06
  Speaker 2: You can. More is possible, which also means that the
• 23:12
  Speaker 2: danger increases as well. So because you can do so
• 23:17
  Speaker 2: much more so quickly... You need to be careful that
• 23:21
  Speaker 2: you don't confuse quantity and speed with quality. Going back
• 23:26
  Speaker 2: to our previous example, you could go out and design
• 23:29
  Speaker 2: something with AI that seems, on the surface of it,
• 23:33
  Speaker 2: to be perfect, awesome, exactly what you want, but you
• 23:37
  Speaker 2: haven't really verified that with an expert. You don't know
• 23:41
  Speaker 2: what you don't know, and it doesn't know either. Unless
• 23:46
  Speaker 2: you are really being incredibly careful and unless you are
• 23:50
  Speaker 2: very confident that you've given it all the relevant information
• 23:54
  Speaker 2: to give you that advice and that you've then gone
• 23:56
  Speaker 2: and verified and validated the advice it's given you, I
• 24:00
  Speaker 2: think that you're still taking a bit of a risk.
• 24:03
  Speaker 1: It's really interesting. For someone who is so deep in tech,
• 24:07
  Speaker 1: it seems to me you sound like you believe the
• 24:11
  Speaker 1: human factor matters. must be there still in big decisions financially.
• 24:18
  Speaker 2: Absolutely. That's my recommendation. I would want to keep a
• 24:22
  Speaker 2: human in the loop, keep an expert in the loop.
• 24:25
  Speaker 2: I think the opportunity with AI and advice and portfolio
• 24:29
  Speaker 2: design is making it more efficient and being able to
• 24:34
  Speaker 2: get better ideas and a more diverse and interesting group
• 24:38
  Speaker 2: of ideas for you to consider and make decisions or
• 24:41
  Speaker 2: sort of put you in a direction, but you still
• 24:44
  Speaker 2: want an expert to say, well, that's all well and good. However,
• 24:49
  Speaker 2: we haven't considered A, B, and C, and that could
• 24:51
  Speaker 2: be catastrophic. The classic one is, do I pay off
• 24:55
  Speaker 2: my home loan or do I contribute to super? I
• 24:58
  Speaker 2: think everyone's thought that one at some point. Mathematically, the
• 25:03
  Speaker 2: answer might be clear, but If your goal is to
• 25:06
  Speaker 2: do something in particular in five years' time and you
• 25:09
  Speaker 2: can't actually access your super for another 15 years, then
• 25:13
  Speaker 2: maybe putting the money into super isn't the way to go.
• 25:16
  Speaker 2: Even though it might be tax efficient and it might
• 25:19
  Speaker 2: give you a better dollar outcome, if you needed that
• 25:21
  Speaker 2: money five years from now, then you've just created a
• 25:23
  Speaker 2: problem potentially.
• 25:25
  Speaker 1: It was right, but it was wrong for you.
• 25:28
  Speaker 2: Correct.
• 25:28
  Speaker 1: Okay, and that's when it goes wrong. Tell me what's
• 25:31
  Speaker 1: the most impressive thing you've seen AI do? It doesn't
• 25:34
  Speaker 1: even have to be in financial world. What did you
• 25:36
  Speaker 1: come across where you just went like, whoa, that's really something?
• 25:40
  Speaker 2: Well, I'm working in technology, and when we get all
• 25:42
  Speaker 2: sorts of documents from third parties, integrations that we have
• 25:46
  Speaker 2: to do, and I can throw in a 200-page document,
• 25:50
  Speaker 2: ask it a couple of questions, and it will give
• 25:52
  Speaker 2: me answers seconds later. Stuff that would have taken someone,
• 25:58
  Speaker 2: days to go and trawl through and fiddle with and unpack.
• 26:03
  Speaker 2: And that's just wild. I still get excited every time
• 26:07
  Speaker 2: we do something like that. It's a great time to
• 26:09
  Speaker 2: be alive and be in tech for sure. Yeah.
• 26:13
  Speaker 1: Well, I hope some of our listeners pick you up
• 26:15
  Speaker 1: on that. It is quite extraordinary and I use it
• 26:17
  Speaker 1: all day long and I find it terrific. I've also
• 26:20
  Speaker 1: found it very useful as an investor. More on computation, really,
• 26:26
  Speaker 1: and calculation. That's what I found it particularly useful for,
• 26:30
  Speaker 1: because I didn't really have those tools before. I just
• 26:32
  Speaker 1: had a calculator. And you can go a lot further now,
• 26:35
  Speaker 1: especially where there's issues with duration or time or timing.
• 26:41
  Speaker 2: It's splendid.
• 26:42
  Speaker 1: Terrific to talk to you today, Daniel Gara of VectorTech.
• 26:45
  Speaker 1: Lovely to have you on the show, and thanks for
• 26:47
  Speaker 1: your observations there.
• 26:48
  Speaker 2: Thanks so much, James. Have a good day.
• 26:50
  Speaker 1: That was Daniel Gara of VectorTech, a financial advisor one
• 26:54
  Speaker 1: time who has become something of a technology force, but
• 26:57
  Speaker 1: an open AI and AI expert. Very useful information. Keep
• 27:02
  Speaker 1: that in mind, folks. Keep the emails rolling, themoneypuzzle at theaustralia.com.au.
• 27:08
  Speaker 1: Today's show was produced by Tiffany Dimack. Thank you.