- 3 years ago
Kaiju ETF Advisors builds, trains, and employs robust artificial intelligence (AI) and machine learning technologies designed to improve fund management decision-making. By empowering these innovative technologies to curate and provide direct management of our ETF, we’re striving to go places no Registered Investment Advisor has gone before.
Category
🗞
NewsTranscript
00:00 (upbeat music)
00:02 - What is up Zinger Nation?
00:04 Welcome back to another episode of Benzinga Live.
00:06 Happy Tuesday, everybody.
00:08 (upbeat music)
00:10 All right, Nicholas, how are you doing today
00:16 on this Tuesday?
00:18 - Hey, Aaron, great.
00:19 That was a nice setup, I like that.
00:21 Pleased to meet you.
00:24 I'm excited to be here.
00:25 Yeah, I'm doing fantastic, thanks.
00:27 It's raining here, we've got a bit of a tropical wave
00:32 passing through, so it's hurricane season.
00:35 Every 10 days, we get a bit of a weather passing through
00:39 and typically doesn't turn into anything,
00:41 but today is just rainy and no overcast.
00:45 - Where are you joining us from?
00:47 - So yeah, I'm in Barbados.
00:50 - Okay.
00:50 - Mostly I spend out here, but between here and Europe,
00:54 it's typically where I am, yeah.
00:56 - Nice, so yeah, we got some people in the chat.
00:58 Looks like they're joining from Florida,
00:59 so they might know a little bit about the rain.
01:02 We're here in downtown Detroit.
01:03 It's super sunny out,
01:04 which is why it's kind of like backlit against me.
01:06 You can't really see the background right now.
01:09 But Nicholas, so before we get started,
01:12 you wanna tell us a little bit about your role at Kaiju?
01:14 You know, we've gotten to know Ryan a little bit,
01:17 you know, now through a couple of different weeks
01:19 of having these conversations,
01:20 but I'm curious exactly, you know,
01:22 where you fit into this Kaiju story.
01:25 - Sure, so I've been looking at the videos with Ryan
01:29 and you guys have been having a great conversation.
01:31 So this is really exciting to come in
01:34 and add a bit of a depth on the technical and AI side here.
01:38 So I'm the director of artificial intelligence.
01:42 And so I lead up the team.
01:44 Typically, you know, my role is going out there
01:47 and looking for guys and gals to join the team.
01:52 We typically look for people from varying backgrounds,
01:56 right, mathematics, physics,
02:00 mainly the sciences, computer science,
02:04 but having a varied background
02:06 and sort of bring these guys together.
02:08 And, you know, I look at my role
02:10 as giving these guys really, really complex
02:15 and exciting problems to work with.
02:19 And if I can keep their work really exciting,
02:23 then I think I've done my job.
02:24 And it's a great place to work here at Kaiju.
02:28 We're a super fun team and very flat,
02:31 completely remote, as I think you may know,
02:34 but we've been doing this for a while
02:36 and it's been a blast and it's really been fun.
02:38 And we are just a really good set of folks here.
02:43 - Yeah, I mean, it sounds like, you know,
02:44 you guys are doing some really cool things
02:46 and using technology that's truly, you know,
02:49 kind of on the forefront of this like new age,
02:52 next wave technology.
02:54 So it's been a fascinating story to learn more about
02:58 through my conversations with Ryan.
02:59 So I gotta thank you guys for coming on
03:02 and giving us your time,
03:03 because like I said, at the top of the show,
03:04 I mean, this has been,
03:06 AI has like been the biggest story of 2023 on Wall Street,
03:10 maybe off Wall Street too, maybe just in general,
03:12 this has been like the biggest storyline of 2023.
03:15 So to have guys like yourself and Ryan Nicholas
03:19 that have come on the show and give insights
03:21 has been very helpful,
03:23 not just for me, but for the rest of the audience.
03:25 So I know we've got started a little late,
03:27 so I don't wanna take up too much of your time.
03:29 You sound like a busy guy.
03:30 Let's get right into it, Nicholas.
03:32 You're the AI guy, so let's start there.
03:35 How do you define AI as opposed to other advanced computing
03:38 that has come before?
03:40 You know, like Ryan and I have talked about quantum trading,
03:43 how that's been around for a while,
03:44 but how is this AI right now that Kaiju is using,
03:48 how is that different from some of these technologies
03:51 that maybe we've already seen in the past?
03:53 Sure, so that's a big question.
03:57 There's a lot to unpack there,
03:59 and I'm really into this space
04:02 and we could spend the rest of the morning
04:03 talking about this.
04:04 So I'm probably gonna have to restrict myself a bit here.
04:08 So yes, you're absolutely right.
04:11 You know, AI has sort of mushroomed
04:13 in terms of the narrative in the media
04:18 and the interest recently,
04:19 and that's been for good reason.
04:22 We can get into that,
04:23 but there's been some real progress
04:24 in the last few years in this space.
04:27 But AI is not new.
04:29 That's the first thing.
04:30 I would probably point back to the '50s,
04:34 1950, a paper by Alan Turing,
04:36 and then a few conferences there
04:37 where the research community in academia
04:42 really got into trying to answer the simple question
04:45 that Alan Turing posed,
04:46 can machines think?
04:49 It's where the movie got its name from,
04:51 the Imitation Game.
04:52 So he said, you know, can it think like a human
04:54 or imitate thinking like a human?
04:56 And so people have been doing this for a long time.
05:01 In terms of the other advanced computing areas,
05:05 things like distributed computing,
05:07 knowledge representation,
05:09 maybe vision, computational optimization,
05:14 these are classes of advanced computing
05:17 that have been developed for decades,
05:20 and they form the basis of much of what we do in AI.
05:25 And so I see AI not as something
05:29 that's just suddenly emerged and sitting out here,
05:32 but on a long, continuous path of development
05:35 of all of the technologies that have come before us.
05:39 I mean, you know,
05:40 obviously it's fully dependent on computing,
05:42 but computing is dependent on solid state physics,
05:45 quantum mechanics, information theory.
05:49 So it's on a long continuity of progress,
05:54 and all of these advanced elements of computing
05:57 that have gone in the decades before
06:01 are the foundations of what we do.
06:03 So yeah, so I really see it as not separate,
06:10 but interdependent, and yeah.
06:13 - Yeah, I mean, that's incredible.
06:16 And I mean, right now,
06:17 I'm sure you've seen this, Nicholas,
06:19 that a lot of people who maybe have been outside of AI
06:22 have been trying to like use this wave as,
06:24 you know, to get in on it.
06:26 You've been around AI for a while.
06:29 You started an AI consulting company almost a decade ago.
06:33 How much more capable is AI today?
06:35 I mean, I'm sure you've seen like a lifetime's worth
06:38 of technological growth in just 10 years.
06:41 Did you foresee it advancing this quickly?
06:44 - So what I would say about that is I think a lot of people
06:49 in research and in this space were quite shocked
06:53 in the last few years about the amount of perceived progress,
06:57 especially with generative AI.
07:00 I think there's been some amount of surprise around that.
07:06 And we can get into maybe a bit of why that is.
07:08 So in some sense, there's been real progress
07:13 and leaps forward in this space.
07:15 But in other senses, it's just a long progress
07:20 that's more linear as well,
07:21 meaning that we get more computational power,
07:24 more access to data.
07:26 These are the oil of AI,
07:28 and AI needs these resources to improve what it does.
07:35 And so on that aspect, it's just a typical sort of linear
07:40 or maybe there's some exponential curve there
07:45 because the underlying data
07:46 and the computational power Moore's law
07:49 is sort of seems to be somewhat exponential.
07:52 So we do see it increasing,
07:53 but it isn't sort of a step change.
07:56 So in one way, surprising, but in another way,
08:00 just it's on a continuing way forward, if you like.
08:06 What else can I say around that area?
08:10 So yeah, I think there's a lot of mixed,
08:12 there is a lot of mixed feelings now in the industry,
08:18 especially in research about what this latest wave of AI
08:23 really means and is it getting us closer
08:25 to sort of general artificial intelligence or not,
08:30 but that's sort of more philosophical, I would say.
08:35 Yeah, so is that what you would say is still lacking
08:37 in the technology is kind of this next step toward AGI?
08:41 So on that note, I mean, I have my views.
08:47 As I said, it does seem that the field is split.
08:52 A lot of people are talking about
08:53 artificial general intelligence in the next five years.
08:57 Well, people have been saying that,
09:00 or somewhere between five and 20 years,
09:02 and people have been saying that for 70 years.
09:05 People have been saying, well-known people have been saying,
09:08 we're gonna have self-driving cars in the next two years.
09:11 I think Tesla said that in '18, right?
09:14 And Facebook has said their mission is to build
09:19 all of the human faculties in the next five years.
09:21 I think they said that sometime around '17 as well,
09:24 vision, natural language processing, cognition.
09:28 It doesn't seem that we're gonna solve that
09:30 in the next couple of years.
09:32 Outright.
09:33 So a lot of people think that we're very close.
09:37 I am more in the skeptical camp here.
09:40 I think that although that they are leaps
09:43 in more linear directions or narrow directions in AI,
09:48 in particular around the large language models
09:53 and what they can do,
09:54 I'm not sure that that sort of generalizes
09:57 to general artificial intelligence.
10:01 And the other sort of thing that I sort of maybe
10:04 didn't specifically address early on in your question
10:08 is the definition of AI itself.
10:10 Well, there is no tight definition of AI
10:14 because there is no tight definition
10:16 of intelligence outright.
10:18 We still are learning about how the brain
10:21 actually does what it does.
10:23 So there is still a large gap,
10:29 and there still is a lot of learning.
10:31 And what I would say is I question the whole sort of field
10:36 of going into replicating a human in a computer.
10:38 Like I wonder the benefits of that.
10:40 I mean, quite clearly, these techniques are very, very,
10:44 very powerful in more linear or specialized areas,
10:48 like finding patterns for your social media feed
10:53 or in network intrusion detection or fraud
10:57 or things like that.
10:58 But in terms of like replicating,
11:00 we've got nearly 8 billion humans on the planet,
11:03 and I kind of tend to get on quite well
11:05 with the ones I know.
11:06 And I wonder the general rarity of it
11:10 is trying to get a ghosting machine.
11:13 I wonder why.
11:14 So I'm not convinced about that aspect of it.
11:19 And I think it's still very far out, if at all.
11:22 Yeah, that's interesting.
11:23 I think that's a perfect kind of segue
11:25 into my next question,
11:27 'cause you were talking about like,
11:28 what's really the benefit of putting human intelligence,
11:32 like we know what humans are good at already,
11:35 why are we trying to replicate that
11:36 instead of focusing on technology,
11:38 doing things that we're not already good at?
11:41 But what are, I mean, you've been in this industry
11:43 for a while, what are the ethical limitations
11:45 of AI investing?
11:47 Do these machines understand the context
11:50 of what they are investing in?
11:52 Will human oversight always be necessary?
11:57 Right, so quite a bit to unpack there again.
12:00 Great questions.
12:02 Human oversight will always be required.
12:05 Like we are not going to,
12:07 or at least to be responsible, right?
12:10 And ethical, like we're not going to develop
12:13 these AI autonomous programs,
12:17 write a blank check and say,
12:19 go off and trade on the markets.
12:22 What we are going to do is,
12:24 so I think you are asking about,
12:28 in terms of responsibility
12:30 and the ethical applications of AI in investing.
12:33 So one of the things I observe, right,
12:35 with people, humans and investing,
12:38 it's immensely emotional,
12:40 even without technology and AI.
12:43 What we do with our money and who we give it to to invest,
12:49 it's a hugely emotional part of our lives.
12:52 And trust is a huge factor.
12:54 So if you are going to take on any aspect of that,
13:00 we have to be very transparent about what we do,
13:04 how we do it, where we get the data from, is it biased?
13:08 And how do we ring fence what we're doing
13:12 with guardrails and rules?
13:14 Like it's not just go off and buy any underlying asset,
13:19 right?
13:20 Don't go and put all risk in one asset or one sector.
13:24 There's diversification, there's risk management.
13:27 These are hard rules that the AI can't break
13:31 and we have to have oversight over that
13:32 and we have to be able to explain that to potential investors.
13:37 So I do think there's a lot to do there.
13:39 I do think there's a lot to do in education
13:41 and explain to people how we do what we do
13:44 and that this isn't magic and hocus pocus
13:46 and that this is about using the really capable aspects
13:51 of this type of AI and machine learning
13:55 and neural nets to crunch vast amounts of data,
13:58 which it's extremely good at,
14:02 and finding patterns and correlations in that data
14:05 and leveraging that.
14:06 I mean, that's not new.
14:07 People have always used data to make investment decisions
14:11 as far back as whenever.
14:13 The challenge is that as a human being,
14:16 you can do a handful of those in a day maybe, right?
14:19 Where this type of technology can just scale that
14:21 and look at hundreds, thousands, tens of thousands
14:25 of data points or underlying assets potentially.
14:28 And I think that's where the edge comes from in this case.
14:32 I think in future, maybe there's gonna be more generalization
14:37 and more context, but at the moment,
14:38 no, computers and AI is not really good at broad context
14:42 in any way.
14:43 It's more linear and singular, if you like.
14:46 - Yeah, and I see AH in the chats asking a good question.
14:50 He's saying, "What's the most concerning or surprising thing
14:53 "that you've seen with AI?"
14:54 I mean, you see a lot of people talking about now,
14:58 you know, "Oh, by year so and so,
15:00 "AI could wipe out all human," you know, stuff like this.
15:03 Like, are some of those concerns legit?
15:05 Like, what are things, as someone, Nicholas,
15:07 who's actually been in this space for 10 years,
15:09 consulted in this space, what are like actual things
15:12 that you're concerned about
15:13 versus what are things that you see
15:16 and you're like, you know, laughing at?
15:17 I can't believe people are talking about this.
15:20 - Yeah, I try not to laugh at it
15:23 because some very serious people,
15:25 and people who have thought deeply about this
15:26 do have a lot to say about this and existential threats.
15:31 But I find it a bit odd.
15:33 I think for one thing, it's not really clear
15:36 how those existential threats
15:38 are going to actually materialize.
15:41 I think for one thing, we're really far away from anything
15:45 that would even come close to that.
15:48 And as responsible citizens, we tend,
15:53 or as a race, right, we tend to be very inquisitive
15:58 and we intend to find out what the universe is like.
16:04 And when we find like super destructive technologies,
16:10 we tend not to use it.
16:11 Like, you know, nobody has, you know,
16:13 nuked the earth to complete death, right?
16:15 And it's very unlikely that we're gonna use
16:20 the power of that technology to do that.
16:22 I mean, it just doesn't make sense.
16:23 And likewise with AI,
16:25 I just see that potentially this is vastly overblown,
16:29 that we are going to regulate
16:31 and we are going to be mindful
16:33 about the potential risks in the future.
16:36 Those types of existential threats in future.
16:38 I do think we have a lot more to be really concerned
16:41 about now and even with AI
16:43 and even before generative AI, right?
16:45 I think we've all seen about the divisiveness of media,
16:50 social media, how that plays into all aspects of our lives
16:54 and the impacts of mental wellbeing
17:00 in younger adults or adults, right?
17:05 So there are concerns that are here and present and now
17:10 that we should be talking about more
17:12 and we should be thinking about and working on now,
17:16 but into the future and those existential threats,
17:20 I think that as we see them emerging in the future,
17:24 I definitely think that we're not gonna sort of
17:26 just hand the keys over to Skynet.
17:28 - Right.
17:28 So I understand, Nicholas, you also mentor students,
17:34 particularly school leavers.
17:36 Can you tell us a little bit more about this?
17:38 How could AI be employed in education to better serve kids?
17:42 - Yeah, so I've been privileged to do that for some time now.
17:47 I don't actually have any school leavers at the moment,
17:51 but that's because we're just changing school years
17:54 and so hopefully I'll get that sorted out
17:56 in the next couple of weeks.
17:58 I've been doing this for a few years.
18:00 I kind of, it kind of happened.
18:02 Someone asked me to get involved and do that.
18:05 They thought that I would be a good person
18:07 to come in and speak to some young adults
18:10 who were about to leave school and some had just left.
18:14 I remember that age in my life.
18:16 - So are these people that leave school
18:18 before they technically finish school?
18:20 - Some, and some that are finishing secondary education
18:25 and thinking about what they're doing next,
18:27 thinking about what they want to do in university.
18:30 So it's quite varied, the sort of backgrounds
18:34 that these young adults come from,
18:36 but mainly speaking, I kind of try to focus on folks
18:40 who just don't know what to do with their lives
18:43 and are sort of thinking about,
18:46 and they look at me and what I do as being like,
18:49 my God, I could never do that.
18:51 Like that's just way beyond my wildest dreams.
18:55 I mean, how would I even start?
18:57 And I try to break it down and say,
18:59 no, this is very achievable.
19:01 This is something that you take in small steps
19:04 and how you would get into this space,
19:08 but there's a huge need for these types of skills.
19:11 And they're very varied, right?
19:13 From, as I said, from scientific background,
19:15 mathematics, computer science,
19:17 and it's about applying yourself.
19:19 And I sort of think about it,
19:21 or my best advice is to them is I was very fortunate
19:26 that I think from about 10,
19:28 I knew exactly what I wanted to do.
19:30 I was going to study physics.
19:31 I was going to find out everything I could
19:33 about the universe.
19:34 And I did not know what I was going to do after that.
19:37 Like I really didn't know if I was going to continue
19:39 to be a physicist.
19:40 I had doubts about that,
19:42 but I just wanted to know as much
19:44 about the universe as I could.
19:45 And so I was fortunate.
19:47 And I sort of think about anyone in this area of their lives
19:51 or this time in their lives,
19:52 it's like if you find something that really intrigues you,
19:55 that you're curious about and can drive your passion,
19:58 that helps so much because that's what drives you forward.
20:02 I mean, you have to apply yourself
20:04 at this stage of your life.
20:05 You have to apply yourself at university.
20:08 University isn't necessarily about just learning content.
20:11 It's about learning to learn.
20:13 And if you can do that and apply yourself
20:15 as a lifelong learner,
20:17 you're going to be extremely successful.
20:19 And I guess that's how I come at it,
20:23 the angle I come at it from.
20:24 And not necessarily being their parent is sometimes helpful
20:29 because it's sort of an independent person,
20:31 not necessarily pushing them in one direction.
20:35 - Yeah, that makes a lot of sense.
20:37 Well, Nicholas, we're running up on time here.
20:39 I do want to give you an opportunity
20:41 in 60 seconds or less to give us a little Kaiju kicker.
20:45 Typically, Ryan gives us some piece of,
20:48 I don't want to say investment advice,
20:51 but some insights from the financial/AI world
20:54 since you're the AI guy.
20:56 Why don't you give us a little bit of insight
20:58 into maybe how people can use AI in their lives
21:02 or just any type of little piece of advice
21:05 you want to leave us with?
21:06 - Sure, sure.
21:07 So yeah, I thought about this.
21:08 I saw what Ryan was doing,
21:10 and I'm not going to confirm it from the sort of trading
21:13 or deep investment angle.
21:15 But what I would say is there's a specific plugin
21:18 for ChatGPT called Wolfram.
21:20 Now, Wolfram is an ecosystem of applications
21:25 that started with mathematics.
21:26 I mean, we use mathematics.
21:27 We can back in physics and chemistry, biology,
21:31 and mathematics to build complex models.
21:34 For decades, that's been an application
21:36 that's well-known in academia.
21:39 And that has been extended to Wolfram Language.
21:42 And also, there's a finance module that goes to that,
21:47 that extends onto financial market data
21:50 and able to backtest or to help you build trading strategies
21:55 and backtest them.
21:56 It's still complex, and it takes a lot of effort
22:00 and learning to get into that whole Wolfram Language
22:03 and applications.
22:08 However, with the ChatGPT plugin,
22:12 that's actually, I think, a real game changer
22:14 because you can use natural language in ChatGPT
22:18 to ask questions like, you know,
22:20 how can I go about starting to build a trading strategy?
22:23 I'm interested in these underlying assets,
22:26 and I'm looking at these signals.
22:28 I'm thinking about these signals.
22:29 Help me understand that.
22:30 And then please use Wolfram, the Wolfram plugin,
22:33 to help me extract some market data
22:36 on a particular equity, for instance,
22:38 and help me build some technical signals on that.
22:43 So I think even though that this area was very, very,
22:47 still very technical and complex,
22:49 with ChatGPT, that's actually gonna be an emerging way
22:53 to help develop your skills and think about that
22:55 and develop ideas, and then using this huge back system
23:00 of really incredible computational power
23:03 and back-testing engine.
23:05 So I see that as something emerging.
23:08 I have played around it a bit just to come into today,
23:13 and it's a bit bumpy.
23:14 It doesn't always work right now,
23:16 but I think it's still early.
23:17 So if you're on the early adopter of this,
23:20 you might be in a good position in future
23:22 if you play around and maybe start using
23:24 this Wolfram plugin.
23:27 - Yeah, that's interesting.
23:27 I actually remember, I mean, this site, Wolfram Alpha,
23:31 I remember, it's been around for a while, right, Nicholas?
23:33 I remember like five, maybe even 10 years ago,
23:37 using it, being able to ask it questions,
23:39 and it would give you answers,
23:40 particularly with math stuff that Google couldn't.
23:44 I thought it was like the coolest thing ever.
23:46 And now that I'm thinking about it,
23:47 ChatGPT is kind of like the next iteration
23:49 of Wolfram Alpha, but now you're saying
23:51 Wolfram Alpha is using ChatGPT.
23:54 But again, this site's been around for a while
23:56 and been able to do some cool things for a while.
23:58 - Yeah, and I think it has been improving over time.
24:02 And ChatGPT is not very good with numbers, right?
24:05 It's good with answering language questions,
24:10 but it isn't very competentially capable,
24:13 but the whole Wolfram language aspect, that is.
24:18 So you can merge this language, competent language,
24:22 if you like, with computational capability
24:25 in the background, and I think
24:26 that could be a powerful combination.
24:29 - Got it.
24:29 Well, Nicholas, it's been great to get to meet you today
24:32 and to learn more about your role at Kaiju.
24:35 Hope you have a great rest of your week.
24:36 Maybe at some point in the future,
24:37 we gotta get you and Ryan on here
24:39 at some point at the same time.
24:40 I don't know if this show's ever had that much
24:42 brainpower on the screen at one time,
24:44 but Nicholas, enjoy the rest of your--
24:46 - That would be cool.
24:47 - All right, let's do it.
24:48 All right, Nicholas, enjoy the rest of your Tuesday.
24:49 We'll chat soon.
24:50 - Okay, take care.