Investing Reimagined In An AI World

Forbes

This conversation between investing industry experts looks at how AI will upend the industry and what opportunities it presents for savvy professionals. This conversation took place at Imagination In Action’s ‘Forging the Future of Business with AI’ Summit in April 2024.   Subscribe to FORBES: https://www.youtube.com/user/Forbes?sub_confirmation=1  Fuel your success with Forbes. Gain unlimited access to premium journalism, including breaking news, groundbreaking in-depth reported stories, daily digests and more. Plus, members get a front-row seat at members-only events with leading thinkers and doers, access to premium video that can help you get ahead, an ad-light experience, early access to select products including NFT drops and more:  https://account.forbes.com/membership/?utm_source=youtube&utm_medium=display&utm_campaign=growth_non-sub_paid_subscribe_ytdescript  Stay Connected Forbes newsletters: https://newsletters.editorial.forbes.com Forbes on Facebook: http://fb.com/forbes Forbes Video on Twitter: http://www.twitter.com/forbes Forbes Video on Instagram: http://instagram.com/forbes More From Forbes:  http://forbes.com  Forbes covers the intersection of entrepreneurship, wealth, technology, business and lifestyle with a focus on people and success.

Transcript

00:00 Okay, who's excited after lunch to hear some panels about investing?

00:06 So, I'm Boaz Thackler.

00:12 I work with Link Ventures as an investor.

00:15 My claim to fame is that I get to speak to John Werner every so often.

00:22 Who likes John Werner?

00:24 Who doesn't?

00:25 Who doesn't?

00:26 Yeah.

00:27 And I'm really excited about this panel because, A, because we have this really amazing high-caliber

00:34 panelist, but also because I both invest, so any AI advancement can help me, and I'm

00:44 always looking for some advancement.

00:48 And B, because I'm investing in companies that are disrupting the investment industry.

00:56 One of the latest is Samir's company, Farsight.

00:59 So this is very exciting to hear all your different perspectives.

01:04 So why don't we start with some intros?

01:07 Lauren, you can start.

01:08 Hi, everyone.

01:09 My name is Lauren Clements.

01:11 I lead up innovation and emerging technology work at Prudential.

01:15 Nice to see you all today.

01:18 Hi.

01:19 Can you hear me okay?

01:22 Perfect.

01:23 Samir Dutta, co-founder and CEO of Farsight AI.

01:25 We automate and deliver proprietary AI agents that help streamline workflows across financial

01:32 services and insurance.

01:34 Prior to this, I was a private equity investor at a firm called General Atlantic.

01:38 Hi.

01:39 I'm Lisa.

01:40 I lead an AI technology center at Fidelity for about the last five years.

01:47 Before that, I led the investing team at Betterment, one of the, I guess, original robot advisors.

01:54 And I'm really happy to be here today.

01:56 Good afternoon.

01:57 My name is John Wu.

01:58 I'm president of Avalabs.

01:59 Avalabs is the team behind Avalanche, a layer one blockchain.

02:04 And we actually help a lot of financial services firms, JP Morgan, KKR, Apollo, to name a few,

02:11 create workflow that's more efficient on a blockchain.

02:14 Prior to being an operator, I was an investor, just like everyone else on this panel.

02:18 I was a tech investor at my own fund, at Kingdon Capital, and at Tiger.

02:26 Amazing.

02:27 So let's start with pessimism.

02:30 We'll go from risks of investments in the AI world to opportunities.

02:37 So future of work.

02:39 Are we afraid that AI is going to replace investors, some portion of workers, or is

02:47 it somehow going to balance itself?

02:49 Lauren, do you want to address this?

02:53 Well, I definitely wouldn't say that I'm afraid of it.

02:57 But for sure, it's going to change how we do a lot of things.

03:03 And I think it's actually kind of an extension of the changes that have already happened

03:07 a lot in the investment space, right?

03:10 So for a long time, we've been using algorithms, we've been using technology to do the work

03:16 with massive amounts of data, with real-time computations, with extremely high-speed trading.

03:22 All of those things have continued to get better and better.

03:25 And it's changed the profile of what you need to do in the investment space, right?

03:31 I think we'll continue to see that.

03:33 It'll hit different positions differently.

03:35 But at the end of the day, if you're replacing someone who's making a decision on what to

03:41 buy or sell with a program that was created by a human who's decided what the goals of

03:48 that bot should be, you're kind of -- there's still that person who's directing it.

03:54 And at the end of the day, there's no right answer to investing, right?

04:02 There's an angle to it about what will other people do, right?

04:07 So if I'm right about what's going to happen, it's meaningless unless I know how the stock

04:11 is going to move or how the investment is going to change.

04:13 And that depends on other people.

04:15 So I think it's going to be a little bit more subtle.

04:19 Okay.

04:20 So you're not a strong believer in kind of like AGI, machines going to determine what

04:27 the machines will decide to invest in?

04:30 Well, they can.

04:31 But at the end of the day, who are they buying and selling from?

04:34 If it's all machines, then who has the edge?

04:37 And so it's going to, I think, will progress the same way we've seen with other technologies,

04:42 like extremely high-speed algorithmic trading, where you'll see some odd behaviors at certain

04:48 times.

04:49 There'll be more breakthroughs.

04:50 There'll be fewer breakthroughs.

04:51 But I think it's not as simple as just computer versus human.

04:56 Yeah.

04:57 Lisa, do you want to address this question also?

05:01 Yeah.

05:02 I'm waiting for my little AI agent right here to help me with everything that I do.

05:10 I think I'm embracing this wave.

05:14 Something definitely is different today, AI versus even two years ago.

05:20 I see AI now as just another kind of computer.

05:26 And that should be really useful for what we do as investors.

05:31 Like the entire value chain.

05:34 So I'm totally embracing this wave of AI right now.

05:38 And with no fear.

05:40 I don't think it will replace humans.

05:42 Humans always have a place.

05:45 We create AI, we destroy AI, and then we create it again with better AI.

05:50 So I have more optimism.

05:53 Oh, yeah.

05:54 I thought it was going to be a little bit more pessimistic.

05:57 Lisa, I'll continue with you.

06:01 Thinking about open innovation versus closed, we're seeing a lot of power concentrated with

06:10 large organizations.

06:12 Many of you work for large organizations.

06:19 Should we be concerned?

06:21 What are the benefits of open source versus closed?

06:25 If you take Microsoft versus Meta?

06:30 I'm really thankful for Meta.

06:32 I'm all about open source.

06:34 The more open, the better.

06:36 Because that's how we progress as humanity.

06:42 And I think there's still going to be differentiation.

06:46 Because the models are going to be commoditized.

06:50 They're going to be open source.

06:52 And when you put all the brain power of all humanity in open source, it's going to win

06:56 over any closed source models.

07:01 I think the differentiation comes from your own data.

07:07 And firms, everybody has their own data.

07:09 Firms have their own data.

07:10 You as individuals have your own data.

07:13 And the models apply to your own data.

07:16 That's democratizing and that's powerful.

07:18 And then the second stage is really how do you represent your own data?

07:22 There's IP there.

07:25 And that's not going to be commoditized away so easily.

07:29 Fair enough.

07:33 John, as both investor, have probably a very interesting view on AI and in the crypto space.

07:43 What is your take on open source?

07:46 You're not going to ask a blockchain guy or a crypto person whether they believe in open

07:50 innovation or not.

07:51 I obviously am a big proponent of open innovation and the benefits of open innovation.

07:56 I'm going to start by going back to your first question.

07:59 I think AI is very important to the investment process and will change how you invest and

08:05 will make it easier to do many, many things.

08:09 That's been the evolution of technology, improving how people work.

08:14 But I think open innovation in the hedge fund investing world or in the VC world may actually

08:21 change business models in terms of how a fund is set up and how it works.

08:26 There are companies out there trying to do this and it's still very early stage.

08:30 I don't know how many people know the company Numeri.

08:34 And what Numeri is basically doing is it is trying to replicate quant like returns that

08:40 generally speaking were reserved for places that have so much scale that they can hire

08:45 the 200 new brightest, smartest PhD kids coming out of MIT.

08:50 And they really had a monopoly over the talent.

08:53 But if you can create a mechanism where all the people in the system are aligned with

09:00 users of the quant models, data scientists, data contributors, and people who are going

09:06 to also fine tune those models, then I can see a world where you're crowdsourcing data

09:13 science and data analytics to the point where it benefits everyone.

09:17 And as a user, the disruption is you actually who cannot get into Renaissance or Two Sigma

09:24 because you're an individual or because you don't want to pay the 220 or two whatever

09:29 they charge, you're actually able to get equivalent type models that you can work yourself

09:36 in by using some sort of payment and some sort of functionality like a token or something

09:43 and get the benefits of that kind of quant type investing.

09:48 So you have to believe that you can crowdsource thousands of people, maybe not full time,

09:54 but getting thousands of part time data scientists and data contributors will be as good or close

10:02 to or maybe even better than that small closed niche of 200 or 300 PhDs being monopolized

10:11 by some big player.

10:14 Thank you.

10:15 Let's move on that note, let's move to a little bit more opportunities in investments with

10:21 AI.

10:22 As a venture capitalist, both investing and trying to support the portfolio companies

10:28 to grow, one thing that we're struggling with a lot is how do we penetrate the legacy players

10:38 as vendors?

10:39 How do the portfolio companies sell to vendors?

10:43 I think Samir, that would be very interesting to hear your experience.

10:48 Yeah, so I can speak to this a little bit from the perspective of someone who's a smaller

10:54 company on the outside and a little bit of how we found success working with legacy players

11:00 that -- and some other large players.

11:04 I think it really comes down to this idea of crawl, walk, run.

11:09 This is nothing new.

11:10 Every time a new technology comes out, that's how enterprises think about adoption.

11:15 And the real challenge is both from the vendor perspective and then also from the perspective

11:20 of the internal champion within the organization, it's your job to together figure out that

11:26 roadmap that gets you from low risk, low opportunity use cases, but then have a very clear roadmap

11:34 that iteratively adds layers of complexity, layers of risk, and ultimately unlocks that

11:38 transformational value.

11:41 And so one of the biggest things I'd say is -- it sounds very easy, right?

11:46 Crawl, walk, run.

11:48 Before it becomes difficult is at each phase of crawling and walking and running, you need

11:52 to align to three stakeholders, right?

11:54 You have the business person who needs to see immediate business value and immediate

11:57 ROI even though you're not allowed to touch proprietary data just yet.

12:01 And secondly, you have the IT people who need everything to be auditable, traceable, high

12:06 security compliance.

12:08 And then lastly, you need to appease the vendor as well, right?

12:11 It needs to be a project that's repeatable enough and scalable enough for it to be worth

12:16 their time.

12:17 So what I found is two companies from the outside can look exactly the same, and the

12:22 roadmap of crawl, walk, run that you come up with is totally different based on who

12:26 the players are and who you're talking to internally.

12:28 >> That's very interesting.

12:30 We'd love to hear now the other side of the equation, the perspective of Prudential, other

12:38 legacy players.

12:39 How do you see this?

12:41 >> Yeah, so I guess my first message to this group is we want to use the newest, latest,

12:48 greatest technologies, AI, all of it.

12:51 And so we want you to know that there's certain things that make that easier for us to do

12:56 or harder for us to do.

12:57 And Samir touched on a whole bunch of them.

13:01 The more that you can tell us about the quality of the output of what we're getting from the

13:08 tools, the easier it is for us to move ahead; right?

13:11 So would I pick a tool that's right 90% of the time, but I have no way of judging the

13:18 10% or a tool that's right 60% of the time, but it's really good at telling me, hey, this

13:24 answer, this is a really good answer.

13:27 I would always take that second one, because I can work operationally around those other

13:35 instances; right?

13:36 Same thing with explainability, oversight, like building into the tool that walk -- what

13:43 was it?

13:44 Crawl, then walk, then run.

13:45 I'm getting ahead of myself.

13:47 If you can build that into the tool and give us a way to fine-tune or say, hey, this is

13:53 what's working, then we can start to take the human out of it faster.

13:57 So -- and it doesn't necessarily have to be, like, you know, a huge part of the product,

14:02 but just thinking that way from the start, hey, what's going to make someone like, you

14:07 know, Lauren or Lisa, like, able to actually put this in the hands of people and start

14:12 using it?

14:14 >> So what do you think is the role of proprietary data, for instance, in that aspect?

14:21 >> Yeah.

14:22 So proprietary data, you know, as we think about what do we build ourselves, what do

14:27 we partner with people on, what do we buy off the shelf, data is a really important

14:32 angle to that; right?

14:33 So I think there's an understanding amongst most large companies that proprietary data

14:41 is like gold in this new AI community.

14:45 So you're going to be very sensitive to wanting to maintain that proprietary angle to anything

14:56 you use.

14:57 So I guess I would add that to my list as well.

15:00 So it's kind of implied learnings from data, all of that stuff, you're going to need to

15:05 be able to talk to that to, you know, people at large enterprises, especially highly regulated

15:11 ones.

15:15 >> So I would love to move a little bit to talk about the product themselves.

15:21 Lisa, maybe you can talk a little bit about your perspective on what is the winning value

15:29 proposition?

15:30 Is it just productivity?

15:33 You know, what Klarna just did with its call centers, reducing costs 40%?

15:39 Or are we more excited about valuable insights that create value?

15:48 >> I'm excited about both.

15:50 But I think productivity is a no-brainer.

15:54 That's the first thing that you're going to tackle.

15:56 Right?

15:57 Go use a computer.

15:59 Go use it.

16:00 Right?

16:01 Go use it for good.

16:02 Go use it for your tasks, for everything that you do in your entire workflow.

16:06 Right?

16:07 In finance.

16:08 So productivity, yeah, I think it's so much easier to build with AI now.

16:14 Like I actually think, like, you know, we're underestimating how much productivity gain

16:20 we can get.

16:21 You know, sometimes people throw numbers like, you know, we'll be 40% more productive.

16:27 I actually think it's like multiples.

16:30 Like it could be like 2X, 3X more productive.

16:33 I don't know.

16:34 Again, maybe too optimistic.

16:36 >> We're going to talk about the investment field.

16:40 Your product is simply your returns.

16:42 I don't really care if it's one guy with the great gut instinct or they have the most sophisticated

16:48 generative AI model and that proprietary data sets.

16:51 Their output is not high returns.

16:53 I don't really care.

16:54 So in this field, it is about returns.

16:58 And I don't think generative AI can be equally used by different functions, different stages

17:06 of investment in different industries.

17:08 For your job, early stage investing, you need to look at that entrepreneur in the eye.

17:12 Does he have that grit to do what is necessary to give your portfolio high returns?

17:17 Or public equity investing even.

17:20 There are fields like, you know, biotech or pharmaceuticals where it's very binary whether

17:28 that drug is approved or not and effective or not.

17:31 I don't think AI will help that.

17:32 What AI will definitely help is improve your process, lower your cost probably.

17:37 Somehow also the inference in all of this gives you some insight so you can actually

17:43 make better investments.

17:44 And that's the hope for most of investing.

17:48 That's a great insight.

17:49 And I think it's a great segue also to my next question about black box versus explainability.

17:59 Would you or your LPs or your investors accept black box that generates returns without explaining

18:07 it or you do need to find some explainability because it's very risky?

18:14 I live through it.

18:15 You know, I started my career as a technology investor at one of the most well-known funds

18:19 in the world.

18:20 We were long, short, fundamental tech investors at Tiger.

18:24 And that was all good until the quants came and they were able to replicate the returns.

18:30 And there are actually very few fundamental discretionary long, short managers today unless

18:35 you're being grandfathered in with large asset bases because money is still sticky.

18:41 So it's going to change.

18:44 But again, it's really just about returns.

18:46 If there's a new methodology, a new gizmo that gives people better investments, endowments,

18:52 pension funds, they'll migrate that there.

18:55 Samir, what is the feedback you're getting from your customers like black box and also

19:03 productivity versus...

19:06 It's a great question.

19:07 I've already learned in the last five minutes.

19:08 Great thoughts here.

19:10 I think the number one takeaway is that productivity to me is more a byproduct of having built

19:18 a good product.

19:19 So inherently, if you're delivering AI to people, investors, in the ways that they're

19:24 already working, in PowerPoint, Excel, other systems that they might be using, that just

19:30 means the productivity is letting you know that people are using it and you're making

19:34 their lives easier.

19:35 I think where the real differentiation of an AI product comes in is in the ability to

19:41 consistently deliver those better returns.

19:45 And that predominantly, a little different in the hedge fund world, but certainly in

19:48 private equity, a lot of that know-how just comes from the senior people who have done

19:54 pattern recognition.

19:55 A lot of the information, frankly, is not written in a document somewhere that you can

19:59 just feed into a RAG system.

20:01 It lives up here.

20:03 And to my knowledge, AI is not yet able to read people's minds.

20:06 And so the real challenge that we've been faced with is how do we accurately get all

20:11 the expertise that someone like John has into an AI system?

20:15 And the answer is, it's the product UX, whereby you allow partner or analyst, doesn't matter,

20:22 you allow them to prod the output of an AI using natural language and distilling their

20:27 thoughts specifically to the AI.

20:30 Can you share an example?

20:32 I know something that you can share from what you're doing.

20:36 Is there any example you can give us?

20:39 Yeah.

20:40 A great example is actually a private equity company we recently worked with.

20:47 They're evaluating a tech market that has two main players.

20:52 And they have a deal team staffed of five people on it.

20:56 And nobody thought to look at -- when they looked at Uber and Lyft and did a whole analysis

21:01 on basically an identical market that was also a duopoly.

21:05 And the AI integrated with their systems was able to surface a five-page report with thorough

21:10 analysis going through retention curves and all these things Boaz, I'm sure, loves going

21:13 through that was not -- the AI was not asking the investors to say, hey, make a decision

21:20 one way or another.

21:21 But it surfaced these very valuable insights that none of those five people had even been

21:25 at the firm when they made the Uber and Lyft decision.

21:29 It's like having unlimited number of first-year analysts.

21:33 But if you ask them to do the job of the more senior folks, there's a lot you need to teach

21:37 them how to do, right?

21:38 Exactly.

21:39 Yeah.

21:40 Okay.

21:41 I think now we can move to a little bit more of a generic question.

21:47 How do we -- what are the most significant changes?

21:52 John, maybe you can address that in a pre versus post-gen AI investment world.

21:59 Samir and I were actually talking about this and I think it was written up in either New

22:03 York Times or Wall Street Journal.

22:05 Literally, and I think Lauren just mentioned it, the function of the first-year analysts

22:09 can be completely replicated now through AI.

22:13 And I think that is one of the most startling changes because most of these kids who go

22:20 to Wall Street either at a fund or in a banking, you know, it's about -- it's not just about

22:28 learning and developing the tools and doing the grunt work.

22:32 It's also ingesting in the culture of the leader, of the founder, of the bank or whatever

22:39 and basically able to project that type of dealmaking or that type of investment strategy.

22:45 So I think the biggest change is the functionality is going to be low-level functionality will

22:52 be taken out of the equation.

22:55 But how do you actually train people who now are skipping that rung and teach them the

23:01 culture and the style and the grit that you need to be a good investor?

23:07 I think that also partner level can be replaced very easily.

23:11 Think about it.

23:12 Let's go straight to partner now.

23:15 I rely on that.

23:16 Lisa, what is your -- you have to think about strategy.

23:21 How do you think about this, the world before and after?

23:25 What are strategies that you're implementing already to face that?

23:29 I think about this a lot, like pre and post-Gen AI.

23:34 You know, well, pre-Gen AI, like I remember I spoke at this conference last year.

23:41 I took an Uber ride home and at the end of the ride, my Uber driver said, "Oh, I wish

23:46 the ride was longer because I want to talk more about chat GPT."

23:51 Right?

23:52 Pre-Gen AI, that couldn't have happened.

23:54 So that was one big -- this is more excitement, I think, you know, more imagination of what

24:01 can be done.

24:03 In terms of like concretely, pre-Gen AI, I think AI was really used as a prediction algo.

24:10 I want to predict something with AI.

24:13 And I think, you know, adoption in finance tend to be low because of that reason.

24:18 Because it was just used for prediction.

24:21 Sometimes the prediction is black box, so we can't use it.

24:24 Sometimes it's wrong, so we can't use it.

24:26 Because the cost of error is very high.

24:28 Right?

24:29 We're not, you know, we're not in a creative field.

24:31 We're not writing poetry for fun.

24:33 You know, we're investing to make returns.

24:36 And you know, we need precision.

24:38 So that's pre-Gen AI.

24:41 Post-Gen AI, like I said, I think Gen AI, post-Gen AI, AI is just another computer.

24:46 Right?

24:47 People throw like terms like semantic kernel.

24:50 Well, that sounds like a computer.

24:51 Right?

24:52 A kernel of a computer.

24:53 Semantic.

24:54 A computer I can talk to.

24:56 So post-Gen AI, I see AI is just a computer that you can talk to.

25:02 And always can talk to.

25:03 You can talk to your data.

25:05 You can ask your data many questions.

25:06 You'll come back and you can ask it again.

25:09 You know, you can look at different types of data.

25:11 And through that iterative process of talking to your data and talking to your new AI computer,

25:17 you're going to invest better.

25:18 And you're going to generate outflow.

25:20 And mitigate risk.

25:22 All in one.

25:26 >> So I heard upstairs one of the panels gave like a wrapper question of give me -- and

25:35 I really liked it, so I'm going to adopt it.

25:37 So one to three words from each of you about investments ten years from now.

25:44 Lauren, you can start.

25:47 >> Oh, you're going to stump me.

25:50 Want to come back?

25:53 >> Okay.

25:54 I know there's a better word for this.

25:57 But I'm thinking like all knowing.

26:02 All knowing.

26:03 Can I qualify it or I just have to leave it at that?

26:06 >> You can qualify it.

26:07 >> I can qualify it?

26:08 I think the real differentiator in hedge funds, private equities, any sort of institutional

26:10 investor is going to be the proprietary knowledge that they've built up over decades of investing.

26:15 That's always been the differentiator.

26:17 And that's not going to change with Gen AI.

26:18 What will change is the number of people you need to execute on those visions.

26:24 And the efficiency at which you're able to draw knowledge from the past.

26:28 >> Okay.

26:29 Three words?

26:30 Okay.

26:31 >> One to three.

26:32 >> One to three.

26:33 >> One to three.

26:34 >> I think open.

26:35 Like open data.

26:36 You know, like a way to aggregate all your data.

26:40 Personalize.

26:41 You know, personalize investing for everybody.

26:44 And the last word is democratizing.

26:47 Democratizing returns for everybody.

26:50 >> I think in ten years, there's going to be two types of investing firms.

26:55 One is that closed source and it's some company that has just got a bunch of proprietary data

27:02 and proprietary models that give you extreme outsized returns.

27:06 And it's very hard to replicate.

27:08 And then the other world will be more of an open world where data is easily transparent.

27:13 The models and where they came from, the efficacy of the models are transparent.

27:18 And then it will be more do it yourself, just like everything turns into.

27:21 So the average individual will be able to contribute to this ecosystem somehow and also

27:25 get access to these models and replicate returns on their own.

27:32 >> I'll say information arms race.

27:35 >> You want to elaborate?

27:37 >> Just I think we'll to your point before about like is it going to be computers versus

27:43 computers, computers versus people?

27:45 I think it's going to be all of the above.

27:48 And then how the dynamics play out.

27:50 Hopefully we'll hit another balance.

27:53 >> Okay.

27:56 Thank you so much.

27:57 On that note, we can move to the next demo.

28:00 Thank you so much.

28:01 [ Applause ]

28:01 [ Silence ]

28:02 [ Applause ]

28:02 [ Silence ]

28:03 [ Applause ]

28:03 [ Silence ]

28:11 [ Applause ]

28:12 [ Silence ]

28:13 [ Applause ]

28:14 [ Silence ]

28:15 [ Applause ]

28:16 [ Silence ]

28:16 [ Applause ]

28:17 [ Silence ]

28:18 [ Applause ]

28:19 [ Silence ]

28:19 [ Applause ]

28:20 [ Silence ]

28:21 [ Applause ]

28:22 [ Silence ]

28:22 [ Applause ]

28:23 [ Silence ]

28:23 [BLANK_AUDIO]

Category

Transcript

Recommended