Latest / The Marketing Meeting with Itir Eraslan / 40. Debunking AI Hype: A Reality Check with Dr. Jeffrey Funk
Transcript
- 0:00Dr. Jeffrey Funk: Sam Altman is not an expert on marketing.
- 0:02He has no idea what marketing people do.
- 0:05Sam Altman is trying to sell OpenAI.
- 0:09And he's not really trying to sell you and use it.
- 0:10He's trying to sell investors on saying it's really big.
- 0:15Because that's how you make money in today's world.
- 0:25Itir Eraslan: Hi, this is the Marketing Meeting and I'm your host, Itir Eraslan.
- 0:30Every two weeks I meet with experts and we talk about topics related to
- 0:34brands, marketing, and businesses.
- 0:36We sometimes add random lifestyle topics too.
- 0:39I hope you enjoy the show.
- 0:46Welcome to the Marketing Meeting Podcast.
- 0:49I'm happy to host Dr.
- 0:50Jeffrey Funk today.
- 0:52He's an expert on the diffusion of new technologies.
- 0:55He has spent over 40 years as a professor and technology consultant
- 0:59advising companies and universities.
- 1:02You might have seen him sharing his insights in MarketWatch, Fast
- 1:06Company, and also in my case, I see him a lot on LinkedIn.
- 1:10Uh, his upcoming book, Unicorns, Hype and Bubbles will be released in October.
- 1:15Welcome, Jeffrey.
- 1:16Dr. Jeffrey Funk: Thank you.
- 1:17Pleasure to be here.
- 1:19Itir Eraslan: Uh, thanks so much for joining me at this
- 1:22very early hours of the day.
- 1:24Um, this episode is different for me because many of the guests that I've
- 1:29been interviewing for AI marketing and brands are usually sharing an optimistic
- 1:35perspective around the technology and AI.
- 1:38Uh, however, I know that you're a voice of reason and you
- 1:42balance skepticism and optimism.
- 1:45Around the technology, uh, in a good place, in my opinion.
- 1:49So I'm really excited to have this conversation with you.
- 1:53Like you were one of the first person to recognize the potential for smartphones
- 1:57during the early two thousands.
- 2:00Could you talk a bit about that?
- 2:01And, and while you are talking about it, I would like to also understand
- 2:05how do you evaluate the potential and sustainability of a new technology?
- 2:11Dr. Jeffrey Funk: Okay.
- 2:12Well, I was, uh, got interested in Japan, uh, way back in the 1980s.
- 2:18Japan is number one.
- 2:19I was trying to understand how Japan was doing well.
- 2:22How could they develop new products quickly?
- 2:25So I moved to Japan.
- 2:27Full time in 1996 and was looking at a number of products, but one of them
- 2:33was mobile phones, and it was just by accident then that when NTT DoCoMo
- 2:38introduced their iMode service in February of 1999, it did much better
- 2:42than any of the services in Europe.
- 2:44Uh, and it was the only country with successful service until Korea
- 2:48launched them a few years later.
- 2:50Uh, but you know, the iPhone didn't really come out to 2007.
- 2:53It starts, it's going to be 2008.
- 2:55So I was visiting lots of Japanese companies.
- 2:58My Japanese was pretty good.
- 2:59I probably visited hundreds of Japanese companies, asking them about
- 3:03their various mobile phone services.
- 3:05And one of the things I realized is that a lot of it revolved around, you
- 3:09get better phones, you get a better processor, more memory, a better display,
- 3:15the resolution in color, or just.
- 3:19Uh, overall better network speeds, those things enabled companies
- 3:23to introduce new services.
- 3:25So, uh, there was a very tight link.
- 3:28Your companies were very specific about what they were introducing, you
- 3:32know, why they were introducing it because the phones had gotten better.
- 3:35So that allowed them to give better services and the rest of
- 3:38the world wasn't able to do so well because the standards weren't met.
- 3:42Uh, set very well.
- 3:44The, uh, the network operators and T Docable and others kind of determined
- 3:48what would be in the phones to make sure that the phones worked right?
- 3:52Uh, and Nokia and other, uh, leading phone makers at the time didn't like that.
- 3:57So, uh, things didn't work out.
- 3:59So that was a very, um, very educational for me in terms of, uh, How markets grow
- 4:05and, uh, the relationship between, uh, the underlying technology and the services
- 4:10that companies were able to offer.
- 4:11Itir Eraslan: In 2007, you taught a unique course on the
- 4:16economics of new technologies.
- 4:18And then that's also the start about being an early critic about,
- 4:23uh, the startups and also the venture capitals, in my opinion.
- 4:27Dr. Jeffrey Funk: Yeah, I was very excited about that.
- 4:29Uh, what happened in Japan and what I learned.
- 4:31So I wanted to talk about that in this course.
- 4:34Uh, and actually started like 2009.
- 4:37And so I started talking about lots of different technology.
- 4:40And I just, every year I was increasing the number of technology I talked about.
- 4:44And I had a lot of students, I had, you know, 50 projects a year being
- 4:47done, group projects being done.
- 4:50And so I would learn a lot from the projects because I would require them to
- 4:53go beyond what I talked about in class.
- 4:55So they would talk about something new and then I would incorporate
- 4:57that in the next semester's course.
- 4:59And so I covered just an immense number of technologies.
- 5:03Uh, I taught this up until 2016.
- 5:06And, uh, by then I was doing big data and there was some mentions of AI.
- 5:11Um, but over that time from like 2009, 2007 years, I was teaching it.
- 5:16I became a pessimist because I began to notice that a lot of the
- 5:19technology I was talking about hadn't really gotten much better.
- 5:23Then when I'd previously taught it, so things like nanotechnology
- 5:27and superconductors for energy and bioelectronics and Toronto's kind.
- 5:31So I, I noticed they weren't getting better.
- 5:33So, you know, and if something didn't get better, then it wasn't going to diffuse.
- 5:37You know, if it's not being used, it has to get a lot better
- 5:40before it's going to be used.
- 5:42So I began to become a pessimist.
- 5:43I also at the end tail end of that 2016, uh, right after that,
- 5:48I noticed that a lot of these.
- 5:50Startups weren't making money.
- 5:51There was a lot of anecdotal data about losses.
- 5:54So I began to become more pessimistic and, uh, the list of technologies for which
- 5:58I was optimistic became less and less.
- 6:01And that continued after I left NUS in the late 2010s.
- 6:07Itir Eraslan: You talk a lot about value and progress when it comes
- 6:11to evaluating the technologies.
- 6:13Uh, I think that's something that's really stuck into my mind.
- 6:17So what's the value and what's the progress that technology
- 6:20is progressing and what's the value that's adding to our lives.
- 6:24And from there, I would like to touch base on the marketing side of things.
- 6:28Dr. Jeffrey Funk: Yeah, well, value.
- 6:31What, what, what are the customers getting in the first thing you want to
- 6:33know is, are the customers using it?
- 6:35And you want to get an idea of there's always going to be somebody
- 6:39using it almost always, uh, you know, maybe some like technologies
- 6:43are too far in the future.
- 6:44Einstein Rosenberg bridge or something, you know, uh, It's too far away that it's
- 6:49not being used, but most technologies are.
- 6:51So you can always find some people using, and then you want to find
- 6:53out what benefits are getting, how much money they're spending.
- 6:56Uh, and then you ask the question, okay, they're using this technology.
- 7:01They have these problems with X and Y, but they like Z.
- 7:05Okay.
- 7:05Is X and Y getting better?
- 7:06Where's the progress?
- 7:08You have to be very specific.
- 7:09Too many people think.
- 7:11They just see all the products coming out and they just see,
- 7:13Oh, there's so much progress.
- 7:14So much is happening.
- 7:15And they hear all these statements by, uh, the tech bros just saying,
- 7:19Oh, this is going to happen.
- 7:20That's going to happen.
- 7:21And they mistake that for progress.
- 7:23They think that, Oh, there's so much talk about it.
- 7:25There's so much progress.
- 7:26So wait, you got to be very specific.
- 7:28about the performance metric you're concerned with and
- 7:32whether there's progress in the relevant performance metric.
- 7:35Itir Eraslan: Uh, in March, 2024, uh, Sam Altman, the CEO of OpenAI said,
- 7:42AI will handle 95 percent of marketing work done by agencies and creatives.
- 7:48He also added, uh, it's going to be easy.
- 7:51Instant and almost at no cost.
- 7:54He's giving it five years, give or take maybe slightly longer.
- 7:59Uh, this is a bold statement and a lot of people in marketing
- 8:02world has been discussing it in non marketing world as well.
- 8:06What are your thoughts about this prediction?
- 8:08Dr. Jeffrey Funk: Well, Sam Altman is not an expert on marketing.
- 8:11He does not have spent his life doing marketing.
- 8:13He has no idea what marketing people do, you know, and, uh,
- 8:17Mira Murati is the same way.
- 8:19She says, wow.
- 8:20It's going to replace all creatives, but maybe those jobs shouldn't
- 8:23have been there first place.
- 8:24You see that they display this lack of knowledge about the intricacies of work
- 8:30and you you can say that the marketing or you know, a lot of stuff for superfluous,
- 8:34but They're there for a reason.
- 8:36Companies do them for a reason, right?
- 8:39So, and there's a lot of details in there they do for a reason.
- 8:43So you have to understand those reasons.
- 8:44You can make these forecasts.
- 8:46So Sam Altman is trying to sell open AI and he's not really
- 8:49trying to sell you and using it.
- 8:51He's trying to sell investors on saying it's really big.
- 8:55Because that's how you make money in today's world.
- 8:58It used to be that you started a business, you satisfied some
- 9:01customers, then you made it a little better, you satisfied some more
- 9:05customers, and the revenues grew.
- 9:07And after a few years, people said, Hmm, that's kind of a good technology, you
- 9:10know, and there's a lot of revenues there.
- 9:12That's a good business.
- 9:13Yeah, I think we should increase the value of that, that company.
- 9:17Uh, you know, either in the stock market or private valuations, but Sam
- 9:21Altman and others don't like that.
- 9:24They want to convince you that it's worth billions, you know, in trillions.
- 9:29I mean, people throw around these terms all the time, uh, trillions of dollars.
- 9:33And, uh, so Sam Altman writes these, these articles, like a few years
- 9:36ago, he wrote one that said the Moore's law of everything, you know,
- 9:39houses and everything was going to get cheaper because of check.
- 9:43And when people say these, these things.
- 9:44Things, these outrageous things that there's no basis of fact.
- 9:48The logical response is this guy doesn't know what he's talking
- 9:52about, but for some reason, these investors have bought into this.
- 9:57Which of course should lead to the end that well maybe these investors
- 10:00don't know what they're talking about because You know this isn't happening.
- 10:05I mean what you want to do is focus on exactly what's happening with
- 10:08chat GTP What exactly is happening?
- 10:11Who's using it and why?
- 10:13There are people using it and there are cases where Seems to be offering some
- 10:17value, but you want to focus on that.
- 10:19You don't want to stay at this very broad level of Sam Altman
- 10:23says this is going to happen.
- 10:24And, and therefore we're going to believe him.
- 10:27No, you focus on specifically what's going on right now.
- 10:30What users value users are getting right now, where's the actual progress
- 10:34and you make your own forecast.
- 10:36Don't just believe someone's forecast.
- 10:39Try to understand their logic and repeat their logic and be able to repeat
- 10:43their logic and make their argument.
- 10:47Itir Eraslan: I think the reason why many people believe in him, well, there
- 10:51are also some other obvious reasons why people believe in him, but marketing,
- 10:56especially in the B2B world, marketing is seen as content marketing, and we
- 11:01have head of content, VP of content acting as if they are the marketing
- 11:05managers or directors of the company.
- 11:08There is.
- 11:09You know, when you think about generative AI, it's about creating content.
- 11:12And if you think that marketing is creating content, then you would assume
- 11:15that, you know, chat GPT or generative AI would replace the role of marketers.
- 11:21I think that's also one of the reasons to that.
- 11:24And I really like how you start talking about predictive AI versus
- 11:28generative AI and those things.
- 11:30Uh, because, you know, predictive AI is just, you know, Seeing the future
- 11:35and forecasting it through machine learning, which has been already in,
- 11:38in the lives of so many professors, so many researchers, uh, whereas
- 11:43generative AI is about creating content.
- 11:45Maybe that's, that's my take on that as a marketer.
- 11:49Dr. Jeffrey Funk: Wow.
- 11:50Generative AI is going to have a bigger impact on marketing than a lot
- 11:53of other functions because the output of generative AI is words and images.
- 11:59It's not going to have a big impact on making cars, or constructing houses,
- 12:03or mining minerals, or catching fish, or even doing healthcare.
- 12:09Because the output of words and images is very different from the output
- 12:12of the industry I just mentioned.
- 12:14There's very different things.
- 12:17And so marketing, yes, it will have an impact.
- 12:19I mean, a lot of marketing is advertising videos.
- 12:21I agree it will help you make videos.
- 12:23I think it also helped make, uh, movies and TV shows.
- 12:27It's not like they're going to make a whole movie, make a whole TV show,
- 12:30but they help in certain places.
- 12:32And then gradually over time, the contribution from generative gets
- 12:36bigger and bigger and bigger and bigger.
- 12:38So that's simply how I would do the forecast because this is the way
- 12:41technology has changed very gradually.
- 12:44And there's some impact of marketing on words, but you see,
- 12:47when you're using marketing, you're really talking about advertising.
- 12:49And I would say that marketing is much bigger because marketing
- 12:52does a lot about understanding what the market for something is.
- 12:55That's very hard.
- 12:56That's what we're doing right now for generative eye.
- 12:58We're trying to stand with the market is for that.
- 13:00And I don't see how generative eye is going to help me determine what
- 13:04the market for generative eye is.
- 13:07The market for generator comes about by having the conversation we're having
- 13:10right now and thinking through what's happening in the market and the uptake
- 13:15and who's using it and who isn't.
- 13:18Itir Eraslan: The other day, a friend of mine, he's a creative director, and
- 13:22we were talking about a commercial and then, you know, just making positive,
- 13:25negative comments and criticizing that.
- 13:28And now he was like, why do we still do ads?
- 13:31You know, I mean, it's so old school because we know that, for
- 13:34example, paid media, it's not working as it used to be in the past.
- 13:39So things may be evolved, uh, in that sense as well.
- 13:43I mean, of course, marketing is.
- 13:45It's not only about advertising or ads.
- 13:47It's a part of it.
- 13:49But I think things might shift around that.
- 13:53I would like to ask you about your opinion about the role of
- 13:56human creativity in marketing.
- 13:59As AI is continuing to evolve, do you see like phases of human creativity advancing
- 14:05with AI as the technology improves?
- 14:09Dr. Jeffrey Funk: Well, I'm not an expert on marketing creativity.
- 14:12I mean, I've, I've seen a few movies and things and read some things.
- 14:17Most of it appears like you've got to have some core idea for some words that
- 14:22are going to strive the new product.
- 14:24And then you try to express those words in some images.
- 14:27And so coming up with the idea, it's very difficult.
- 14:30Now I can imagine that generative AI can help you try those
- 14:33ideas out much more quickly.
- 14:34So you can create some videos, you have some words, and you think this sounds like
- 14:39a good marketing campaign for the product.
- 14:41And then you input the words and you get a bunch of images.
- 14:44And so you can try things much faster with generative.
- 14:48So I, I believe that that's going to happen.
- 14:51That can help.
- 14:51And that, Very good creatives are gonna use it.
- 14:54Well, I mean, there's gonna be a lot of people trying, but unless
- 14:57you're really creative in the first place, you're not gonna be able to
- 15:00use the generated vibe very well.
- 15:03You know, it's not gonna be very Mm-Hmm.
- 15:04impactful.
- 15:05Itir Eraslan: Mm-Hmm.
- 15:06. I think I see it in a way, for example, you know, Adobe is the
- 15:09creative platform that you can use, you know, generate graphics and so on.
- 15:14And it's a big tool for the creatives, like the designers and so on.
- 15:18So.
- 15:19When, uh, if we think that way, I mean, a graphic designer was a graphic
- 15:23designer 100 years ago, and it's a graphic designer now, but with the
- 15:27tools of Adobe, then it's more enhanced.
- 15:29I mean, it doesn't necessarily diminish the role and make the role, uh, fight
- 15:34away, but I see this as an enhancement of, uh, the creativity process.
- 15:40Dr. Jeffrey Funk: Well, it'd be very incremental.
- 15:42Think of some of the things that people say, uh, computers, I mean,
- 15:46AI isn't going to replace people.
- 15:47People use AI, it'll replace people who don't use AI.
- 15:51And they never really talk about what they mean by that.
- 15:53I mean, you think about people who use computers, replacing people
- 15:58who didn't use computers, right?
- 15:59Everybody uses computers now.
- 16:00So obviously they've replaced the people who didn't use computers.
- 16:04You know, the AI people kind of think, Okay.
- 16:06I know AI and all these statistics and all this, and all this has
- 16:09become necessary in the future.
- 16:11And that's going to be the future.
- 16:13I think, well, wait a minute.
- 16:14The people who are using computers today don't really know computer
- 16:18architecture or know how computers work.
- 16:20They certainly don't know how summit conductors work and all
- 16:22these integrated circuits work.
- 16:23They don't know any of that.
- 16:25They know a bunch of.
- 16:27You know, kind of superficial stuff about computers about, you know, how
- 16:31to use the software and you click here, you click there and it can do this.
- 16:35It can do that.
- 16:36Where do you find information?
- 16:37You know, all this stuff that.
- 16:39A true engineer would say it's superficial, but it's
- 16:42necessary to do your job.
- 16:43I see AI as being the same way.
- 16:45It's going to require people to understand some, you know,
- 16:49prompt engineering type thing.
- 16:51It may require them to understand something about statistics and things
- 16:55like that, and you know, how often something makes errors, but probably not.
- 17:00It's probably going to be more, well, it makes these kinds of errors, so you
- 17:03have to check for those kinds of errors.
- 17:04Uh, and, uh, it doesn't make these kinds of errors, so you can
- 17:09usually trust it when it says this.
- 17:11And that will take a while to understand when you can trust it, what type
- 17:15of things you can trust it with, and what type of things you can't.
- 17:18Um, and you get these videos, and sometimes they produce an
- 17:21extra arm or a leg or something like that, which may hurt you.
- 17:25They may also give you some historically, or something inaccurate.
- 17:30But you may say, well, for these kinds of videos, marketing videos, we
- 17:34don't care about the accuracy so much.
- 17:36You know, it's not like we're doing a documentary where we have to have
- 17:39the people look exactly the way they actually looked 150 years ago.
- 17:45Um, and so these are the kinds of things that I think that.
- 17:48People will have to learn.
- 17:49They'll have to learn when AI makes, you know, what are the
- 17:52advantages and disadvantages?
- 17:53And you can't just say, wow, it's early years of AI.
- 17:57It's going to get so good and everything.
- 17:59I'm perfect.
- 18:00No, it's not going to get perfect.
- 18:01It's going to get a little better, but not a lot.
- 18:04And so you better understand where it works.
- 18:07And where it doesn't work so well, and that's going to be the challenge
- 18:11Itir Eraslan: going out of the marketing scope for a bit.
- 18:15Are there any specific AI applications that you believe
- 18:20the game changers for businesses?
- 18:22Because.
- 18:23A lot of companies are investing heavily, uh, on AI without knowing too much
- 18:28about it or like just trying to, you know, jump to the wagon and helping it.
- 18:33Do you see some specific areas that will make the most impact, especially when
- 18:38you look to the one or two years horizon?
- 18:41Dr. Jeffrey Funk: Well, I think these, these advertising videos
- 18:44and then a lot of editing or videos and TV shows is the biggest.
- 18:49Um, Mm-Hmm, . I, I think the generative I and the tech side will have less of an
- 18:54impact because there isn't a lot, there's books and things, but my understanding
- 18:59is the books that, that are written with generative aren't that good.
- 19:03And, you know, it's really hard to write a bestselling book really hard.
- 19:07Even, you know, lots of smart people can't, it's very hard.
- 19:11So I don't see generated by having much of an impact there.
- 19:16Um,
- 19:17Itir Eraslan: you mean in like, uh, book writing copywriting on those areas?
- 19:22Dr. Jeffrey Funk: I mean, like, uh, writing a book, like, uh, you know,
- 19:27and helping an author write a book, you know, maybe it can help checking,
- 19:32you know, maybe editing, maybe editing a book, and you're looking
- 19:35for inconsistencies or something, but.
- 19:38I have a feeling that it's not going to be that big a deal.
- 19:41Um, there are some other areas that people like to talk about.
- 19:45I mean, certainly software code, but appears that software generated by
- 19:49makes too many errors creates too bloated, too much software code.
- 19:53So it proposes solutions that have too many characters.
- 19:58You don't want to have too many lines of code, right?
- 20:01We already have too many lines of code.
- 20:02These software solutions, just millions of lines of code, and
- 20:05then somebody has to check them and figure out where there's a problem.
- 20:09And so if you're using generative AI that creates more code than a human
- 20:13would to solve a particular problem, It's going to be harder to check.
- 20:17And that kind of gets me to what one of the big things I
- 20:21emphasize, which is process.
- 20:22You know, a lot of these economists and stuff, they, they think it's too
- 20:25complicated to think about processes.
- 20:27And so they just focus on tasks and well, they have these 10 tasks and,
- 20:30uh, generally I can help you with each task and the user will just
- 20:34go from this tool to that tool.
- 20:35And this isn't the way the world works.
- 20:38This isn't.
- 20:39What enabled productivity to improve.
- 20:43You see, I grew up, I, a lot of my early career was in just in
- 20:46time manufacturing and processes.
- 20:49You know, it wasn't about a person just seeing how many parts they could produce
- 20:53and let them stack up in the corner.
- 20:55No, it was producing just in time so that there was close feedback
- 20:58between the downstream process and the upstream process in order to
- 21:02make sure things were done right.
- 21:03That's how you achieve lower defects and how you improve
- 21:06the productivity of factories.
- 21:08The same thing was done product development.
- 21:10People looked at this back in the early 90s.
- 21:13Books came out on how Japan was able to develop products faster than other
- 21:18companies in the world because they were doing a lot of things concurrently.
- 21:22People were collaborating very closely and not just having it
- 21:25passed downstream and then passed downstream in the past downstream.
- 21:28They were doing work at the same time concurrently sharing ideas.
- 21:33Now, I have a suspicion that in the marketing field, there's
- 21:36a lot of collaborative work, a lot of sharing of ideas, right?
- 21:41The, and the good ideas, uh, come partly from collaboration.
- 21:46There may be some core ideas and, you know, the stars produced, but putting it
- 21:50all together, you know, making sure that it all works well, that message works
- 21:54well, that the advertising video works well, requires a lot of collaboration.
- 21:58So you have to think about process.
- 22:00So you have to think about how is generative AI going to help in that
- 22:03process, that creative process.
- 22:06So understand what has to be done in that creative process,
- 22:09uh, and how you catch errors.
- 22:12And you know, when errors aren't caught, what's the implications.
- 22:16So this, this is important, I believe for any, Application for AI.
- 22:21It's all about process, understanding processes.
- 22:24And, you know, I'm not the first to come up with this.
- 22:26There was a very, a bestselling book around 1990 called re
- 22:30engineering the corporation by James Champion, Michael Hammer.
- 22:33It was all about this.
- 22:35And it just had all these simple examples of processes and companies
- 22:38and how there was a lot of work that they realized wasn't really necessary.
- 22:43And other work that was more important and work that needed to be done in parallel.
- 22:47So in order to implement AI really well, people need to think about the processes.
- 22:52What are the processes?
- 22:53Uh, don't just think about my work.
- 22:55You know, that's what a lot of these proponents of generative AI are doing.
- 22:59They're just talking about their work.
- 23:00Oh, it improves my productivity like crazy.
- 23:02You know, says one crypto CEO.
- 23:05commenting on my posts, you know, and it's like, I wonder what this guy's doing.
- 23:10You know, he's just writing messages on, on LinkedIn and he says, wow,
- 23:14this is, this is really helping me.
- 23:16Um, but you know, it's not getting any work done.
- 23:18It's not improving the productivity of anything.
- 23:22Uh, and so we need people to think in depth about this.
- 23:24And I know people are, there's people, but there's a lot of people aren't.
- 23:28And one of them is Sam Altman, right?
- 23:29Cause Sam Altman doesn't understand any of this, or at least his comments
- 23:33suggest he doesn't, Miramarati doesn't, uh, her comments suggest she doesn't
- 23:37understand it and there's a whole lot of people who aren't talking about this.
- 23:41And, you know, even economists, I have problems with a lot of economists who
- 23:45they do these studies and they, they, they don't refer to these things.
- 23:49They just talk about, Oh, the average.
- 23:51Person in a call center was able to use general IT rat faster than they used to.
- 23:56They don't talk about what happens when they make an error, right?
- 23:59When they make an error and something gets transmitted to a client, it's wrong.
- 24:02There had, there's, there's a process by which that has to be fixed.
- 24:05We're not talking about that.
- 24:06Right.
- 24:07But yet these processes are important when there are errors.
- 24:10How do we deal with them?
- 24:11There's a process, right?
- 24:13And you have to think about how is generated AI going to impact
- 24:17on the process as a whole.
- 24:21Itir Eraslan: Um, the other day I was reading, uh, another professor's
- 24:25note, which says, uh, marketers need to know what's going on.
- 24:29Uh, learn how to talk to a machine, how to talk with generative AI, because SEO,
- 24:35you know, search engine optimization, uh, would not work as it is working right now.
- 24:41So you have to understand how to work with generative AI, which
- 24:44means people start to use per pixelity, uh, chat GPT to do search.
- 24:49And he says that this is different than.
- 24:52How we used to do SEO.
- 24:54So what's your thoughts on that?
- 24:57Dr. Jeffrey Funk: Well, some people use the term prompt engineer.
- 24:59You need to know what to prompt it with.
- 25:01And I think that's way too simplistic.
- 25:03I mean, what they're really saying is what I said earlier.
- 25:05You have to understand when it makes errors, when it doesn't, when
- 25:11you can trust it, when you can't.
- 25:13Um, you know, we all did this with, with Google search engine, right?
- 25:17We all look for information with, with the search engine.
- 25:21And we know that if we ask what's the population of Australia, then
- 25:24most likely the number, first number it gives you is correct.
- 25:29But if you're asking about something more complicated, like who should I vote
- 25:33for Donald Trump or, uh, Kamala Harris.
- 25:36Well, there's going to be some controversy in there.
- 25:39And you know, a lot of things are in between those two, but nevertheless,
- 25:43you're going to have to understand when you think the generator guy gives
- 25:47you, when you can trust the answers and when you can't and, uh, and how
- 25:52you ask things such that you're more likely to get a more accurate answer.
- 25:57So you have to understand the technology, which I was mentioning earlier.
- 26:02Itir Eraslan: For example, I, I started doing some search, uh, when
- 26:05I'm using ChatGPT or Perplexity, uh, so I started doing search on brands
- 26:10saying, okay, what's the best brand for milk, oat milk, those types of things.
- 26:15You know, in order to understand how the search option works.
- 26:18Uh, but the thing is that.
- 26:21It's not like the search that we have at Google right now.
- 26:23Of course, Google search is also quite screwed because of all the,
- 26:27you know, a CEO, all the branding and ads, uh, but it's not there yet.
- 26:32So I think it's just like a game off of, you know, all these seven
- 26:37companies trying to win, uh, you know, because Google search is a big thing.
- 26:42And now People are using, suddenly using Chachapiti or Pyrexite to do search.
- 26:49Uh, your book is coming up in October.
- 26:52Uh, how long does it take for you to write the book?
- 26:56Dr. Jeffrey Funk: Well, writing is very easy.
- 26:58I mean, if you spend your life writing, it's very easy to write.
- 27:01It's, it's hard to get a book published.
- 27:04So I, much of this book was written, Three years ago, I was trying so hard
- 27:10to get published because see, I noticed these things so long ago, I was warning
- 27:14about the startup losses back in 2018, 2019, because I had anecdotal evidence
- 27:19that most startups were losing money and over time I got better and better data,
- 27:23but you know what people don't care,
- 27:27you know, the VCs and the startup companies, they don't care
- 27:30because they want to hide this.
- 27:32And the professors who teach entrepreneurship, they want
- 27:35to hide this, they all want to hide this and it hasn't blown up
- 27:40because there's so much easy money.
- 27:43So these startups, they can't make money.
- 27:45I mean, 90 percent of these unicorn startups are losing money.
- 27:48They can survive because they just cut back on their growth.
- 27:52They just cut back, but they're not ever going to make money.
- 27:55And people use this term zombie.
- 27:57Company, company that has, uh, losses, big losses, uh, but it has enough
- 28:03revenues to pay for the interest on the debt, uh, they're not going to recover.
- 28:08So this is what I've been dealing with and trying to convince
- 28:10publishers this is an issue.
- 28:12I mean, publishers are like, you know, academic publishers, first of
- 28:16all, are hard because you, uh, they will publish these kinds of books.
- 28:19But if you're not really famous, it's not only that they don't
- 28:23want to publish you, they can't get anybody to review your book.
- 28:27Because a famous professor isn't going to review a non famous professor.
- 28:31So then you have to go to the business books, the how to books.
- 28:35But you have to how to do something.
- 28:37You see, so my book is about how to Avoid, uh, investing in bad companies
- 28:44and bubbles and things like that.
- 28:47I'm not an investment person.
- 28:49That's not my expertise, but I had to, to write this kind of book because
- 28:54that's what published what they want to, how to book, how to do something.
- 28:57So I'm trying to tell people how to interpret market signals and things.
- 29:03Itir Eraslan: So I will have two questions about that because you
- 29:06are often highlighting the excessive hype of AI and the book is about
- 29:10unicorns, hype, and bubbles.
- 29:13Uh, do you foresee a crisis, uh, stemming around the overhype
- 29:18of AI, uh, in the coming years?
- 29:22Dr. Jeffrey Funk: Well, I mean, if you, if you add up the increase in
- 29:27market capitalization for the top tech companies since January, 2023, You
- 29:33come up with the figure of 10 trillion.
- 29:36Now, I'm the only person that noticed this, I guess, because
- 29:38I never see anybody mention it.
- 29:39Because when I say 10 trillion out there, a lot of people say,
- 29:42where'd you get that number?
- 29:43Where'd you get that number?
- 29:44Because no one else says it.
- 29:45But it's 10 trillion.
- 29:47I mean, all you have to do is go through Microsoft, Google, all these companies.
- 29:51I mean, it's huge.
- 29:52I mean, Microsoft share price has almost doubled since January 2020.
- 29:58You add it up, it's 10 trillion.
- 29:59That's a lot of money.
- 30:01And you look at, uh, you know, the revenues for this AI.
- 30:05I mean, yeah, the cloud company revenues, but the actual
- 30:08sophomore revenues are very small.
- 30:10You know, and other people notice that with, uh, that the biggest is probably
- 30:15a few billion for, uh, open AI, but none of these other companies and, um,
- 30:22Microsoft just announced some earnings.
- 30:25And they, they don't say anything about co pilot, number of co pilot users,
- 30:29number, amount of co pilot revenues.
- 30:32Uh, and of course that's what caused Alphabet stock to drop last week, was
- 30:35they didn't say anything about revenues.
- 30:38They didn't say probably need details.
- 30:40Um, and so people are beginning to notice.
- 30:43So wait a minute We we expect to see these kinds of details We just don't want to
- 30:46hear about well, your advertising revenues went up 20 since last year, which is
- 30:50that's pretty good most companies but the problem is that microsoft's I'm, sorry.
- 30:57It was uh Microsoft revenues had gone up 20 percent in the last year.
- 31:02Um, and that's pretty good.
- 31:04But the stock price doubled Yeah You know, it's up like 2
- 31:07trillion in market capitalization.
- 31:09We're looking for bigger numbers than just a 20 percent increase.
- 31:12We're looking for something that suggests that all of this hype
- 31:15about AI, this increase in market capitalization, 10 trillion is warranted.
- 31:20is justified.
- 31:22So that's what the market's looking for.
- 31:24So at some point, the market's going to say, well, it isn't there.
- 31:27You know, I know some people will say, well, they're just not talking
- 31:30about the co pilot number, co pilot users and revenues, but they're there.
- 31:35They're going up.
- 31:36And one thing I've learned through my life is that companies will
- 31:40brag if they get the chance.
- 31:41If they have something to brag about, they're going to brag about it.
- 31:44So if there were some good numbers there, they would be talking about them.
- 31:48They wouldn't be shy.
- 31:49People aren't shy.
- 31:50These companies aren't shy.
- 31:51They're concerned about their share price.
- 31:53It's their primary concern.
- 31:55They're going to do anything to move that share price up.
- 31:59So what, you know, if they had these numbers, if they had good numbers
- 32:03for copilot, Microsoft would talk.
- 32:07Itir Eraslan: I mean, on my side, from the advertising world, I'm expecting
- 32:10that open AI will soon, will start.
- 32:14Getting ads, uh, for the search results that they are just like Google search,
- 32:18you know, you get to pay that's, uh, and I'm waiting for the moment
- 32:21that these companies, uh, generative AI companies are now turning into
- 32:26advertising companies, uh, taking ads.
- 32:29Dr. Jeffrey Funk: No, that's a very good point.
- 32:30I mean, I, some of us raised that year, a year ago when people talking
- 32:34about opening, I is going to eliminate, you know, Google search because
- 32:37it's going to be able to end better.
- 32:38You know, the first question is, well, how are they going to monetize it?
- 32:42One question is how are they going to monetize it?
- 32:44Are they going to charge people for it?
- 32:46If not, how are they going to offer ads?
- 32:49You know, and so those kinds of details, they never came out and when they
- 32:54don't come out, what do you conclude?
- 32:56Well, they don't have an answer to it.
- 32:58They don't have one.
- 32:59Sam Altman is again just talking around the issue.
- 33:02Uh, it's not that you have a bad question that you're misunderstanding things.
- 33:06No, you're right on.
- 33:07You understand it.
- 33:08That's the problem.
- 33:09That's the question you got to keep asking.
- 33:11And the reason that Sam Altman and Mira Murati aren't answering the questions
- 33:16'cause they don't have a good answer.
- 33:17They don't know how to do it.
- 33:18Uh mm-Hmm.
- 33:20. And I imagine that they're not gonna come up with a good way to do it.
- 33:24Itir Eraslan: Um, the second question and the last question about the book
- 33:27is, uh, what are the insights, what's one insight that you can give, uh, to
- 33:33avoid the, these pitfalls, uh, about.
- 33:36Hypes on bubbles.
- 33:37Dr. Jeffrey Funk: Well, what I say is you have to understand.
- 33:40You just don't listen to people's Forecast examinals.
- 33:44You have to understand them.
- 33:45You have to be doing the forecast yourself.
- 33:47You're understanding the logic behind the forecast uh, and so, um One part is
- 33:54about startups and all these startups are losing money and a big reason
- 33:57they're losing it because they didn't do a technologic, sophisticated business.
- 34:02And when you look back, all the successful startups over the past 50,
- 34:0560 years, it was like some companies like networking companies, computer
- 34:10companies, software companies.
- 34:11It was all something technologically new that enabled this new startup to succeed.
- 34:17You can go through any from Oracle, Microsoft, they all
- 34:20did something radically new.
- 34:22There wasn't being done that was much better.
- 34:24Well, that's not the case with a lot of these food delivery startups
- 34:28and, uh, even health care startups.
- 34:31There is nothing really new.
- 34:34It's kind of something very small.
- 34:36Um, for the companies that did offer new technology, they
- 34:41tried for the main market.
- 34:42So think about VR and AR.
- 34:44We've been hearing about VR and AR for I mean, I've been hearing about for 30
- 34:47years and certainly with startups, I mean, you go back to our Oculus Rift.
- 34:53It was acquired by Facebook more than 10 years ago.
- 34:57People have been talking about this a long time, you know, and here we
- 35:01are, even Apple can't do well with it.
- 35:03So my conclusions, they all kind of assumed this was easy.
- 35:08And my book goes into the reasons why it's not easy.
- 35:10So in a lot of ways, my book goes into how the progress is very slow in these
- 35:14technologies and the companies should have been aiming for a much simpler thing.
- 35:18You know, some niche user, right?
- 35:21You got to understand that niche user, someone who's willing
- 35:24to pay a lot of money, uh, for something that isn't that great.
- 35:30Now, this is the case all the time with new technologies.
- 35:32You have to find that niche user who's willing to pay for
- 35:35something that's not that good.
- 35:36And, uh, but they didn't.
- 35:39And so I show that the progress is very slow on all kinds of these technologies.
- 35:43And now I know that a lot of marketing goes, I don't want to go into this
- 35:46stuff, you know, and you know, whether it's progressing or what, but, but, you
- 35:50know, that's the core of understanding whether it's going to be big.
- 35:54Right?
- 35:54Is it progressing?
- 35:56You know, so, when I was, you know, a young man, in graduate school, you know,
- 36:01personal computers were big in the early 80s, and I wasn't really following it very
- 36:06closely, but it was hard to avoid because I was at Carnegie Mellon, and I was in the
- 36:11news, uh, we had to use computers to write our articles, write papers and things.
- 36:17You know, people were talking about Moore's Law all the time, and about
- 36:20how it was, you know, the computers would get faster and faster and faster.
- 36:24Because there's a direct link between the better transistors
- 36:27and the faster computers.
- 36:29So that was very easy to forecast.
- 36:30And now you look at generative AI and you think, well, are the frequency
- 36:34of hallucinations going down?
- 36:37You know, people talk about all these new products and, oh, they have higher
- 36:39resolution and, oh, I can use these inputs to get these outputs and,
- 36:44but that's not the core question.
- 36:46To me, it's all about what is keeping people from using more generative
- 36:50AI for text or even for video.
- 36:53It's hallucinations.
- 36:55So are the frequency of hallucinations going down?
- 36:58I see a little impact, not much.
- 37:02And so that's what the book is about.
- 37:04It's about all these performance metrics, about thinking about a technology,
- 37:07you know, like drone delivery or.
- 37:09You know, even Starlink thinking about how much faster, how
- 37:12much better is it getting?
- 37:13How fast is it getting better?
- 37:16That's a core question.
- 37:17If you want to think about a new technology, it has to be getting better
- 37:20really fast or else it's not going to go from little use to lots of use.
- 37:27Itir Eraslan: I mean, I feel that I'm also on the correct path of asking those
- 37:31questions because I was interviewing with someone and then she says that
- 37:35now I can, instead of creating.
- 37:39One post in an hour, I can create a hundred posts in an hour.
- 37:44And I said, I mean, why would you need a hundred posts?
- 37:48What's the reason to that?
- 37:49I mean, like, there's only like a few people that you can talk and those
- 37:53people can consume that much of content.
- 37:57Why would you need to produce hundreds?
- 37:59I mean, I don't want to produce that.
- 38:01Uh, so I'm happy that, uh, I was able to talk to you because, uh, I also, you know,
- 38:06watch a lot of your videos and that's.
- 38:07Also shifted out of thinking, uh, two last questions about more lifestyle.
- 38:13What is one book that you would recommend, uh, reading
- 38:18Dr. Jeffrey Funk: about this topic?
- 38:20Well, I'm sure that, uh, Gary Marcus's book is going to be very good.
- 38:24It comes out in September.
- 38:27I like Emmanuel Maggiore's book, Silicon, which comes out in September,
- 38:31but he also wrote a book last year, Smart Until Dumb, about AI.
- 38:35They're very funny books.
- 38:37I mean, I read the book, Silicon, that's coming out.
- 38:39He sent me an advance copy.
- 38:41But they're very funny books, you know.
- 38:43So my book isn't funny like that, because he has been working with these, These
- 38:48people and these projects, and he's talking about the things they say and a
- 38:54lot of these crazy things where, you know, they, they, they, they didn't need AI to
- 38:58solve a problem, but they used AI because it would look better on the resumes and
- 39:01it looked better as a project description.
- 39:03And, um, and so he's talking about really source of the hype, um, you know, that,
- 39:09that people want to say, talk about AI, because it makes them look good.
- 39:14You can put it on the resume, makes them look like they're really advanced.
- 39:19So those books are good.
- 39:20There are others.
- 39:20I mean, you can read, uh, Darren Acemoglu's new book this year, uh,
- 39:27but it's more academic, most, most practitioners don't want to read
- 39:31academic books, it's got too much theory and stuff, uh, so these
- 39:35kind of funny books are better.
- 39:38Itir Eraslan: And the last question is all about, uh, every time
- 39:40coffee, uh, in Singapore, uh, do you have a favorite coffee place?
- 39:44Because I have one in Singapore.
- 39:46Dr. Jeffrey Funk: Well, I don't drink coffee, but I, uh, there's a place down
- 39:51the street from me that I meet a lot of people because I don't walk very well.
- 39:55So I say, if you want to meet, you got to come to near my house.
- 39:59And, uh, so I meet at this place, Tiong Bahru Bakery that
- 40:04has coffee that people like.
- 40:05I seem to like it.
- 40:06I usually drink, just drink some juice or something.
- 40:11Itir Eraslan: That's good.
- 40:12I mean, coffee is one of my favorite things.
- 40:14So, I mean, I don't drink too much.
- 40:16I just drink one per day, but it's an important moment for me.
- 40:20Uh, thanks so much for joining me this morning in Singapore
- 40:24and I hope to talk to you again maybe after the book is released.
- 40:29Dr. Jeffrey Funk: Okay, my pleasure.