Generative AI is one of the most exciting and disruptive technologies to emerge in recent years. It has the potential to revolutionize the way we live and work, and it is already having a major impact on many industries. With the rapid advancement of Generative AI technology, businesses are scrambling to keep up and adapt to the new reality. In this episode, Paul and Yadin discuss the rapid growth of artificial intelligence, its impact on the workforce, and how to make all of this tech easy to understand.
Generative AI is one of the most exciting and disruptive technologies to emerge in recent years. It has the potential to revolutionize the way we live and work, and it is already having a major impact on many industries. With the rapid advancement of Generative AI technology, businesses are scrambling to keep up and adapt to the new reality.
As a technology leader, it's critical to understand the implications of generative AI and how it can be harnessed to gain a competitive edge.
In this episode, we hear from Paul Roetzer, CEO and founder of the AI Marketing Institute. For the last decade, he has been immersed in the emerging capabilities of artificial intelligence and how it can be leveraged to help enterprise companies. He has watched closely as AL/ML has evolved into the large learning models (or LLMs) that have astonished the world with their ability to create, analyze and perform tasks that rival the capabilities of many knowledge workers.
---------
“What I always tell people is, AI's not gonna replace you. But people who use AI will replace people who don't. Basically in any career path, any knowledge work, any creative path. I do think, though, that the disruption to these professions is gonna be probably more significant in the near term than we expect.”
“The best advice is, you just have to embrace the fact that the technology is going to continue to advance in a really, really rapid way. And you have to have the resources internally. Maybe it's an AI council, maybe it's a couple of people that are focused on this emerging area. You've gotta stay at the forefront of it and see around the corners because the lead time to know what's coming next is gonna keep getting shorter. Right now, we could probably reasonably project what life looks like 12 months from now. Anything beyond that is just a fool's errand.”
“The blessing of ChatGPT in my world is it woke people up to the power of AI in good ways and in bad. Now people can envision the downsides as well. So now we can actually get to the important conversations around ethics and responsible application of AI and impact on workforces and things like that.”
“I feel like we need more conversation around AI as a leverage point to create better working environments, better jobs, and more fulfilling lives for people, because otherwise I think it's just for nothing. Like, it's just another productivity gain and a missed opportunity to improve people's lives.”
---------
02:23 The rapid growth of AI
05:45 Staying up to date on emerging tech
07:01 Who’s going to do the innovating?
09:20 How to organize your AI team
011:39 Why education is so important
13:53 How ChatGPT works
14:40 What’s terrifying, what’s exciting?
16:37 How language models work
19:02 Will AI replace workers?
21:56 Embrace technology’s quick growth
23:22 Your whole organization should understand AI
27:58 How to build an intelligent company
31:44 The gifts ChatGPT brings
33:43 The future of AI
36:17 Pitching AI to the board
37:59 How to get involved
---------
Paul on LinkedIn: https://www.linkedin.com/in/paulroetzer/
CIO Exchange on Twitter: https://twitter.com/vmwcioexchange
Yadin Porter de León on Twitter: https://twitter.com/porterdeleon
[Subscribe to the Podcast]
On Apple Podcast: https://podcasts.apple.com/us/podcast/cio-exchange-podcast/id1498290907
For more podcasts, video and in-depth research go to https://www.vmware.com/cio
0:00:00.7 S1: My fear is in this economy that there isn't gonna be enough thought put into it, people are just gonna take the obvious action of reducing workforces because AI can help us be more productive. And I'm trying to head that off. I'm trying to do everything we can to get the conversation out there so that people make more educated decisions around the future of their teams and their workforces.
0:00:22.9 S1: Welcome to the CIO Exchange podcast where we talk about what's working, what's not and what's next. I'm Yadin Porter de León. Generative AI is one of the most exciting and disruptive technologies to emerge in the recent years. It has the potential to revolutionize the way we live and work, and it is already having a major impact on many industries as well as the role of many knowledge workers.
0:00:42.3 S1: With the rapid advancement of generative AI technologies, businesses are scrambling to keep up and adapt to the new reality. As a technology leader, it's critical to understand the implications of generative AI and how it can be harnessed to gain a competitive advantage. In this episode, I speak with Paul Roetzer, CEO and Founder of AI Marketing Institute. For the last decade he has been immersed in the emerging capabilities of artificial intelligence and how it can be leveraged to help enterprise companies. He has watched closely as AI/ML has evolved into the large learning models or LLMs that have astonished the world with their ability to create, analyze and perform tasks that rival the capability of many knowledge workers. Most importantly, throughout our conversation, he urges everyone in the organization, especially technology leaders, to be having active and productive conversations now, so they can be prepared for what's coming next.
0:01:36.3 S1: Paul, let's start by establishing the insanely rapid pace of growth. I've heard you talk several times... Do you know... And one of the interesting things that I've heard you describe is the number of AI vendors, SaaS companies, and the like. By the end of 2022... So ChatGPT launches November 2022, by the end of that year, how many generative AI vendors were there, roughly?
0:02:00.1 S2: So the Sequoia venture capital firm came out with a generative AI landscape in the middle of October, and they had 30 on there. A week later, they came out with another one and it had a 100. Then CB Insights came out with one in November, still pre-ChatGPT and it had 250. My assumption at the moment is there's probably about 2500 generative AI tools and companies that have been built in the last five months, and I'm guessing 10,000 by the end of 2023.
0:02:31.5 S1: From 30 pre-ChatGPT to about 10,000 in 2023. And how many new announcements come out each week in the space?
0:02:42.1 S2: It's honestly really hard to keep track of... There was this one story, I think, I've shared before on our podcast maybe. But it was early March, I was on my way to San Francisco, actually, and San Diego. And in that morning there was like six major announcements. It was the day that GPT-4 came out but Google also announced their plans for Bard, and this company called Figure had announced plans for a humanoid robot, Grammarly had announced something, and Microsoft had announced... It was just wild. So some days it feels like you live a whole year in a day, [chuckle] AI.
0:03:14.4 S1: You get off the flight and then... And then you have a dozen announcements that have already just happened while you're in flight.
0:03:19.9 S2: Yes. Luckily I had the $8 WiFi on that flight so I was keeping track of everything, and then when I landed in Denver for my transfer, that was actually when OpenAI announced GPT-4. So I was processing all of that, and then in my... In your ascent you're without WiFi for 30 minutes, and I was worried I was gonna miss something in the 30 minutes I was ascending. [chuckle]
0:03:40.9 S1: The 30 minutes. It's happening so fast and you're... The ascent in your plane may mean that you just... You basically miss out on AI. And just that's it. You took a flight and then now all of a sudden the machines have taken over the world by the time you land.
0:03:53.7 S2: Before November, this wasn't the case. Things didn't happen quite this fast. And I feel like if I just unplug for three or four days, I just can't even imagine what I'll miss. So, I think, I'm going to have to stay plugged and at least keep up every few hours on what's going on.
0:04:09.5 S1: Just the fact that you mentioned that, I feel like that is something illuminating to how fast things are moving... 'Cause we're in a hype cycle right now, aren't we? We are in a very steep hype cycle...
0:04:21.6 S2: Yeah. I think, it's a legit hype cycle. [chuckle]
0:04:23.8 S1: So you think there's actually me behind the hype cycle, so we're not... From a Gartner perspective, we're not about to crash into the trough of disillusionment or are we?
0:04:32.3 S2: I think we're years away. So all this hype that's happening, I'm not so sure that most CIOs, CMOs, CEOs... I'm not sure that most have a grasp on the depth of AI and the use cases. The hype cycle right now is just around language models and ChatGPT and that's just the tip of the iceberg.
0:04:50.5 S1: How do people stay up to date on this? You're afraid of taking a day off. And when it comes to then project this out to... I'm working in an enterprise company, and because I get lots of questions now I have work streams to be able to address some of those questions. And so it seems overwhelming. And we have high profile firms like Andreessen Horowitz pouring money into some of these companies, knowing they're not gonna be around in three years. Or maybe not even... By the end of the year. I'm a CIO, I'm in an enterprise company, I've sales, I have marketing, I have engineering and other lines of business beating down my door saying, "We want these tools. How do we integrate these tools? What are our strategy around these tools?" What do I tell them?
0:05:28.7 S2: It's a really tough challenge right now when you look at the tech stacks and for the CIOs that have to solve for this, it's challenging. Because we've always had startups and emerging tech companies showing up on the scene, it's just never been at this volume. And so, I think, that there's gonna be a lot of shiny tools, a lot of really interesting AI applications. There's gonna be a lot of pressure from different functions within the business, like Marketing, sales service, OPS, HR, finance that are gonna want to be... Experience these latest and greatest tools. The challenge is that all the big platform companies that your stack is built around are also building these capabilities.
0:06:03.2 S1: Exactly. Exactly.
0:06:03.9 S2: Like Microsoft, Google, Adobe. They're all building them too.
0:06:07.4 S1: So, why would I sort of go out and start experimenting with all these different point solutions... You say, what... Like 10,000 of them by the end of the year. Why do I need to look at this big field of 10,000? Why don't I just look to my rep at Microsoft? Or reach out to Google and say, "Hey, lookit, I wanna bring this to my tech stack, but I wanna do it in secured and controlled. I wanna do it in a an uniform fashion where I'm not doing onesies, twosies, and managing a billion different licenses for some small point solution. Two things, one is a CIO, like how... Why wouldn't I just take that approach to my tech stack? And the second part of that too, I'd love that you speak to is as a SaaS company, why am I even building function or it's gonna even just be a function in larger tech stacks.
0:06:44.8 S2: It's tricky because the big players aren't gonna necessarily always be the innovators here. And it's not a sure thing that they're going to bake in. The technology that you're seeing right now is really appealing. So if you just think about the generative AI space with image generation and language generation and audio generation and video generation and coding and things like that. The innovation and like stay on the video one for example, if you're marketing and sales team wants to be doing video generation, like text to video kind of stuff, there's really only a couple players in that space and they are not the traditional tech companies. And these companies are raising hundreds of millions of dollars. Like they're legit startups backed by major VCs and oftentimes by buy Google and Microsoft and Amazon themselves are investing in these companies.
0:07:30.2 S2: So, I think, there's gonna need to be a mix of staying on the cutting edge with a team that probably focuses on the emerging technologies and companies. And then trying to keep as closer as you can on the big players that are core to your tech stack already and what their roadmaps are because they're having to evolve really quick. And then on the application side, if you're a SaaS company trying to build these. You gotta be really honest with yourself about what's defensible because it's very possible. You're just gonna build... I mean language models are a great example. If I go get an AI writing tool, am I gonna need it in six months? Am I just gonna be able to do that, write in Microsoft or write in Google whatever where they're just gonna have those capabilities baked in.
0:08:09.4 S1: Yeah. Am I gonna have a co-pilot where I'm not gonna have to copy and paste this into it? Well, I just have a Microsoft co-pilot and it's really that that paper clip that was in Windows long time ago that was Adobe's coming back and the paper clip's gonna jump in. [chuckle] And it's gonna write for me to get out those... I'm dating myself by having that example. So I wanna dig into sort of how like having a tight team that stays out ahead. That's what's the roadmap of these companies? How are they their functionalities gonna benefit my tech stack and seeing what their capabilities are so I can build some of the things in? Before we get into that, though, what happens when, say for example, legal comes in and shuts it all down, do I fight that fight? Do I work hand in hand with legal just to the front end? 'Cause that's the first thing, the first step is, well great, even if I know tomorrow Microsoft's gonna have this or Google's gonna have this just because we can, should we? And is that the first step that that like say a CIO should be taking?
0:09:02.8 S2: I think that there needs to be a cross discipline team internally that works on two things, responsible AI principles, so how you're gonna use this stuff, and then the generative AI policies. So how are you going to apply these generative AI capabilities in image and language and video and code and audio. Because there are some very big issues around copyright. So the legal team needs to be a part of this. I just saw a great example recently of like they were comparing Adobe's image generation to MidJourney V5 and MidJourney just crashes Adobe. But MidJourney appears to have all kinds of trademark and copyright issues with the images they're outputting where Adobe doesn't because it appears Adobe was trained on a cleaner data set. Then MidJourney as the CIO and as the head of legal.
0:09:54.0 S2: And all these like need to come together and agree on this because the marketing team could just look and be like, "Oh this is amazing. Let's just use MidJourney." Well, that may open you up to lawsuits. I don't know, like this is where you really need to be having these conversations and a level set across the C-suite of what is possible, but what are the threats within this and how do we navigate those and help our team see through around the corner a little bit?
0:10:17.5 S1: Yeah, because I think one of the dangers here and especially some... One of the risks around the hype cycles is that this is creating a lot of excitement. You use it and it almost feels like it is indistinguishable from magic. It does things that are delightful. It does things that are helpful and there then does things that are incredibly productive with certain use cases. And so you have sort of a larger momentum for rogue users wanting to go and jump on these tools and leverage these tools and experiment with these and create all this stuff, move fast and break things. And it becomes even harder for you to fight that desire because there's so much innovation happening in this space and so much value that could potentially be created. And how are you seeing 'cause you work with a lot of different companies. How are you seeing them address all the excitement versus how then to also tame that excitement, enthusiasm with a sense of responsibility and ethical operation models and all those different pieces? I'm asking you, Paul, to basically solve the AI problem for the enterprise right now as the AI oracle for marketing. So if you could do that first us that would be great.
0:11:21.5 S2: The best I've seen the best examples is people start with education first. This is what VMware has done. You look at how do we level set across the team, what exactly it is that we're doing here, what is possible, what is this technology capable of? And then putting some basic principles and guidelines in place for the team around the use of this technology. So I feel like the education is the absolute first component to this 'cause where you see it go wrong is the CIO has to deal with these requests from all the different business functions could be coming from directors, VPs, C-Suite, whatever. But the other challenge we're now seeing is top down from the CEO. So the CEO sits at... I had this, I did a talk for a group of like a 100 CEOs a couple weeks ago and I have had calls with these people where they're like, "I didn't even... I wouldn't have believed it.'
0:12:09.5 S2: If my CIO had come to me and said, we're gonna go all in on AI, I would've been like, "No, that's probably not the right play yet." But having sat in the room and heard what's happening, I now have CEOs that are pushing down to their teams like, "We are going all in, we are gonna become an AI emergent company. Let's do it." And now you have CIOs that are reacting to the CEO pressure one and you're getting it from the under of like, "I need apps to do things." And it's a challenging environment but I, again, I think education is the absolute critical step for every organization.
0:12:40.2 S1: No, I think, that makes perfect sense because there's a lot of what I like to call cultural technology that hasn't been developed yet. There's technology, great, you have this, then there's awareness and then you have to build a cultural technology that allows people to actually leverage it. The habits that need to be built, the understandings, the interoperabilities between teams, the workflows, all of that has to be put in place so the technology can actually be leveraged to the degree that it can be.
0:13:01.4 S1: And let's take a pause right here too 'cause you talked about there's a bunch of CEOs in the room. They wouldn't have believed it until they got in their room, they've seen it, they touched it. Some have experiment with it, some on who are listening to this right now have heard about it, have may have even dabbled in it. But let's take a moment too to talk about what is it actually doing and we'll get specific and we'll just talk about generative AI for text. So how good is it chatGPT, other tools at producing something valuable to people who are working in marketing, for example. How good is it at doing what it says it's going to do?
0:13:35.4 S2: So the general models you can get out on the open market like ChatGPT where it's just trained on the internet. It's not trained in vertical specifics, so it's not trained on SaaS companies. It's not trained on insurance companies or health care companies or whatever your domain is. It's not customized for that industry. And yet if you give it a strategic challenge within that industry like, "Write me a go-to-market strategy for a new insurance product," or "Write me a business plan for launching a SaaS company." Like it can do it extremely well, surprisingly well.
0:14:08.4 S1: And I think that's what gets people excited. You see it spit out something.
0:14:11.4 S2: Excited and terrified.
0:14:12.5 S1: Yeah, [chuckle] and terrified. And we can explore both of those too because I think it's really important to be able to create that distinction of, okay, what is terrifying? What is exciting?
0:14:23.3 S2: So on the exciting side, I think people see the near-term opportunity to drive efficiency and productivity, and that's obvious to everyone. Anyone who tests one of these tools, you can immediately see, oh, okay, I can maybe write emails faster, develop proposals faster, build business plans faster, create outlines faster or I can personalize. Like if this stuff's connected to my CRM, I can actually infuse very specific details about someone in there... If it's a sales, what the latest pipeline activity was with them. Are they a customer, are they not a customer? How much did they spend last year? Like all of that can be developed as you're writing the email, that stuff can just be populating the email itself.
0:15:02.0 S2: So it is very apparent, very quickly that this stuff can dramatically improve efficiencies and productivity. That's exciting, but it's also terrifying because what does that mean to processes, to talent, to HR? Once you start looking at this, you start saying, "Oh my gosh, so do we need as many people as we have? Are we just going to produce 10X the work and have 20% less staff or are we now able to do all this other stuff?" So depending on how the organization views AI and how they're going to leverage it, it can be really scary or it can be really exciting. In my world, I live in the middle. It's exciting and terrifying all at the same time every day. [chuckle]
0:15:45.7 S1: It's not clear-cut exactly how you need to approach it. What are the benefits? What are the actions? What should I do? And to just kind of go back to talking about what it is, and it does all the things you just described sound extremely valuable, but just give us a sense just so we can level set, what is it actually doing when you prompt something like ChatGPT, "Hey, write me a business case or write me a plan, or write me a go-to-market strategy." What is it actually doing underneath when you type those words in and hit submit? So it's not taking a step back and thinking for a second, saying, "Wow, what's the big picture?" It's actually just taking what's the first word, what's the second word, what's the third word, and just the next word is going based on the prompt.
0:16:23.0 S2: In the case of generative AI, specifically AI writing tools, it's making predictions about words in a sequence or in a sentence basically. And so what it's been introduced with these language models. And again, when I say language model, like OpenAI has GPT-3, GPT-4, those are like the core language models. ChatGPT is sort of like the interface. You have Cohere, Anthropic, Quora, like everybody's building their own language models. What those models do is they go learn from a bunch of text data mostly. They can also learn from videos and images. But they learn largely from text data, and they learn basically where words appear, the probability of a word appearing in a sentence, given the context of everything else in that sentence or in that document. So it's just making these predictions about words. That's the simplest way to think about what it's doing, at least, at the language model level is predicting words in a series.
0:17:19.5 S1: And I think that's good. It's good to level set too to kind of understanding about what is happening in this process. There's not a whole lot of deep thinking that's happening about, well, what's this and what's that? It's basically saying in the context of your prompt and the information spin's learned on, it starts spitting out words in context and predicts what the next word should be. And that's really good to sort of say that's the fundamental foundation of the technology because there's obviously some places where that might go wrong. Of course, in full disclosure, say, I went into Jasper, one of these interfaces and said, "Hey, you're a podcast host, and you have get, you know, and your guest, Paul Roetzer, is gonna be on the show, and write 10 deeply engaging questions that spur a conversation with Paul that would help CIOs, enterprise companies gain value perspective on what actions they need to take in the wake of the new gener genre of AI tools. Here's Paul's background" I paced your whole background, everything you've done and all this stuff that you've done. The results were absolutely terrible. They were horrible.
0:18:16.5 S1: They were just like, Paul, tell us about your career history and what you did when you were... It went off just like, I'm looking at the bell curve of like what it's been trained on. And it all probably is like, this is the way you should conduct this particular interview. It doesn't know what style you want to do. It doesn't know how you want to take the conversation forward. It says, okay, let's be trained on information in the past, but it doesn't necessarily know what comes next, what the next leap, what's the next step in the conversation. So aren't totally in trouble yet, right, Paul?
0:18:49.0 S2: [chuckle] What I always tell people is AI is not gonna replace you, replace writers, designers, architects. People who use AI will replace people who don't basically in any career path, any knowledge work, any creative path. I do think though that the disruption to these professions is gonna be probably more significant in the near term than we expect. There's limitation to these models. They hallucinate, they make stuff up because they're...
0:19:15.9 S1: I like the way you put that they hallucinate. I think that's interesting. Could you...
0:19:20.0 S2: It's actually the technical term for...
0:19:22.0 S1: The technical term?
0:19:23.0 S2: It thinks it knows something but it doesn't know. So it'll just make stuff up. It might not actually know who I am. It's a possibility that whatever language model Jasper was using for that prompt 'cause they'll use different models depending on the question that, that model didn't really know who I was. Now you gave it some additional context but it'll actually just kind of make stuff up. It pretends like it knows who I am. So that can happen and that's the hallucination. It can make up dates, people, historical figures all this stuff. It can just make it up, and it sounds really authoritative when it does it.
0:19:56.7 S1: Yes. Quotes are my favorite too where it's like, Can you give me something... Some quotes by this person? And they'll basically just make up a quote. This person said this. And it sounds great into the context of what you're writing. Whoa, that quote works really really well there, but the person never said it. They just made it up.
0:20:09.4 S2: So that... Humans will be in the loop for the foreseeable future. GPT-4 is seven-month old technology. So what we're experimenting at the very forefront right now which is mind-boggling some of the uses for it, it's actually older technology and it's got quite a number of guardrails in place that prevent it from doing its full capabilities. So I do think that as we enter a world where GPT-5 and 6 and 7 emerge, things start to look really really different for how good these things really are human-like...
0:20:37.8 S1: Exactly. So that... We have stuff in the lab right now that's seven months ahead of what we're already astounded by. That rate is growing at a steep curve if not exponentially. And so we'll have in the next year or two something that's just fundamentally different than even what we're looking at right now. For example we're typing text in a little box which is already... It's gonna seem absurd to us probably in a couple of years. And there's gonna be better audio and text interfaces as well. And I would imagine that it'll even start to predict what we're going to type in. For keyword, there's gonna be "Hey look, I already know what prompt you're gonna give me and here's the best possible prompt that you could do for what you're looking for." Prompt you better than you can prompt and you're gonna create something even better than you thought you could create. [laughter] Is that a technical term, Paul? Is that like hallucinating?
0:21:28.1 S2: That's gonna get... It's gonna get very weird. [chuckle] It's gonna get very weird I will say. It's the best term I can come up with for... I mean, all I can say to the CIOs, CEOs anybody 'cause it's overwhelming honestly. If you're not deep in this space, it's overwhelming to me and I'm deep in this space. The best advice is, you just have to embrace the fact that the technology is going to continue to advance at a really really rapid way. And you have to have the resources internally, maybe it's an AI council, maybe it's a couple of people that are focused on this emerging area, you've gotta stay at the forefront of it and see around the corners because the lead time to know what's coming next is gonna keep getting shorter.
0:22:07.3 S2: Right now we could probably reasonably project what life looks like 12 months from now. Anything beyond that is just a fool's errand to try and guess what this technology's gonna be like. You need to know three to six months ahead, "Oh my gosh GPT-5 is going to come out." Let's say it's this fall. I'm not saying it is but let's say it's this fall. What could it possibly have the ability to do that this one doesn't? You need to be thinking about that. You're gonna have to have people that are pondering that because you may make a big buying decision as a CIO in this summer and then have that three-year contract that you signed be obsoleted because some major shift happens this fall. And so that's what I'm saying is like, you have to just accept that it's gonna happen at an accelerated rate and you have to be prepared to stay at the forefront of that.
0:22:52.4 S1: Yeah. That makes sense. Along that line too, what do you think fundamentally, technology leaders and those business leaders who are trying to make those guesses and those bets, what do you think those sort of fundamental errors are that people are making right now? How should they start shifting their thinking?
0:23:09.5 S2: I still think it comes back to education. I think it's a lack of understanding of the technology. So if you don't know how large language models work and you go buy a major AI writing platform and you scale it across an entire team of thousands of people who are in a big enterprise, and now all of a sudden we have AI writing tools infused into marketing and sales and service and everybody else in the organization has access to use these tools, and if the CIO or the people making these decisions can't explain a language model to you and the flaws that they have inherent and the differences between the different models that are available on the market, you got a really big problem because now you have a core tech embedded into your company that you don't even understand. I came out of journalism school. I am not a computer science major. I'm not a machine learning engineer.
0:23:53.6 S1: You're one of the AI Oracles now though, Paul.
0:23:56.1 S2: Yeah but anybody can learn this. That's my point. I'm a storyteller. I just do research and figure out how to make this make sense to people and try and decipher what the researchers and engineers are saying so that other people can understand it. But none of the topics in AI are complex enough that any business leader can't understand this stuff very quickly. And so you don't need to go back to school and get a master's degree in this stuff, you don't have to spend a year on it. You just need to pick topics and go deep on them and in a week or two, capture enough knowledge to have educated conversations with your team.
0:24:31.8 S1: Now, I think that's what fascinates me too is that you said you're a storyteller but at the same time those who use AI who embrace AI will be around and those who don't use it, don't embrace it may not be from a knowledge worker perspective. And a way to approach it when you're looking at education and I think when you say education, I think, it's really what you're talking about is, you need to educate yourself, you need to educate your team, you need to give them the tools so that they can go out and stay up to date on what these latest tooling is. It's not the latest release of this particular product. It is, what is this technology?
0:25:06.8 S1: Because I think I've heard you in some of your talks before, you talked about how there's a few truly revolutionary technology shifts that have happened in our lifetimes. A lot of them have quite a dramatic. But this is one of those dramatic ones from the internet to mobile technology to now, we're looking at tectonic shift in artificial intelligence and large learning models in generative AI. And that is something that well, if you look at the past, well if companies didn't embrace mobile technologies, then that is a huge strategic disadvantage for them. It's not an advantage, it doesn't help you leapfrog the competition but it makes sure that you don't get left behind. And on that point, that's the big risk that companies have if they're not embracing this. It's not that this is going to be a differentiating technology from just the technology itself, it's gonna be there's a risk they're gonna be left behind.
0:25:55.4 S2: Oh yeah. I published a thesis in May of 2022, months before ChatGPT that I thought the future of all business was AI, native AI emerging are obsolete. You either build a smarter version of a company, any industry you can think of. Build a smarter version from the ground up where AI is infused into everything, all business functions. AI emerging is, you are an existing entity and you find ways to become AI first. You think about everything through the lens of how can we make this smarter, this process, this technology or you just become irrelevant. And depending on the industry that irrelevancy may happen in six months, like if you're a SaaS company, it could happen overnight. Look at Google, like OpenAI came after Google, nobody thought that anybody could go after Google and such. So it could happen fast or it could take, if it's healthcare or higher education, like it might take six years, seven years, 10 years, I don't know. But it's gonna happen.
0:26:45.3 S1: Yeah, one of the key pieces in that sort of paradigm that you're describing is proprietary data. And if you have something like a proprietary dataset and then you apply AI that's where you can really create your defensible differentiation. And 'cause I was listening to a CIO of the major petroleum companies and his big challenge was making sure that he got the right data to the right people at the right time. The holy grail, this is not a new quest. Everyone's trying to get there, but artificial intelligence then makes that horizon, that quest a lot closer and the journey a lot faster. Do you feel like from a tech stack perspective, from a culture perspective, from a data democratization perspective, how AI is really gonna help companies transform themselves in a way that will actually allow them to differentiate themselves for those who don't have a proprietary dataset?
0:27:45.7 S2: At its core, you're trying to build a more intelligent company. You're trying to leverage data in real time for applications across the organization to make better decisions, to do things faster, just to do things smarter overall. And data is at the core of all of that. And so if you think about where this is gonna go from a strategy standpoint, from a creation of content standpoint, from a predicting outcome standpoint, data is fundamental to all of it. And so, yeah, the organizations that one, have the data and two, structure it properly and three, have a vision for how to infuse it into building smarter processes and technologies, that's a really hard combination to compete with, if you're in an organization that doesn't do those things. It's gonna be hard to comprehend how much more efficient and how much more intelligent organizations will be versus their peers when they embrace these technologies and infuse them in versus those that don't.
0:28:40.2 S1: Yeah. And so I think you used a great example of Bloomberg pulls GPT technology onto this 30, 40 year proprietary data set in the financial services and investment community. And how powerful is that? What kind of lessons can other companies take from that example?
0:28:56.9 S2: Well, and even if it's... So in their example, yeah, it's 40 years of financial data that they're able to train a language model on. So even if it's only an internal tool, like imagine if you're in a big enterprise and you have decades of data sitting there, and right now how do people find it?
0:29:11.5 S1: It's a mess.
0:29:12.6 S2: They go to a server, they do a keyword search, they look for some, yeah. It's like, it's not easy to find.
0:29:17.6 S1: It's a total mess. And if Joe or Sam or Susie leave the company and they were the masters of that data, they knew where to find it...
0:29:21.2 S2: Gone.
0:29:23.0 S1: All that intelligence goes with them.
0:29:26.1 S2: And instead you have a language model that you can prompt with text. And on top of that language model is all of the intelligence, all the knowledge base throughout the history of the company. Everything. And I can now just ask questions through an interface, whatever it is, I want to know what trade show we were at in 2020 and how many leads did we get from it to what was the sales in 2018 when this event happened? And like anything you can imagine asking, it's just, it's literally a query away that's really, really powerful. Even if it never leaves internal. Like it's just for your stakeholders.
0:30:03.2 S1: That one excites me. Of course being that you are very deeply into sort of AI and marketing and how that connects to sales because I've been sort of through that whole life cycle, whole funnel. And if you had something that could query all of this data, then you're not a salesperson in the field trying to connect to your CRM, trying to figure out what's the last thing your customer did, when did they do something new? When I'm on the road driving, I need a hands-free experience where I could then talk to a verbal interface with this going on in the backend. And then all of a sudden you're now seeing sort of that value of that marketing brings to the field.
0:30:38.3 S1: You're not spending your a ton of your time doing queries and hitting your CRM and then filling things in... And we go deep into what the possibilities are. The possibilities are endless from digital assistance large language model creating these huge query efficiencies. But I guess what I'm looking for too is where, education is a big thing and educating themselves is great, but when do we need to get to a point where we need to fundamentally think differently about what work means? What does knowledge work mean? And you as a storyteller and a knowledge worker work with lots of different companies. You get in front of CEOs. How is that conversation progressing of what does work mean now?
0:31:24.2 S2: Yeah, it's not progressing enough.
0:31:27.7 S1: It's going slowly, isn't it?
0:31:30.5 S2: It was going really slowly until ChatGPT, the blessing of ChatGPT in my world is it woke people up to the power of AI in good ways and in bad. People can now envision the downsides as well. So now we can actually get to the important conversations around ethics and responsible application of AI and impact on workforces and things like that. So I think that throughout 2023, there's gonna be a lot more studies emerging. I've seen a couple just in the last week about the projected impact on knowledge workers and it's gonna come down to each organization trying to understand and comprehend what is basically an incomprehensible thing that's about to happen to knowledge work and trying to prepare for that. My fear is in this economy that there isn't gonna be enough thought put into it and people are just gonna take the obvious action of reducing workforces because AI can help us be more productive. And I'm trying to head that off. I'm trying to do everything we can to get the conversation out there so that people make more educated decisions around the future of their teams and their workforces.
0:32:34.6 S1: Yeah because what I'm seeing it and you can please give your thoughts on this too. I'm not seeing it replacing people, I'm not seeing it. Like I say with that example where I'm trying to get it to just generate some thought-provoking questions that really stretch the conversation take it to the next level. It's falling down there. Maybe, with better prompting, maybe with a different model, maybe there might be something we can do there, but it's not replacing people and especially with the patented trademark rulings of, "Hey, you don't really own this if you generate it." It's a utility now you own what you generate, you can't trademark it. I guess, I don't understand why a company would then look at it and say, "I need to reduce headcount" unless you have a say a quarterly goal of reducing headcount or being more efficient and that just becomes something that's more of a PR activity rather than actual thoughtful effort of, "Hey, I know that I can actually increase productivity by X and therefore I'm gonna apply this tool and I'm gonna do it thoughtfully this way." I think that's what you're talking about being the dangers.
0:33:32.5 S2: Yeah I think it's realistic to look out over the next year and say that the average knowledge worker will become 20 to 50% more efficient at their job. So you need 20 to 50% more staff than I do in a given day. So if I have 10 people on my social media team, I'll just pick an area in marketing, if I have 10 people on there who monitor, chat online and create social shares and engage with people and like they have all these fundamental things they do. They have massive audience, they're doing a lot of it. If you take those 10 people and say collectively we're gonna improve their efficiency 20 to 50%, but there's no more work to do. Like we're still gonna do the same thing, but the AI's gonna do a lot of the social monitoring, sentiment analysis, it's gonna drill drafts of the social, like it's gonna do a lot of the work, pick the images, do the hashtags, all that stuff.
0:34:17.2 S2: Do we need 10 people to do that or can we do it with four? Like that's the question you're gonna have to be faced with. We don't have 20% more work to do in that area. Now there's a lot of areas in business and knowledge work where you can say, "Yeah, that's great, we're gonna get 50% improvement" but there's 50 things we never get to do that we can easily reallocate those resources to do. That's my hope is that we look at the wishlist of things that don't happen and say, "Yeah, we're not gonna like get rid of our staff because of this. We're gonna redistribute them to all these things we never get to do that can make us a better company and grow faster and grow better." That's the hope. I just think with the economy the way it is, you're seeing it in tech is like leading the way where there's these, all these layoffs that happen in tech and none of that had to do with AI outside of maybe Meta in their last cut of 10,000. But there's gonna be a push to leverage productivity from AI in the tech world. And then I could see that trickling into other industries in the next year.
0:35:15.5 S1: Okay, well, we have a a spot where we call take it to the board and so, I think, this is floating right well into that. And when you're having that sort of, you're there, let's say the CEO and the CFO are there by your side and you're the CIO sort of having a board level conversation and you talk about what you just sort of mentioned of this is how we think we should approach AI and integrating of AI into our organization. What is that board level story, you feel like, that resonates from a business value perspective on how the company should be approaching this and it's a technology great, but ultimately it has business outcomes and from your perspective, let's say, you would be specific and use marketing, let's say marketing sales. What's that story that you tell at the board level?
0:36:03.0 S2: The board wants to peer performance, like you have to tie everything to performance first and foremost. And AI certainly fits into that conversation around how you see it impacting that, but I think they also are the shepherds of the culture and the brand and everything else. And so you need to start to build support for the idea that AI can enable us to become a better brand, a better culture, a better company by creating more fulfilling lives for our employees. And the way we can do that is by leveraging AI to not add 10 more hours of work to their week, but to actually maybe reduce a little bit so they're working actual 40 hour weeks instead of 60 hour weeks.
0:36:44.8 S1: Exactly.
0:36:46.2 S2: I see there's a positive here and it's not some like idealistic world where I think people are gonna start working 25 hour weeks and still make what they're making. I'm not saying that. I'm just saying like, we should understand that life is meant to be lived and not spent 60 hours a week working in an office. Whether it's your home, remote home office or like, so I just... I feel like we need more conversation around AI as a leverage point to create better working environments, better jobs and more fulfilling lives for people because otherwise, I think, it's just for nothing. Like, it's just another productivity gain and a missed opportunity to improve people's lives.
0:37:24.4 S1: Yeah. Do you think enough people are talking about this right now?
0:37:26.8 S2: No.
0:37:27.4 S1: Not even close?
0:37:31.4 S2: No.
0:37:32.5 S1: No? More conversations with people who are intelligent individuals who are thinking about this, who are talking about this, who are moving this forward. What's the best way that they can keep, what are some of the great ways sort of takeaways of how can I educate myself? How can I educate my team?
0:37:46.9 S2: You gotta find the thread of AI that is fascinating to you that will make you passionate. It might be bigger picture thing, like I'm just obsessed with understanding the big picture and where it all goes. So I pay attention to everything that might not be of interest to you. You may just wanna be a better leader. You may just want to be better technologist. Maybe you're more interested in this whole, like how do we build better lives and a better society with it? Maybe that's interesting to you. So, I would just find the thought leaders in that, in your lane that you find interesting. Whether they're podcasters or books or teach online courses or wherever they share their information, I would follow them. Twitter is great. I follow a ton of AI people on Twitter, so you gotta find the part that's interesting to you. And then who are the people to follow in that space? There's no, I mean, our institute is largely marketing, but our podcast is very general business society marketing. So that's a good starting point for people, but overall, you just gotta find the right people in the space that you find fascinating that you trust to help kind of guide the way here.
0:38:41.8 S1: No, that makes sense. What has been your work, like with marketing AI Institute, what has that work been like and that shift in work been like from 2022 to 2023? How has your world shifted?
0:39:01.6 S2: That's a really good question. I've been in marketing for 23 years. I've never seen a shift in mindset and demand for anything change the way everything changed for AI since November 30th, 2022 when ChatGPT came out just like from a metric standpoint, our website traffic pre-ChatGPT was up 7% year over year. It's now up on average 80 to 85% a month, year over year.
0:39:25.5 S1: That's a bit of a shift.
0:39:26.8 S2: The podcast went from average of like 70 downloads the first seven days to like, I think, we're probably approaching 3000 downloads the first seven days, this was like four months. I did 10 to 15 AI talks all of last year. I probably did 15 to 20 in March. It's just insane. And, but it leads to all these amazing conversations because we're talking to venture capital firms and major enterprises and school systems like heads of universities, heads of primary schools, like everyone is in the same boat. Nobody understands it, everybody gets that it's massive and nobody knows what to do about it. And so I just don't... I don't feel like there's enough hours in the day for us to help as many people as I would like to help figure this out.
0:40:06.2 S1: Yeah, and I think you said it best on your podcast when you said you know this is really a thing when your mother-in-law is text messaging you with questions. [laughter]
0:40:18.2 S2: I did. It was great. I loved it.
0:40:24.6 S1: Yeah. So with that, Paul, thank you for your thoughts. Thank you for your perspective. I could go on for, at least, another couple hours on this, but we're gonna end it right here. I really appreciate you jumping on the CIO Exchange podcast.
0:40:32.0 S2: Enjoyed it. Thanks for having me.
0:40:33.9 S1: Thank you for listening to this latest episode. Please consider subscribing to the show on Apple Podcasts, Spotify or wherever you get your podcasts. And for more insights from technology leaders as well as global research on key topics, visit vmware.com/cio.