<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=2006193252832260&amp;ev=PageView&amp;noscript=1">

34 Min Read

[Marketing AI Show Episode 26]: McKinsey's State of AI 2022, Lensa AI, and AI Will Eat Software Companies

Featured Image

Wondering how to get started with AI? Take our on-demand Piloting AI for Marketers Series.

Learn More

McKinsey’s annual survey, Lensa AI, and Paul’s LinkedIn post about software companies take center stage on this week’s Marketing AI Show podcast episode.

For five years now, McKinsey has been conducting its Global Survey on AI by surveying thousands of business leaders on how they use artificial intelligence. This year’s findings are now out in McKinsey’s State of AI in 2022. There’s a ton to unpack, but three big points jumped out at Mike when he was reviewing the report.

Next up, an app called Lensa AI is taking the world by storm—and creating a lot of controversy as it does. Lensa allows you to upload photos of yourself, about 10-20 selfies, then it uses AI to generate dozens of avatar photos of you in many different styles, from serious and realistic to fantastical. The app has been generating $1 million or more in revenue per day as everyone and anyone can now create images of themselves using AI. But controversy ensues.

The third topic stems from Paul Roetzer's LinkedIn post and a Twitter thread that are getting a lot of attention. In both of them, Paul outlined why anyone who runs or invests in a software company needs to have a concrete plan for considering how AI will enhance and disrupt the business. Specifically, Paul mentions that “If I was running a software company right now, I would be aggressively exploring how recent advancements in AI (and the ones that are coming in 2023) could disrupt the business.”


Paul and Mike discuss these three topics, plus they’re back with some interesting rapid-fire takes on other news. Listen to the podcast below or in your favorite podcast player. 


00:05:12 Lensa AI

00:09:28 McKinsey’s annual report

00:12:48 Software companies and artificial intelligence

00:23:34 Rapid-fire questions

Links referenced in the show

Watch the Video

Read the Interview Transcription

Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.

[00:00:00] Paul Roetzer: It's a really interesting time and it's an interesting space. There are a lot of tech leaders who are very aggressively pushing open sourcing of all this stuff and getting the AI in the hands of the users. And their belief is that society will figure it all out on its own, that it's better that you democratize access to this power and let the people figure out where the limits.

[00:00:23] Paul Roetzer: Rather than allowing a few select tech companies to keep the guardrails in place

[00:00:29] Welcome to the Marketing AI Show, the podcast that helps your business grow smarter by making artificial intelligence approachable and actionable. You'll hear from top authors, entrepreneurs, researchers, and executives as they share case studies, strategies, and technologies that have the power to transform your business and your career. My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.

[00:00:58] Paul Roetzer: Welcome to episode 26 of the Marketing AI Show. I am your host, Paul Roetzer, along with my co-host Mike Kaput, who is the Chief Content Officer at Marketing Institute and my co-author for the book, marketing Artificial Intelligence, AI Marketing in the Future Business, which is available now in print, digital, and audio.

[00:01:18] Paul Roetzer: What's up, Mike, now much. How's it going? Well, I don't know. I feel like we have a lot to follow on after our episode last week. If you didn't listen to LA last week's episode, go back and and listen to that. It was, it was a pretty wild episode. We've actually heard from quite a few people who listened to that one, who had their minds blown by everything that was going on in ai.

[00:01:40] Paul Roetzer: Um, yeah, like us, you kind of took a couple weeks away from talking about it. A lot happened and. This week, I wouldn't say we have quite the level of, uh, AI announcements. Uh, we did today have, uh, the announcement by the US Department of Energy, I think on nuclear fusion, which is a pretty big deal. . It's a little outside of our area of expertise on this show, but, uh, that I think qualifies as groundbreaking news for the week.

[00:02:08] Mike Kaput: Yeah. Someone tweeted about saying something to the effect of, it's possible in 2023 we'll have the beginnings of artificial general intelligence and nuclear fusion that's viable. And they were like, what a time to be

[00:02:22] Paul Roetzer: alive. Yeah. , no, no big deal. just like the most transformative things in history. Yeah. Uh, Yeah.

[00:02:28] Paul Roetzer: So, uh, we got, got a bit covered today. This episode is brought to you again by our piloting AI from Marketers Online course series, which may have launched by the time you're listening to this. It's coming out on Thursday, December 15th. So Mike and I recording this episode on Tuesday, December 13th. Uh, they usually go live the day after we record it, so, The new piloting eye series is something Mike and I have been working on, uh, intently for a few months, but realistically probably a decade.

[00:02:54] Paul Roetzer: So it's sort of combining everything we've learned about artificial intelligence. It's taking the book as sort of a jumping off point and really expanding much beyond that with some original thoughts and frameworks and a lot of updated content around everything that's happening with generative ai. So if you're at the point where you're really trying to figure out how to, uh, pilot AI and, and infuse it into your career in your organization, 2023, definitely check that out.

[00:03:19] Paul Roetzer: There's 17 courses. We came in at about eight hours of, of total learning time. Now you can obviously go to, you know, one and a half times or whatever speed and hit them 'em a little quicker, but it is, it is a lot. Um, but it's, you know, having gone through it. Mike, I know you did 10 courses on like intro to different categories.

[00:03:37] Paul Roetzer: I did seven courses. Use case, problem-based, state of the industry, all these other things. It was a really cool process to go through. I mean, there's so much happening and I think even since we published the book in June, I feel like so much had happened that as I was creating my courses, it was like, I'm so glad we did this

[00:03:56] Paul Roetzer: Like it helped me sort of wrap my own head around what was going on and what needed to happen. Piloting ai.com. If you're interested in those courses, they're gonna be available on demand. You can have access to everything as soon as it goes live. There's, uh, content downloads. There's an exam at the end so you can earn a professional certificate from the institute for doing it.

[00:04:18] Paul Roetzer: Um, and then, yeah, I mean, Mike and I prob will probably schedule some stuff with different cohorts of the groups, um, that go through it and do some live stuff and do some AMAs and things like that. Yeah, definitely check that out. It is again, piloting ai.com. You can learn all about that new series. And with that, Mike, I'll turn it over to you for our three big topics if you're new to this format.

[00:04:40] Paul Roetzer: Mike and I, each week we try and pick three topics to talk about, uh, and we kind of run through 'em, try and do about a half hour. Last week we doubled that without knowing it. Um, but yeah, about about 30 minutes or so, we try and keep it too. So, uh, probably about five to seven minutes per topic. And then we end with a rapid fire.

[00:04:56] Paul Roetzer: We've got.

[00:04:59] Mike Kaput: Awesome. Well, yeah, we've got, um, you know, I don't know what a normal week looks like anymore in ai, so I don't know if this is light or, or a heavy week or whatever. It's just the game has changed so much. But first up, we have something that is taking the wider world by storm. There's an app called Lensa ai, and this is.

[00:05:20] Mike Kaput: A hugely popular app that allows you to upload photos of yourself, um, about 10 or 20 selfies. Then it uses AI to generate dozens of different avatar photos of you in many different styles. So some are serious and realistic, some are fantastical. And from what I've read recently, this app launched a couple weeks ago, max.

[00:05:42] Mike Kaput: It's been generating a million bucks or more in revenue per day as basically anyone and everyone that I. Uh, interact with online has commented on it, used it, even people that had no interest in artificial intelligence, as far as I could tell. And so it's really this interesting example of. A massively popular app that's sort of broken containment, so to speak, outside the walls of, of, you know, people that follow ai.

[00:06:10] Mike Kaput: Now, everyone is asking about these AI avatars, and there's been a ton of controversy around it too, because you. In some cases the app has made some problematic changes to people's body types, skin colors. There's been some high profile cases of kind of needlessly, you know, sexualizing images of women in particular.

[00:06:30] Mike Kaput: Um, and it also seems that lens may be using your personal photos to further train the product. So there's kind of an extensive but murky. Privacy policy related to it. Um, not to mention, the app itself relies on stable diffusion, which is an immensely popular deep learning model trained on images online.

[00:06:51] Mike Kaput: And in some cases artists have complained that that model has kind of illegally duplicated the look and style of their work because it was trained on images online and artists had no say. If whether or not their art was used to train it. So there's this kind of massively popular AI photo app that actually touches on all these other really important issues kind of related to artificial intelligence.

[00:07:16] Mike Kaput: So Paul, in your opinion, Ken, why is Lens

[00:07:18] Paul Roetzer: such a big deal? I mean like anything like what happened with Dolly when it became readily available, like we saw with Chat G P T, it's accessible. It's easy to use. It's no code like anyone can experience the ai and it seems like magic when you do it. And so I think that there's gonna be an explosion of these tools that are using AI to do creative things that have some value to people, whether.

[00:07:46] Paul Roetzer: Creating images of themselves that they can share on social is a value. Um, you know, giving 'em a chance to experience the technology, whatever the, the value is to the end user. I think you're gonna see an explosion of these creative applications of artificial intelligence. And I think that's why it's just something that be, it's easily shared.

[00:08:05] Paul Roetzer: Um, you know, and so I think you see that application, but like we saw at chat, G P T. I had people asking me about that at like Dad's basketball night, . Like I had people coming up to me who've never talked to me about ai. I didn't even know, knew I did anything with ai, but they follow what we're talking about in some cases.

[00:08:22] Paul Roetzer: So people are coming up all the time and asking me these things. I'm getting text messages from people who've never cared an ounce about ai. who now all of a sudden are seeing it for themselves and like, oh my gosh, like that mind blowing moment. And I, again, it goes back to this, it, it feels like magic.

[00:08:38] Paul Roetzer: It's, it's a trick. You can show your friends, you post it on, you know, social media and it's something fun and different. And so I think it's just because it's so accessible and in many cases, free or low cost, no code. Um, so yeah, I mean that's my, my general perception. I haven't used it, but I think you. In some ways it's, it's representative of how far we need to go in society to understand ai, and I don't even know that we're ever gonna get there.

[00:09:08] Paul Roetzer: So I, this is a more like a pie in the sky idea for me. It's like somebody sees this stuff and immediately you just jump in and upload your photos and or give 'em access to Instagram or whatever. Like it's just, oh, awesome. I'm gonna output these cool things too. When you and I see something like this, my first reaction is, well, how are they using the training set?

[00:09:28] Paul Roetzer: Like, what, what are they going to do with my photos? That's my first reaction. Anytime I give any AI tool, anything, any data set? Um, first thing I think about with social media feeds and how they're learning from behaviors, I th you know, I think about it in anything we do where I know AI is present. How are they using what I'm doing to train the data?

[00:09:47] Paul Roetzer: And I, I don't think the vast majority of society would ever think about that question. And so you have a tool like this where just people just race in and give 'em whatever they're asking for and then after the fact, like, oh wait, like maybe it's training. I mean, and that's, Cathy did a great job in her article of sort of summarizing.

[00:10:05] Paul Roetzer: Hey, but Cathy knew going in, she did it knowing that like, it's like, okay, let me go test it, but let me also call out like potential privacy concerns. And you make that evaluation like, okay, whatever. It's gonna get 20 of my photos. Like it's not the end of the world. I could probably scrape that offline anyway.

[00:10:21] Paul Roetzer: Um, but again, I think just helping. People have a greater understanding of how this stuff works and learns and how the things you create and what you input into it is used in a training, uh, case. Then you get into this other realm you were talking about, about, okay, if enough people have done this and we know it's outputting, um, you know, questionable content or things that have bias in them, or, uh, demean certain, um, you know, demographic or geographic background, whatever.

[00:10:54] Paul Roetzer: That that, that's a massive issue too. And I think the other thing we're gonna see with all of these is there's just a race to release things right now just to ship products and. Again, the big companies like Google can't do it, but these, all these other little companies are gonna spin out and just start releasing crazy stuff that's gonna have all these inherent flaws and biases in it.

[00:11:17] Paul Roetzer: And I, I don't wanna become numb as a society to it, but it sure feels like that's the direction we're heading is a bit of a numbness toward privacy concerns and um, you know, invasion of privacy stuff.

[00:11:30] Mike Kaput: Yeah, and I think that's a really good point too. Maybe for some of the listeners that are just starting to discover this technology, thanks to things like Lens is.

[00:11:42] Mike Kaput: You would think like, well, aren't there a ton of rules and laws around this? Why are they able to do it? And it's like, well, no, there's no . Not yet. I mean, there's not anything that is dramatically overreaching in my experience, that would prevent anyone from doing what Lens AI is doing. And to your point, if they can, they will.

[00:12:01] Mike Kaput: And we've talked about that idea in the marketplace is this stuff is moving so fast and people are releasing things so fast that there's not a lot of guardrails outside. Personal preference, I suppose, in terms of the founders and the people releasing technology.

[00:12:18] Paul Roetzer: Yeah, and I, I just don't, like you say, I don't see 'em coming like there.

[00:12:22] Paul Roetzer: I don't know how, I don't know how the government keep up with where this is going and how quickly it's going. Like I put something on LinkedIn, This week around, or maybe zend around education, like, I feel this way about education at all levels. Grade school, high school, college doesn't matter. There's no way to keep up with what AI is enabling.

[00:12:42] Paul Roetzer: But, but it, you know, from an education perspective, you have to at least start asking the hard questions around, what is this tech? How's it gonna affect us? How's it gonna affect the way, uh, you know, people learn and their experiences they have in and outta classrooms. And I feel. Government needs to be asking a lot of questions, but I, I just don't know how they keep up.

[00:13:02] Paul Roetzer: Like, I don't know that there's enough people within government that even understand what's happening, um, to really do that much about it. But then we're in a situation where we're relying on the tech companies who obviously have incentive to not have a bunch of guardrails, , um, to self-police. And I don't, I don't see that getting, getting us there either.

[00:13:24] Paul Roetzer: It's a really interesting time and it's an interesting space. There are a lot of tech leaders who are very aggressively pushing open sourcing of all this stuff and getting the AI in the hands of the users. And their belief is that society will figure it all out on its own, that it's better that you democratize access to this power and let the people figure out where the limits.

[00:13:48] Paul Roetzer: Rather than allowing a few select tech companies to keep the guardrails in place or the government. So I, I don't know how you put it back in the box, I guess is what I'm saying. I think we're, I think the future is rapid advancement and capabilities, rapid deployment through open source technology, and for better or for worse, the reliance on humanity to figure.

[00:14:14] Paul Roetzer: Where the guardrails are, um, and how we handle this capability. Cause it's only gonna get more powerful.

[00:14:21] Mike Kaput: Those are some great thoughts. That's a great way to put it. Um, I like that a lot. And in terms of kind of figuring out how. To figure all of this out, you know, our second topic to cover today is that, you know, there's some of the bigger consulting firms out there and companies that do research are actively looking into artificial intelligence adoption across, uh, businesses.

[00:14:45] Mike Kaput: So McKinsey actually for five years now, has been conducting its global survey on ai and they survey thousands of business leaders on how they use artificial inte. And this year's findings are now actually out in the state of AI in 2022 article or um, summary that they put out. And there's a ton to unpack in that.

[00:15:07] Mike Kaput: So it's definitely worth checking out the show notes and the link, but. Three big things really jumped out at me, and I know you've got a lot of thoughts on connecting the dots here, but first, um, they found over the last five years of data that they've collected, the AI adoption has more than doubled in the last five years based on these self-reported findings.

[00:15:27] Mike Kaput: Um, They also found that the average number of AI capabilities being used by organizations, uh, like natural language generation, computer vision, those have also doubled. And that natural language text understanding has become one of the more leading AI applications being used now. In addition to that, they're also seeing AI investment increase.

[00:15:49] Mike Kaput: And what's really interesting is, aside from the increases year over year, uh, today, the biggest reported revenue impacts from AI investments, um, in what area of the business is now actually in marketing and sales, which is really interesting to see. And that's compared to areas that were not related to that at all.

[00:16:11] Mike Kaput: Um, back in 2018, things like manufacturing and risk management were kind of the bigger applications where companies were using AI to actually achieve revenue increases. And then last but not is kind of what really jumped out at me as well, is that when it comes to gaining more AI talent, They found that the most popular strategy among respondents was actually re-skilling their existing employees, given how difficult it can be to source AI talent.

[00:16:41] Mike Kaput: Now this is across both. You know, um, computer programmers, data scientists, machine learning engineers, more building the technology, but it's also related to analysts, to AI product managers, uh, and business analysts who are working on these projects. So, and the most common approaches there are actually things like self-directed online courses and certifications, which we just talked about and I thought was really fascinating.

[00:17:08] Mike Kaput: I mean, what were your kind of initial thoughts? The findings, McKenzie put.

[00:17:14] Paul Roetzer: So, yeah, I mean McKinsey is obviously one we follow for a long time. Anybody who's attended, like our intro to AI for marketers class, or really heard me give any keynotes on the topic, will often cite the McKinsey Global Institute study from 2018 that looked at the nine business functions, including supply chain marketing and sales ops, and compared, um, like the the growth potential, where the value creation was gonna come from, and even in then.

[00:17:41] Paul Roetzer: They found marketing in sales to actually be the largest value potential area. So I, I think the. I don't know. I don't know that it's discouraging, but you know, you talked about that adoption has doubled since 2017, but it also has largely plateaued over the last three years. So it's kind of been resting like right around like 50% or something of the people they survey and I think they surveyed around 1500 executives, I'm assuming it's mostly large enterprises.

[00:18:08] Paul Roetzer: I, I glanced at the. Um, methodology. It didn't say the size of company, but I, I'm assuming it's like billion dollar plus mostly companies based on McKinsey's, uh, history. So, you know, I think there has been a bit of a plateauing. You know, we've talked, I, I think I've talked on the podcast before about the Firmi paradox and how I felt that way with AI of all these years.

[00:18:30] Paul Roetzer: So again, Quick recap. Firmi Paradox is, um, you know, Firmi looks into the sky, I don't know, hundreds of years ago, says, where's all the intelligent life? Um, so we have this vast view universe, and yet there's, there's no intelligent life that we're aware of out there. It's a paradox. Um, I've always felt that way with ai.

[00:18:46] Paul Roetzer: It's like, it seems like it should just be everywhere. Like you look at what AI is capable of doing. . You look at intelligent automation, you look at improving personalization, you look at improved and enhanced decision making and creativity and predictive capabilities and all these ways you can reduce costs and accelerate revenue.

[00:19:01] Paul Roetzer: It seems like, well, obviously everybody should be doing ai and yet we look around and we try and find CMOs who understand it and brands who get it and are willing to talk about it. And we couldn't, like for years we struggled to find case studies and I was, it was always. Frustrating to me and I couldn't figure out why.

[00:19:20] Paul Roetzer: And then I, you know, a lot of different things kind of came to light in the last few years as to why that was happening previously. And so I think in general, um, I'm not surprised that it's plateaued. I do think that that is going to change in 2023. So I think what we have experienced with generative ai.

[00:19:40] Paul Roetzer: and the accessibility of the technology, the simplifying of use cases, um, the movement toward, you know, no code, less training data, all of these things that are making AI way more accessible now, I could see that dramatically increasing to where it goes from like 50% and, and maybe it jumps 10 full points next year or something.

[00:19:59] Paul Roetzer: I could, you know, really see it jumping. Um, so I think there's gonna be an acceleration of it. So, um, the re-skilling part jumped out to me as you had highlighted, because I don't know what the other option is. Honestly, like if you think about the building and integration of artificial intelligence historically has been a technical endeavor.

[00:20:18] Paul Roetzer: Like you've needed AI researchers and machine learning engineers and people who can build things. What we need now is people who understand what the technology is and how to apply it for business cases. Mm-hmm. So we don't necessarily need. To go hire 20 AI researchers or 50 machine learning engineers.

[00:20:37] Paul Roetzer: If you're a brand in any industry, retail, manufacturing, insurance, financial services, doesn't matter. What you need are marketers and ops people and business leaders who can be trained in what AI is and what it's capable of doing. Like figure out what predictions need to be made and tell the machines to make those predictions and then know what to do with the predictions.

[00:20:58] Paul Roetzer: I think that there needs to be a massive, uh, reskilling of the workforce very quickly because there are gonna be some roles and specifically tasks within roles that are gonna be obsoleted very, very fast. . And so there's an opportunity to reskill people now, uh, as we know, universities aren't gonna shift fast enough to do it.

[00:21:18] Paul Roetzer: Hmm. Uh, in our own research with Drift, we do the state of marketing every year. We know it's not happening internally. I think it's like what 13% of companies have any sort of internal AI training. We ask about what are the barriers to adoption Two years in a row, lack of education and training has been the number one.

[00:21:35] Paul Roetzer: 63%. I think of people said this year that that was the main issue. And that's, that's a big part of the reason why we built the piloting AI for Marketers series is like we know that people need it and there's nowhere to go for it. So our feeling was, well, let's build a step-by-step journey to get people there.

[00:21:51] Paul Roetzer: So I think, uh, so one, I think it's gonna accelerate the adoption number is gonna. The re-skilling is essential and, and we have to do it. And then they get into, um, looking at what they call, was it high performers? I think? Yes. Whatever they call their classification. Yeah. So they get into what, like what separates the companies that are seeing the disproportionate value creation from ai.

[00:22:16] Paul Roetzer: And I, I liked the chart. So there's an exhibit, and again, the link will be in the show notes. You can check this out. Um, but it's says, organization scenes. The highest returns may are, are more likely to follow strategy data models, tools, tech and talent best practices. And in strategy, the number one indicator is have a roadmap that clearly prioritizes AI initiatives.

[00:22:37] Paul Roetzer: Linked to business value across the organization. Seems obvious, and yet find me one that has that like, you know, think of if you're listening to this, think about your own organization and say, is that true? Do we have that? Chances are no, and you probably don't even have anybody internally who can build it.

[00:22:52] Paul Roetzer: That's, that to me is like the biggest fundamental issue right now is the, the, then you go to number two, have an AI strategy that is aligned with the broader corporate strategy and goals. Senior management is fully aligned and committed to organization's AI strategy. Next is have a clearly defined AI vision and strategy.

[00:23:08] Paul Roetzer: Like just take those, what is that? Four, and like ask yourself as a listener, like, do you, do you have any idea how to even get to those four things, like chances are no. If you're talking to Fortune 500, sure. Like they maybe have hundreds of AI people on staff, AI engineers, researchers, ML engineers, whatever.

[00:23:26] Paul Roetzer: Um, they can get there. They have CIOs, they have chief digital officers, chief data officers, they got everything. If you're a middle market company and you have no one on staff that has AI or ML in their title, chances are you have no clue how to get to those four things. And so that to me is the major, um, obstacle in 2023 and beyond.

[00:23:46] Paul Roetzer: Is you have the this critical need to understand AI and build a roadmap for the next few years, and we have a lack of business talent, not technical talent. We have a lack of business talent who get that and know what to do about it. . And so that's, you know, we've alluded to this before with, with the institute and some of the work we're looking to do more in the consulting realm is there's just no answers.

[00:24:09] Paul Roetzer: There's nowhere to go for this. And so we see almost like a necessity to build, um, solutions to help organizations figure this stuff out because I, I, we talk to 'em all the time, they don't have a clue. And again, I have, I have lots of CEO friends, lots of friends in VC and private equity, and nobody knows what.

[00:24:28] Paul Roetzer: So that was, that was my main takeaway is like, it kind of looks like the previous year's report, like not a, not a hell of a lot is different in terms of the findings. The numbers moved a little bit here or there, but the overall takeaways are about the same. But I feel like the end of 2023 is a very transformational period.

[00:24:47] Paul Roetzer: Where all of a sudden all these corporations are gonna figure out that they need to understand and apply this stuff, and we're gonna see a lot of change in this, these numbers, and hopefully in their findings come fall of 2023.

[00:25:00] Mike Kaput: Yeah, absolutely. And you actually, um, have the honor of being the focus of our third topic this week because you actually highlighted very recently on LinkedIn, uh, and on Twitter.

[00:25:13] Mike Kaput: How software companies need to actually have a concrete plan for considering how AI will enhance and disrupt their business. And I haven't checked the post in the last hour or so, but I know that you've gotten a ton of comments and really interesting conversation on LinkedIn specifically about this idea.

[00:25:32] Mike Kaput: That specifically, you mentioned that quote, if I was running a software company right now, I would be aggressively exploring how recent advancements in AI and the ones that are coming in 2023 could disrupt the business. Now you give some tips in the post as well that we can walk through. But first off, I wanted to know why are you writing this now or why is this on your mind now?

[00:25:56] Paul Roetzer: Listen, you, we talk to lots of SaaS companies. I have friends who run SaaS companies. We have partners of ours or the institute that are SaaS companies. Um, you know, I built my agency around SaaS companies. So we've lived in that world for, you know, I started my agency in what, oh five and by 2000. 13 or 14, like 80% of our portfolio were SaaS companies.

[00:26:21] Paul Roetzer: Like most of the work I did in the teens was for SaaS companies. So we've kind of been through a lot. We've seen a lot of cycles with SaaS and the more I look at existing SaaS and our tech stack, so you think about the companies we use as the institute do what we do as a media vet and education company.

[00:26:39] Paul Roetzer: Um, if you think about. The different software that we're being exposed to and the things we're learning about with generative AI and how transformative those could be. Again, if you just go back to last week's episode and you listen to the things we were saying, and then you have that lens when you're looking at software that we use as a, as an institute today, it's all I can think about is like, wow, they're, they're in a tough spot.

[00:27:05] Paul Roetzer: Like so many of the software, uh, applications that we. I could tell you how to build a smarter version of them right now, like, I mean, you and I do it all the time. Like there's all these manual, repetitive processes in every workflow that we have to do. Our marketing, our sales, our service, our operations.

[00:27:25] Paul Roetzer: And so you just think about all the software we use. I mean, we're a five-person company and we probably have what, 25 or 30 SaaS products that we use to run our company in all these different areas. And I would say, I don't know, like 10 to 20% of 'em are legit AI technology. So we have all of this legacy software that's running our company that's about ai , like we, we have an AI company being run by legacy software that has no intelligence in it.

[00:27:54] Paul Roetzer: And so I, I think part of it is from just a constant awareness. And then as we were building the piloting AI courses, it became very top of mind for me. I was thinking a lot about. And then just a number of recent conversations, cuz I also get into like VCs and private equity and angel investors. Um, I have a lot of friends who are angel investors.

[00:28:16] Paul Roetzer: I have friends who run venture funds. I have friends in the space. And then we follow a lot of like, you know, the VC world on Twitter. Like I, you know, kind of keep close tabs on what's going on there. , and I know a lot of these people invested heavily in sas, uh, rightfully so. I mean, it has been the place you made money the last decade.

[00:28:36] Paul Roetzer: Um, and I've told of, you know, different people, uh, in conversations like, Hey, heads up, like as a friend, I'm telling you, take stock of your SAS portfolio, whatever SAS you're invested. Make sure they know what the hell is happening because like 2023 is different. And they need to, they need to have a plan.

[00:28:57] Paul Roetzer: And so after having said this so many different times in private conversations, I was like, I, I, this needs to be said out loud, . Like if you run a SASS company or if you invest in SASS companies, I really, and I, I've said it in different topics, like I don't know how you have a business in three years if your, if your product isn't infused with, So someone is gonna come along and build a smarter version of whatever it is you do.

[00:29:22] Paul Roetzer: I don't care what your software does, and I don't care if it's a point solution or a platform. The AI that we're seeing now and what we're gonna see next year is going to enable a lot of different people with the vision to build smarter software to come in and take your market share. And so, I think it was just one of those, like spur of the moment.

[00:29:43] Paul Roetzer: I mean, literally I was typing that on my phone, like I was, we were waiting to go to dinner last night or something, and it was like I had this thought. I was like typing it out on LinkedIn and told my family, hold on a minute, gotta get this thought outta my head. So yeah, it wasn't anything like, I sat around and set an editorial calendar and said, okay, on December 12th I'm gonna publish this thing about SaaS companies.

[00:30:02] Paul Roetzer: Uh, it was just a collection of a bunch of conversations and current events that sort of led me to be like, okay, I just gotta put this out there, kinda like I did with the education thing last week. It's just, yeah, I dunno. Sometimes you just have these thoughts and like dots connect in your head and you just wanna say something.

[00:30:17] Paul Roetzer: Yeah, but it's not just software. I mean, that's a number of comments called out like we're, it's universal. It's software in particular, like it's just mm-hmm. so critical.

[00:30:26] Mike Kaput: So I guess one point I wanted to add here or ask about is that, you know, as part of your post, you had mentioned that SAS businesses may not be able to handle the speed of AI advancements coming next year.

[00:30:39] Mike Kaput: So why do you think that is? I. It would strike me that some of these companies should be on the bleeding edge of what is technologically possible. I realize a lot of advancements have come out of nowhere, almost, uh, in the world of ai, it seems on how fast they've moved, but why have they been caught Flatfooted.

[00:31:01] Paul Roetzer: I don't think most executives understand ai, and I don't think the investors who have enabled them to build to the point they're at and understand ai. I, I just think understanding of AI is so limited across the business world right now that they just didn't see it coming. Like, you know, and, and, and part of it is the, a lot of.

[00:31:22] Paul Roetzer: SaaS products are great. The companies are great. They're run by phenomenal, brilliant people, and AI just wasn't their thing like it. Mm-hmm. , maybe it just seemed. To them to sci-fi or too abstract or just not a huge priority. And I mean, you can, it's not hard. Like take the publicly traded companies, go get their earnings reports, how many times they mention AI in the last two years.

[00:31:44] Paul Roetzer: Um, go to LinkedIn Sales Navigator. I think I've mentioned this kind of hack before, like go on LinkedIn Sales Navigator and just run a keyword search for job titles of how many AI and ML employees they have in the company. And if you're talking about software companies with hundreds or thousands of employees and they have like, Or 10 AI ML people, there's a really good chance that they don't value AI enough yet.

[00:32:06] Paul Roetzer: And I, I think that's gonna change. But, um, and we've watched it. I've been, I've never published this data. I've been running these numbers every month for the last like, year of the number of AI ML people at these different companies and trying to monitor and see if there was a change. If, if we hit an inflection point where all of a sudden you would start to see this increase across different industries or companies.

[00:32:30] Paul Roetzer: So, yeah, I, I don't know. I mean, I think it's just three years ago, I, I could have heard an argument that we weren't at a point where you could truly commercialize artificial intelligence, yet we hadn't had the major breakthroughs in language and vision capabilities that are enabling what we're seeing right now with generative ai.

[00:32:49] Paul Roetzer: And so I guess you could say, you know, people were kind of busy trying to figure out the pandemic along the way and yeah, trying to deal with the economy. And it's not like CEOs had nothing else to think about for the last three. . But if you took the last three years to just survive and kind of figure out where the next round of funding was coming from or how to avoid layoffs, then you were doing important work.

[00:33:09] Paul Roetzer: But you may have in the process if you didn't have the right people on your team missed what was happening with ai and now you gotta play catch up real fast.

[00:33:17] Mike Kaput: Yeah, I mean, it's, for anyone listening, please go read the entire post, because the comment section is special. You're going to realize how, how common this is happening across different industries, different types of businesses, and I think we're seeing a lot of people wake up to what you just talked about.

[00:33:34] Mike Kaput: Are you ready for a couple quick rapid fires before we wrap up here? All right, let's do it. First up. So a company called CO Here, which is one of the top, uh, kind of AI outfits in the world, um, has some of the bigger names in AI behind it. They just released a multilingual language model that is one of the, um, most accurate in terms of being able to work with over a hundred different languages in terms of producing accurate language results across those languages.

[00:34:06] Mike Kaput: And there's three areas that the model. Is being tested out as being possible use cases for, so there's semantic search, so multilingual semantic search that improves the quality of search results across all these different languages. Again, I think it's more than a hundred. Um, Aggregating customer feedback.

[00:34:28] Mike Kaput: So you can actually organize crust customer feedback across hundreds of languages. And then there's also cross lingual content moderation, so you can actually identify harmful content, uh, in hundreds of other languages. Now, I think it's still early days for the model, but I just wanted to kind of get your thoughts on this just given how quickly we went from.

[00:34:50] Mike Kaput: Having on our end, you know, models that work in our native language astonishingly well to suddenly, oh my gosh, a hundred plus languages may now have these incredible capabilities too.

[00:35:02] Paul Roetzer: It's a huge growth area and obviously if you're a, you know, international, Uh, organization. This stuff's gonna be super relevant to you.

[00:35:10] Paul Roetzer: There are a number of big players. Most of the I research labs are working on rapid advancements and multilingual, like Meta does a ton of stuff with this, and I know they've had some breakthroughs in 2022. I'm a huge fan of cohere. I think that it's a company to watch for sure. Uh, it was had support of Jeff Hinton, um, you know, who was the, kind of the guy who coined the phrase deep learning back in 2011, 2012.

[00:35:36] Paul Roetzer: Um, so I, I would be paying very close attention to anything cohere does. They're very legit company. I have zero inside knowledge, but I would imagine like Google and other companies are gonna be, you know, tripping over themselves to maybe a acquire co here if co here wants to be acquired in the, in the future.

[00:35:55] Paul Roetzer: Um, I know one of their co-founders actually came outta Google Brain, um, Aiden Gomez. So he was a, he did student researcher, Google Brain Team and, um, Hinton's Company in 2012 got acquired by Google. So there's like some history there. That was part of the reason I think that. So, yeah, I mean, just. Again, these are so many of the things I think we try and present on this weekly is the stuff to keep an eye on in the future.

[00:36:22] Paul Roetzer: Like sometimes we'll give you, just go try lens or go try do Dolly, or go do chat G P T, and you can go do the thing today. Other times we're giving you stuff to help you connect the dots as you start to recognize trends and patterns in the innovation that's happening. Multilingual and multimodal are two of.

[00:36:39] Paul Roetzer: The bigger things to think about. Um, multimodal being, you know, video, uh, language, like written text, spoken texts, like the ability to, like different modes. Um, so yeah, I mean, multilingual is a huge area. Pay attention. Pay attention to, okay, here, pay attention to innovations and multilingual.

[00:36:59] Mike Kaput: And so speaking of languages, I mean, I, I, we are also seeing another massive release from DeepMind, uh, a tool called Alpha Code.

[00:37:10] Mike Kaput: And the languages we're talking about here are programming languages because Alpha Code is actually an AI system that can produce computer programs, and it can do it at a competitive level. So, Alpha Code actually achieved. They estimated the rank was within the top 54 per percent of participants in programming competitions.

[00:37:34] Mike Kaput: So machinists actually competing within the top 54%, you know, on the day of release of participants at solving, uh, new coding problems. Essent. The same types of challenges and interviews and problems you would be solving either to get hired as a programmer to contribute to a project as a programmer, or to solve a challenge at your company if you're a programmer.

[00:38:00] Mike Kaput: So they're, you know, deep mind is very clear that this is not replacing everyone who does computer programming, but it is pretty astonishing. We're at a point where it can already do this level of coding work on its own. Um, this quickly.

[00:38:18] Paul Roetzer: Yeah. This was one that, when I saw it, I, I know I tweeted something about it, and I think I actually sent it to a couple of friends because.

[00:38:25] Paul Roetzer: When Chad g p t comes out, you and I are writer, we can go in and assess it. Like I can go in and be like, pH, man, this is gonna change things in five minutes. I'm not a coder. I can't go in and test this and realize it. I can just look and say, oh, science is publishing this paper. Mm. And demos thinks it's a huge deal.

[00:38:42] Paul Roetzer: and Jan la Koon's retweeting it from Meta's Research Lab. So we, I, I actually rely more on the AI network I follow for the context of how big of a deal is this. Yep. And this, that actually played into me putting that software post on LinkedIn because I look at this and I think, okay, let's say I run a software company and I have 500 employees and 60% of them are developers, like writing code or whatever.

[00:39:09] Paul Roetzer: Do I need them? Mm-hmm. like, does this, is this like a year out there? It's gonna change things because I also saw with chat, G P T, people were using it to write code and they were hacking it to be able to function. Is that, and I, from what I was hearing, it was doing really well. Yeah. And so if you say, okay, so.

[00:39:27] Paul Roetzer: Chat, G p T can be used to write code. Uh, this alpha code thing seems to be getting really good at writing code. Meta did some stuff and you start to, okay, this is one of the themes of investment where they're doing r and d and it sure seems like they're already made making major advancements. And again, if I run a software company 12 months from now, am I gonna see a leap forward and is it gonna change the dynamic of my hiring?

[00:39:52] Paul Roetzer: My entire HR system. I have no idea what the answer to that is, but it sure seems like it's something I would be wanting to ask right now. If I was running a big software company or if I was building, um, things I would probably want to be understanding, is AI going to be able to do some of what my people do or at least make my team more efficient?

[00:40:11] Paul Roetzer: I'm not saying I have to get rid of parts of my team, but can I double our production next year? Can I ship twice as many features? Mm-hmm. , because I can. New AI coding capabilities to increase our productivity and our output. Again, I have no idea, but I It sure seems like it's a question worth asking. If you build software for a living,

[00:40:33] Mike Kaput: Yeah. And I think additionally with that, not only building the software, but who does the software serve, right? So you mentioned we have all this, um, legacy technology and it's like, well, why aren't the SaaS tools we're using, functioning this way for marketers? Right? You know, why aren't, why don't we have more co-pilots in the existing tools we already have that are powered by ai?

[00:40:55] Mike Kaput: That would be. in immediate place, I would be looking to create value, whether, I mean A, whether it's in analytics, writing, advertising, what have you. We have tools that do all these things, but do any of our existing platforms really enable that kind of 10 x or a hundred x productivity gain that some of decoders may be getting in your

[00:41:15] Paul Roetzer: future?

[00:41:15] Paul Roetzer: Right? Yep. Yeah, for sure.

[00:41:18] Mike Kaput: Um, all right, the last topic here is runway ml, which we talk about pretty frequently. They just raised a massive amount of funding, uh, at a 500 million valuation. But what's fun and um, Astonishing about runway ML is how quickly they appear to release new features. They come out, it seems like with a new feature, a new product once every week or two.

[00:41:42] Mike Kaput: And the most recent one, they're a suite of all these different products that are helping you create and edit content. Uh, the most recent one is called Backdrop Remix, where you can give any photo infinite backgrounds, which means I could take a picture. Say a product that we're selling or something I want for a marketing campaign or a piece of collateral, and then tell it to put in multiple different backgrounds just generated from scratch based on whatever I would like to see.

[00:42:10] Mike Kaput: So if I don't wanna see it in a living room and I wanna see the product say in a garden or outside or what have you, it's very simple to just swap that out and create hundreds of variations of that type of photo. Uh, in seconds. Paul, what'd you think of this when you saw

[00:42:25] Paul Roetzer: it? I haven't tried that one yet, but I do, I think I've mentioned runway before.

[00:42:30] Paul Roetzer: Um, I'm a huge fan. We, they're not a sponsor of ours, so this is just me as a, you know, customer. A user. Uh, I love it. I mean, I created an infinite image example. I might have talked about that last week for our piloting AI series. So they have. I, I don't know, about 30 or so, pre-trained, whatever they call 'em, applications, features, tools, I think is what they call 'em in their AI magic tools.

[00:42:53] Paul Roetzer: And they're, they're like stupid, easy to use. Like any, anybody can go in. You don't have to be a designer, a videographer, anything to go in and, and do these things. And I think that's what I love about it. I'm pretty sure the pro license is like 12 bucks a month, something like that if you pay annually. So I think I paid 144 bucks.

[00:43:11] Paul Roetzer: One year subscription and, uh, just go and play around. And they do, they , I mean, they've dropped like five new tools in the last three weeks, I feel like. And it's definitely one of those like ShipIt cultures. They have a lens like app where you can go in and train it yourself. They have an ai, they just introduced an AI training capability where you train it on your own data sets and then it has the ability to infuse those into, I think, If I wanted to train, train it on, say my face, and then build an infinite image on my face, I could do that.

[00:43:43] Paul Roetzer: I could train it and then pull that image into the infinite image. And it's just really smart stuff. And so they're a company that I would just pay attention to, pay a few bucks, give it a try, you know, test it out for yourself. It's just a fen phenomenal way to experience ai. And anytime you get to experience a tool like that, you just develop a deeper understanding of what's possible.

[00:44:02] Paul Roetzer: And then, I mean, the thing I. In the course, uh, after I build the show, how to build the infinite images that your imagination is your only limitation. Like, that's the way I feel about AI right now is it's anything is becoming possible. It's just understand what it is and what it's capable of. And then it's just whatever you can imagine building or, or creating or producing.

[00:44:25] Paul Roetzer: It seems like you can or you will be able.

[00:44:29] Mike Kaput: That is a great place I think for us to end it here. That is really empowering and I think people can also realize these tools can be incredible examples for your leadership team, for your c e o, for the people that you're trying to talk to and convince that might not exactly get it.

[00:44:44] Mike Kaput: So Paul, thanks again as always for awesome insights into the world of ai. I am sure as we speak, uh, the brief. Next week's show is already getting overpopulated here. .

[00:44:58] Paul Roetzer: Yeah, it is. It's wild. Maybe we'll do, yeah, we'll probably drop one next week, cuz then we're heading into the Christmas holiday. Um, so yeah.

[00:45:05] Paul Roetzer: We'll, we'll drop one. I and I think we're working in between We'll, we'll try and keep on schedule. So for our regular listeners, which increasingly we're hearing from, and thank you again if you reach out. We love hearing feedback. Um, I think we'll try and, uh, try and stay on schedule through the holidays.

[00:45:21] Paul Roetzer: Uh, it's fun to do. I feel like if we wait until early January, no, man, we're gonna have a three hour episode and play catch up, so, uh, okay. Yeah. Thanks. Thanks again for joining us. Reach out. We love hearing from you. Be sure to, um, subscribe that we do, uh, not only publish these on podcast networks, but it is on YouTube.

[00:45:40] Paul Roetzer: So if you ever want to, you know, watch it on YouTube or check it out there, um, you can also go get, get it from there. And then the latest episodes are always, uh, added to our blog as well. So subscribe to the Institute newsletter and then again, if, if you're ready to figure out all this stuff and what it means to your career in your business, check out that piloting Act for Marketers series.

[00:45:58] Paul Roetzer: Do you have any questions? Cathy, from our team is usually active on the chat there. It's actual human, it's not chat, g p t . Um, you can, you can actually interact with a, a human, um, during normal business hours and even after sometimes. Uh, so yeah, reach out and have, ask us any questions you'd like. And other than that, um, till next week we'll talk to you then and uh, have a great week.

[00:46:19] Paul Roetzer: Thanks Mike. Thanks Paul.

[00:46:21] Paul Roetzer: Thanks for listening to the Marketing AI Show. If you like what you heard, you can subscribe on your favorite podcast app, and if you're ready to continue your learning, head over to marketing ai institute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.

[00:46:43] Paul Roetzer: Until next time, stay curious and explore ai.

Related Posts

[The Marketing AI Show Episode 58]: Big ChatGPT Updates, A New Autonomous AI Agent, Vertical-Specific LLMs, McKinsey’s State of AI Report, and New Google AI Search Features

Cathy McPhillips | August 8, 2023

This week's Marketing AI Show covers ChatGPT updates, autonomous agents, vertical-specific LLMs, McKinsey’s State of AI report, and new AI search features from Google.

McKinsey Predicts AI Will be as Impactful as the Steam Engine

Ashley Sams | September 12, 2018

Our team reads tons of AI and machine learning news every week so we can share the top stories with you. Machine learning articles worth reading this week include McKinsey's AI predictions, Google's new What-IF machine learning tool, and some of the most notable machine learning use cases.

McKinsey: AI Will Have the Most Impact in Marketing

Ashley Sams | April 26, 2018

McKinsey’s study of artificial intelligence in more than 400 use cases in companies across 19 industries concludes that marketing and sales are one of the top fields to be positively impacted. Here are the highlights.