This week on The Marketing AI Show, Marketing AI Institute founder/CEO Paul Roetzer and Chief Content Officer Mike Kaput talk about three AI stories in the news and add their take on these developments and what it means for marketers and business leaders.
On October 4, the White House released what it calls an “AI Bill of Rights,” a document that offers a blueprint of “five principles that should guide the design, use, and deployment of automated systems.” It is not binding in any way, legally or otherwise. But it's an important initial effort by the US government to draw attention to the impact of artificial intelligence on our daily lives.
In an exclusive interview with McKinsey released in late September, Dr. Kai-Fu Lee, a world-leading AI expert, investor, and author spoke with McKinsey about how business leaders can use AI in their operations and what it means to be a truly AI-driven company.
Meta, the company formerly known as Facebook, just released Make-A-Video, an AI system that turns text prompts into machine-generated video clips. This means you can type in a prompt, like “a horse drinking water,” and Make-A-Video will understand the prompt and create a video clip of it in a specific style. This type of generative AI does for video what a tool like DALL-E 2 does for images: it creates unique visuals from a simple text prompt in seconds. The tool isn’t yet publicly available like DALL-E 2 but has major implications for businesses and creators.
Timestamps
00:02:29 AI Bill of Rights
00:13:26 Dr. Kai-Fu Lee
00:27:17 Meta’s Make-A-Video
Links Referenced in the Show
- Blueprint for an AI Bill of Rights, White House
- Building an AI-driven company with Kai-Fu Lee, Chairman and CEO, Sinovation Ventures, McKinsey
- Make-A-Video, Meta
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: 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.
[00:00:20] My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.
[00:00:29] Thanks for joining us for episode 20 of the Marketing AI Show, the podcast that makes artificial intelligence approachable and actionable for marketers and business leaders. I'm your host, Paul Roetzer. He is my co-host Mike Kaput. Mike, how's it going? Good. Mike is the chief content officer at Marketing AI Institute and also my co-author for our book, Marketing Artificial Intelligence, AI, Marketing, and the Future of Business, which is available now in print, digital and audio, which was an adventure. I did record the audio 18 hours in the studio. Let me know how it goes. I will not listen to it. So if you listen to the audio, let know. All right, before we get started with this week's episode, I want to take a moment to thank our sponsor, Persado.
[00:01:14] Persado is the only motivation AI platform that generates personalized communication to immediately motivate individuals to engage and act. Organizations that use Persado benefit from an extensive customer motivation knowledge base, enabling them to hyper personalize their communications at scale.
[00:01:33] Clients who incorporate Persado's innovative motivation AI increase conversions by an average of 41%. Unlocking millions in unrealized revenue. I wouldn't mind doing that. . We need to be using Persado to motivate people to buy books and online courses and come to events. I agree . So check out persado.com for motivation ai and thanks again to Persado for being a partner with Marketing AI Institute.
[00:02:02] So we got a lot to cover today. We're going to do it in about 40 minutes or less, maybe a few more. Let's get into the,
[00:02:09] Mike Kaput: Great. We've got a few really exciting topics I want to kick us off. We're going to take a few topics, few interesting developments, things in the news. There's a ton going on in the world of ai and I would say the past few months especially have been particularly crazy and exciting, which we're going to kind of dive into.
[00:02:29] But first off so actually last week and on October. The White House actually released what it calls an AI Bill of Rights. So this is a document put together by the science team on the, at the White House that offers a blueprint of these five principles that they have identified. And these are supposed to guide the design, the use, and the deploy.
[00:02:57] Of automated systems. And so we'll kind of dive into what those five principles are. But at a really high level, they've outlined principles like developing safe and effective systems putting in place protections for algorithmic discrimination to prevent that, ensuring data privacy, providing enough notice and explanation of how AI tools are being used to consumers, and then also offering human alternative.
[00:03:25] Considerations and fallbacks if something goes wrong or if consumers want the option to do so. So before we dive in, it's really important to kind of note that they're actually the AI bill of rights That's not in any way legally or otherwise binding from the White House, but I, I do think despite there is a lot of commentary.
[00:03:44] Rightly so online that, you know, perhaps it could have more teeth, it could be expanded, but I actually do think we both agree. It's a really important and interesting step forward with the government actually getting involved in how artificial intelligence works, how consumers engage with it, and how it impacts our daily lives.
[00:04:05] So that, that's kind of what immediately jumped out to me. We could dive into each of these. Principles, but I think the first point I would have is at a high level, this is definitely. An important development regardless of, you know, how much of it actually ends up being implemented into law? Like, what did
[00:04:23] Paul Roetzer: you think about that?
[00:04:25] Yeah, I mean, when I saw it, for sure, I put it on LinkedIn right away, and I think what I said was not legislation, you know, not anything formal, but it, it is a, an important step forward. I think what concerns me and what we often talk about the institute and you know, some of our content, and I think even in the book we have a chapter on responsible ai.
[00:04:44] Is consumers are certainly aware that their data is cons absorbed by tech companies like they, they know their data is going to use. Whether it's informing an algorithm within TikTok or you know, what, predicting purchase behavior. Like they know that the data is out there and that it is being used by brands and technology companies to make predictions about their behavior and to drive actions.
[00:05:10] I think generally people would get that concept. How it does it and how the AI works is a very abstract thing to even business leaders we talk to. Mm-hmm. . So to the average consumer who is targeted by ai, You know, to drive either personalized ads or communications or offers of promotions or, you know, even when it starts working into making predictions related to health or finance or wellness or risk, you know, insurance mortgages.
[00:05:39] Like there's all these ways. AI is just everywhere in our lives today, and the average consumer has no clue, like how that stuff works or that it's even there underlying the technology. And so I don't think it's reasonable to believe that consumers. Can, can really understand this and take control at a, at a wide scale level of their own, like protect themselves basically from the ai.
[00:06:05] So then the next step is you rely on the companies that are building and using the AI to make decisions that are always in the best interest of consumers. And let's be honest, that is not going to happen. So while tech companies can set their own guidelines and put guardrails up to protect the consumer, at the end of the day they have.
[00:06:25] They have financial goals they're also trying to achieve, and there is often a, a, a imbalance internally within these companies about finance over. Human good and wellbeing, I guess. Yeah, the AI for good, and we've seen this play out in, you know, some pretty high profile things that like Google and like target the, you know, the historic example of Target, you know, using some data to make some predictions about customers that backfired.
[00:06:48] But that it's, you know, it's everywhere. And so I, I just, I feel like. Government needs to get involved. I'm not necessarily big government guy, like everybody, you know, government should come in and do everything, but I, I don't think in this case that the tech companies alone are going to solve this. I don't think they have the motivation to solve this at a very high level.
[00:07:06] Yeah. So, yeah, I mean, I think it's, I think what people do is read it like, we're not going to spend next 45 minutes talking about the Bill of Rights and going through, like it's a blueprint for they're, they're basically laying out what it could. I think everybody should read it or at least read an in depth summary of it and see what they're trying to do.
[00:07:20] And I think what you should do is connect it to your company. Like our belief is that every brand and every tech company should have like a, an ethics guidelines, an AI ethics guidelines. So if you're a brand buying AI technology or leveraging it on the data you consume, You should have rules in place that guide your team on how you are going to use what AI enables because it gives us superpowers and we have to be able to have guidance for our teams internally about how we will use this.
[00:07:52] Now the same goes for if you're building it. I think we use the example of Adobe in the book and their ethics policies around the development of AI technology. It's the stuff is moving so fast, the guardrails are going to have to keep being rebuilt and moved and rebuilt and moved because things that weren't even possible three months ago, we'll talk about some of it in, in the next topics.
[00:08:12] Things that weren't possible are now possible, and now the questions will, well. Do we use image generation technology? We vi video generation technology, Do we create synthetic reps? Like all these things that most companies don't even know you can do. And so that's where I think it's just critical that not only is the government stepping in and saying, Okay, we, we need a point of view on this and we need to give some guidance.
[00:08:36] I think the action step for marketers and business leaders is you cannot font this, like push it forward for a few more years. Like, Oh, we'll get to the ai. It's, it's too late then like you need to be thinking from the ground up. How do we ethically apply AI to what we do? And you have to have the, the consumer.
[00:08:54] Has to be human centered. It has to have the consumer in mind in every decision that's made. So that, I mean, that's kind of my high level takeaway from this without getting into all the details. Yeah. Is the impact for the individual brand and for the tech companies building it is the importance of thinking about the ethical use of ai.
[00:09:09] Responsible use of AI in your company?
[00:09:13] Mike Kaput: Yeah, I, I couldn't agree more. And I think that when you say business leaders should go read this, it's not only interesting and important to read, it's actually, I think a decent set of guidelines. For what you need to be thinking about in your own company. You and I both were just at the Digital Now Conference in New Orleans, and I remember when we ran a workshop there with a bunch of different nonprofit and business and association leaders.
[00:09:40] One of the first questions that people started asking was really, really difficult and interesting questions around ethics and the morality and the. Responsible use of this technology. So regardless of what type of business you run, I mean these five principles. First off, safe and effective systems, you should be protected from unsafe or ineffective systems.
[00:10:04] I mean, business leaders need to understand not all of these systems work in the way they're intended. There's always the possibility for them to go wrong. Even things too, like number two, algorithmic discrimination protections. We have seen many different examples of algorithms that actively or accidentally discriminate against certain groups, certain data types, and a lot of it's unintentional.
[00:10:29] It's not even built. With that intention in mind, but it's on the brand to figure out exactly how these things are working and impacting things like your brand equity, your business operations. I mean legal, there's plenty of legal ramifications. And then just very quickly things like data privacy, understanding how explanations of how the tool itself actually works.
[00:10:51] Oftentimes, the people building it can't always a hundred percent even tell you what is going on behind the scenes. And then last but not least, The, one of the principles is about human alternatives if possible, giving consumers the ability to opt out of some. The results of these technologies, these are all front and center issues that brands are going to have to worry about very, very soon if they're not already.
[00:11:15] Paul Roetzer: And that, that one in particular caught my attention, the opting out. I, I, I think it, it, the, in principle, what they're trying to do is important. Unfortunately, I think what would happen is it'll be like the cookies, like everyone just says allow like fine . I mean, it's almost. When you know, if an example here of it being applied would be like Facebook or Instagram where you can opt out of ad targeting, like personalized.
[00:11:38] You're still going to get ads, right? They just won't be relevant to you. So would you rather see relevant ads or irrelevant ads? Right. Right. But again, that's why it's like, it's an important step. It puts the conversation out there. It sort of stakes these five core areas as things that we should be thinking about.
[00:11:53] And like you said, you should be considering these in your business. And what we hear from you know, like Tim Hayden and Chris Penn, like people we talk to about privacy call Dubey, but pan data. , what I always get told is just assume the regulations are coming, like mm-hmm. , you should just put best practice in place now to make your AI human center.
[00:12:15] Just assume the restrictions are going to be there. We had a speaker at Macon Gemma who was from Europe, and I was saying like, it wasn't it harder to do ai. And she goes, No, no, no. Like we have the benefit because we already have the restrictions. So we're building our AI with more restrictive.
[00:12:31] Oversight and governance, which actually is great cuz you're going to have to do it in the US eventually. We're already there, right. And so her point was like, it actually makes things easier because there's some guardrails in place. So yeah, I just, again, as you think about scaling the use of ai, if you're in a bigger enterprise, and this is like top of mind.
[00:12:50] This is the kind of stuff you've gotta be thinking about from the ground up and figuring out who within the organization needs to be involved in these conversations. Because AI adoption is not just a bunch of individual use cases to truly scale it, you're going to have to become like an AI emergent company where AI's infused into everything and there's going to be a lot of hard decisions that are going to have to be made and, and those decisions going to have to be reevaluated as the tech keeps evolving.
[00:13:16] Mike Kaput: I think that's a really good lead in to the next topic here. Because that idea of AI emergent companies, how do you become an AI driven business, an AI driven company? We found a really great interview released just a couple weeks ago with from McKinzie, where McKinzie interviewed Dr. Kai-Fu Lee, and if you don't recognize that name he is actually a world leading AI expert.
[00:13:40] He's an author and an investor. He gave this really, really in depth and great interview on how business leaders can actually become drive their companies to become ai first AI emergent companies regardless of, you know, if they have historically been a traditional. A more traditional business and you and I both have been familiar with Dr.
[00:14:01] Lee's work, his book, AI Superpowers, China, Silicon Valley, and the New World Order that I remember that, you know, a couple years ago being extremely formative for both of us as we were learning more about artificial intelligence and building out the institute. He has since gone on to become a chairman CEO at Innovation Ventures.
[00:14:21] And so he actually runs, I believe it is out of China. Firm that backs hundreds of AI companies so very much like say an Andrew ing. He is very on the forefront of actually investing in real world applications of AI in the corporate world. So I thought some of these answers he gave and points he brought up were really, really interest.
[00:14:40] I would highly recommend we'll link in those show notes to the entire interview. Business leaders of all types should absolutely read this as soon as possible. It's really, really important. But what really jumped out to me are a couple of things. The first is he is, he pinpoints exactly something that we've been seeing in the last few.
[00:15:00] Which is, we are once again at an incredible turning point in the field of ai, especially as it relates to business. So Kai-Fu Lee actually points out that he thinks we're at a similar stage today to where we were in 2012 when Jeff Hinton and his research team blew the doors off the problem of computer vision.
[00:15:21] So Jeff Hinton and his team were one of many teams that. Crack the code on deep learning and really driving a decade of innovation in artificial intelligence, specifically through teaching machines how to recognize objects identify objects. Now, Kai-Fu Lee thinks we're in the same spot, but in terms of language, he actually thinks that some of the recent developments in language ai, natural language processing, He says, quote, We're about to enter a second golden period of AI investment thanks to this.
[00:15:53] And I think we've seen that in some recent conversations we've had with some really knowledgeable people. Paul, what's your, what's your
[00:16:00] Paul Roetzer: take on that? Yeah, I mean, I agree like. Who leads AI superpowers book was very influential for me to get like a global, like macroeconomic view of where AI was going.
[00:16:11] And then he has AI 2041 was his most recent book. Yeah. So yeah, I mean he's definitely one of those people. I, I look at as one of the main thought leaders, people who are actually doing it and seeing tons of stuff. So I, when he talks, I listen like it, you know, it's somebody I really pay attention to. A few thoughts.
[00:16:27] I I definitely agree on the language thing, and we've been saying that for a while. I, I've said on stage, like, if I was just, if I just ran a billion dollar fund, I would just invest in language tech. Like I, I truly believe that you could just focus on that and make it domain specific and there's just so many applications for language.
[00:16:44] So that, that is key. And as you'd alluded to, We talk to a lot of people, a lot of it's sort of off the records, like research labs, major tech companies and stuff that we can't really get into. And, and what, what, what I generally would guide people is like, if you look at like a DALL-E 2 or stable diffusion or we're going to talk about, you know, metas make a video minute.
[00:17:10] If you look at this stuff and you're, your mind is blown, like how in the world does that stuff? Just assume that the research labs are already 12 months ahead of what you are seeing. So the stuff they're releasing today probably was in a pretty advanced stage a year ago. So, What's happening is all of the sudden it, it feels like in the last like five months, we've hit this tipping point where the AI is very, very real and getting, getting better at an exponential rate.
[00:17:42] So a lot of like the image generation tech, the language generation tech, it's relatively new, but it's advancing so fast. And again, when we talk to the people who are building this stuff or working in the labs on this, The things they're talking about I wasn't hearing a year ago. Mm. Like they're, and, and even, I mean, there's some public stuff, like you could go follow some of the open AI people you know, some of the, the deep mind people like.
[00:18:07] They're, they're much more regularly talking about things like agi, like they're, they're almost talking about like these major leaps that are coming are inevitable and near term. So
[00:18:20] Mike Kaput: agi, meaning artificial general intelligence. So the ability of a machine to do. Everything and anything at a human or beyond super
[00:18:31] Paul Roetzer: level.
[00:18:32] Level. Level. Yeah. Yeah, yeah. Which is why Open AI exists. It's why Deep Mind exists, like there are these people within AI who believe this is imminent and possible, and the chatter. Around its imminence seems to be accelerating. So again, I'm just like, I, you know, I'm not looking at a study where like somebody changed it.
[00:18:50] Oh, within three. I'm just telling you like we talk to a lot of people and you can follow a lot of these top AI researchers, and the conversations are happening very quickly that we are heading towards some pivotal moment in AI's development. Now what I will say is whether that happens or not, Whether we even get to like sentient where these language models understand that they exist and like this thing people worry about.
[00:19:14] And there was the big article in Washington Post a few months back about the Google researcher who thought their language model Lambda had become sentient. Yeah. What we're thing is like, it doesn't even matter. Like if we get to agi, doesn't matter. In the near term because the language models are so good and they're going to get dramatically better as the machine learns to reason as it develops, like a reasoning ability to understand what it's actually being asked and stuff.
[00:19:41] It's not sentient itself. It's not, AGI doesn't need to be, but these language models are getting so good, so fast that I, I mean, to look out a year, two years, three years from now. It's really hard to fathom how advanced things are going to get. And so when Kafu is talking about this stuff, like he talks about some forward looking things, but he also anchors it back to like very practical things that you can do to become an AI driven company.
[00:20:07] And so I think it's just, it's really important that people understand. Like, you may be new to ai. Most people are, And he even talks about like single digits, like the adoption curve of AI within corporations, which I agree with. Like, I, I do not think that we are very advanced in the adoption of AI at all.
[00:20:24] But I, I think it's coming. Way faster than most business leaders think it is to the point where you're not going to have an option. Like AI is just going to be infused into everything in the next few years, and most companies just are not ready for it. So I I, I know you high like a few of the things that he identified as like what is an AI.
[00:20:42] What do you have to do to become an AI driven company? And I think those are really helpful.
[00:20:46] Mike Kaput: I think that that's a really interesting point, is that him leading into this interview, talking about this next golden age of investment in AI development, but then he readily admits that, yeah, you said it's a single digit percentage of people that are fully using AI to the extent they should.
[00:21:03] Then you have to look at that and say, Oh my gosh, like oh, nine out of 10, if not more, of the businesses and leaders and marketers listening are not remotely close to using AI as much as they should, if at all. So then he brings it down to Earth and says, Okay, so what does it actually mean to be an AI driven company?
[00:21:24] What do you do? And it's very, Intelligent and unsexy steps. It says, he says, First you have to become data driven. Without data, you simply cannot do anything with artificial intelligence, whether that's internal data or you're using a language model, for instance, that relies on billions and billions of parameters.
[00:21:44] So, His first out of the gate advice is you need to invest in digitizing every single thing you can digitize at your company or find out how it can be. I love, he really brings up the point that data and data storage is no longer a cost center. You cannot think of it that way. It is in fact your most valuable asset.
[00:22:05] A huge mindset shift is required before you even. Using artificial intelligence. And then from there, honestly, I think his advice really echoes a lot of the things we tell audiences. Students of our AI academy consulting opportunities is, look for this low hanging fruit. Look for things that drive cost savings to begin with.
[00:22:29] I mean, you don't have to necessarily from day one. Transform every single thing you do to become some futuristic AI organization. Though the long view is important, look for things that just take people a long time to do and look for things that machines can do in a faster fashion. I mean, and that can include making decisions.
[00:22:50] I mean we always talk, we with so many business leaders about, you know, find. A practical use case that take, that's repetitive, that's data driven, and that might be making a prediction. You probably have a really good initial use case for artificial
[00:23:06] Paul Roetzer: intelligence. Yeah, I think that's the key. And you know, I always love when we read this stuff from these people that we have such respect for and they're, they're echoing what we're advising people.
[00:23:16] And Andrew, same way like, and so much of our thinking is formed through the years of how. These leaders and researchers and entrepreneurs talked about ai and I think that's the key is like that intelligent automation, the cost reduction, the efficiency drivers from every report we've read in the last like five, six years.
[00:23:34] Mm-hmm. , that's the starting point. Like you can do the big thing. You say, Oh, we're going to become an AI driven company, AI first company. That's fine. You, you probably should. It's going to take years, and it is like your data, your business intelligence, your decision making, your staffing, these are all big things and they're going to take a while to play out.
[00:23:50] But you can start with intelligent automation of a repetitive task now, and it might be something that saves you 20 hours a month or a hundred hours a month, or a thousand hours a month. Depending on the business and you know how often you do it. But that's a very tangible starting point. And the other thing he talks about is like these, this idea of for the leadership team, like vivid working examples mm-hmm.
[00:24:10] five years ago, if a, if a CEO came to us and said, Well, show me ai. Like I don't, I don't get it. It would be hard to show, like you would've had to have gone and find like something within IBM Watson where you can analyze a data set or something that wasn't super tangible. Right now it's like, let me pull up DALL-E.
[00:24:27] Let me pull up character ai. Let me pull up this like the AI test kitchen from Google. I just got my beta, I didn't even tell you this. I got my access this morning to the AI test kitchen from Google. It's. I can, I can show you a dozen examples right now, and most of them either cost you nothing or less than 50 bucks a month to use.
[00:24:44] So now all of a sudden AI become super tangible. I think like my, my working theory of why we're hitting this tipping point with AI is because it's gotten so good, so fast. That the cost of creating AI technology has dramatically lowered to where you can create this very practical use case technology for very affordable rates or free that people can actually see and experience the ai.
[00:25:08] And that is, we did not have that three to five years ago and now it's so readily available and in everything from your iPhone to, you know, Adobe, to all the tools you're using as consumers, it's just everywhere. So it's much easier to explain AI to people and show them it at.
[00:25:24] Mike Kaput: Yeah, it's, I've noticed, and we've definitely capitalized on this, it's much easier in front of say, an audience or you know, a workshop to get people nodding along and saying, Oh, this is actually important technology when you show them.
[00:25:40] Oh, this text prompt I just typed into DALL-E 2 invalidates the need for me to be an artist to create the image, the beautiful image that the machine just created based on my idea. I mean, just seeing this stuff regardless if they don't, if they glaze over at the technical terms or it seems too far away.
[00:25:58] I think the last 12 months has been a game changer with
[00:26:01] Paul Roetzer: some of this. And that's like for us, like we do consulting and we run workshops on this stuff. So like our piloting AI workshop, that's the basic premises. You get the different people in the room and we talk about what are the things you spend a bunch of time doing that are repetitive, that are data driven.
[00:26:15] Mm-hmm. . And you can actually create this very tangible list of like, well here's 25 things that we spend more than 25 hours a month each on. And some of these are purely repetitive. There's no reason a machine shouldn't be doing this, and that's how you start to prioritize use cases. Mm-hmm. . So I think again, it's, it's just becoming easier to sit down and have those conversations very quickly.
[00:26:36] I mean, for us, like we'll sit down for like three, four hours with a group and you can walk away with dozens of ideas and then it's just like, well, which one do we start with? So again, I think three, four years ago, ai. Was a big thing that probably cost a bunch more money and needed a bunch of structuring of the data and there weren't a ton of logical use cases.
[00:26:53] You could start with tomorrow and anytime you or I do a talk or you want to work workshop, it's like, No, we can have you using AI tonight. Like it's, this isn't that hard. Like here's some starting points if you just want to go experience it. And we just, we couldn't do that before. Yeah.
[00:27:07] Mike Kaput: Absolutely. And I think that's a good transition into our third topic and final topic of the podcast today, which really shows you what is possible here.
[00:27:17] And so meta formerly, Facebook recently released an AI system that will take text prompts and turned them into full machine generated video clips. So where something like DALL-E 2, you type in, draw me a picture of a horse drinking water. Make a video will take the same prompt and actually create a realistic looking video or a video in whatever style you choose that is moving.
[00:27:44] It's an image, it's a video clip of a horse drinking water. So this isn't publicly available yet, like DALL-E 2 is that they do state that their intention is to release it, I believe, commercially, but it shows how quickly we've gone from overall good language models that can now. Understand human language and understand the connections between objects to breathtaking static art to incredible video generated simply from a text prompt.
[00:28:17] And I mean, this is, it's just mind blowing to see how I'd highly recommend everyone look up, you know, metas make a video and just even take a look at some of the sample clips they have here. And it's just jaw dropping that I don't think we would've even been. Considering this to be possible a few years ago.
[00:28:37] Paul Roetzer: Yeah, and I, I think like, so at a high level, like one, you could look at the, make a video and be like, Eh, it's okay. Like video's not great. Like that's not cinema quality video. That's not the point. . Right, right. Again, assume. Whatever you are seeing is probably 12 months behind where they are. So, you know, what's coming is probably the key here.
[00:28:57] The, the thing to know is this is sort of like you hear about prompt engineering is a term. Mm-hmm. , that was we, we heard at our conference this year a lot. It's this idea of giving the machine a prompt. So in this case it's text to video, so you're prompting it with text of what you want it to create and it's generating a video.
[00:29:14] We the same thing with DALL-E 2 and Mid Journey and Stable Diffusion and all these image generation technologies where you're giving it a prompt. So human language, a text. To say, create this image that looks like this, that is in this environment, that's in this style, and it creates it. What's happening is we're moving towards this text to everything environment where we're going to be able to prompt the machine to create anything we can imagine.
[00:29:42] Right now it's images and videos, but in the future it's going to move to all kinds of things. And as the language models that are the underlying thing, enabling a lot of this get better and better. The machine starts to really understand what it is you're asking. So the way we were talking about as a team yesterday, we were in a team meeting, brought this up.
[00:30:00] Is think about like Google searching. Like for those of us old enough, Used to have to do boon searches. Like if I wanted to find something at the library or the early days of the internet, I have to like, and, or like I had to really, really good at trying to tell the machine exactly what I wanted and the results kind of were, you know, sucked a lot of times cuz it just wasn't understanding it.
[00:30:21] Then Google comes along and build a smarter search engine. Now think about like how good that search engine is when you ask it something or put in a few words mm-hmm. and how it can predict. What it is you actually want, because it's understanding the context of what you're asking. It knows where you are, where you've been, what you've searched before, what you've bought.
[00:30:41] Like it can factor all of these things in and say, Oh, Mike is actually looking for this, and it can confidently give you like the number one result. That's that thing. Now apply that to image generation, language generation, business plan, generation, presentation, anything you can imagine creating. And say, Well, what if the machine could create those things?
[00:31:01] And all I had to do was give it a prompt with my voice or with text, and it understands me like the search engine does. Like it's so good at understanding what I'm asking. I don't even have to be that good at prompting it or like tuning that prompt, adapting the prompt to like be involved. And that's the, the key takeaway for me here is like, yeah, it's mind blowing what they're doing.
[00:31:22] Runway ML is another one to look at. They actually announced their text to video before me did, but they didn't come out with it yet, but they're teasing it real heavily. There's, we'll put all this in the show notes, but there's some really cool stuff with stable diffusion. The guy who created the stable diffusion model.
[00:31:36] And runway ml. But the key is you're going to hear a lot more about this and you're going to start seeing it infused into the language generation. So like there's a few I've seen recently of these G P T three language generation tools, like the right AI writing tools mm-hmm. , where the image generation is connected to the, like, go Charlie, didn't they do it?
[00:31:56] I don't think their GPT three. Hyper write is another one. So as you're writing or the machine is writing for you, It'll actually automatically build images to go with the texture writing, not stock photos. It'll build images. It's basically taking the paragraph or the sentence from the blog post as a prompt for what image to build.
[00:32:16] And all I have to do is generate image. And so now imagine like you're writing a blog post, paragraph by paragraph, just telling the machine, Write the next paragraph. Now write me a paragraph on this, a paragraph on that. And as that's happening, an image is being generated to go with. We're there, like, I'm not talking about three years now, like that's what's happening right now within some of these tech companies.
[00:32:35] So I, I just think like it's really important for people to realize we're not talking three to five years out. We are talking about stuff that is here right now, and it is, it is improving so quickly. And so it's not like this whole idea of like text, anything may sound crazy to people, but I'm telling you, it's not , right?
[00:32:56] The people who are building this stuff are very, very confident that they are making major leaps and it's going to change everything for marketing, for sales, for customer service, for business. It's, it's a wild time. And again, Mike and I look at this stuff all. And they're like, I'm getting to the point where there's like two or three times a week where I'm like, What?
[00:33:15] Yeah. That used to happen every two or three months. Yeah. And now it's twice a week. Yeah. Where I'm just like, mind blown. Gotta send it to Mike. Like, Did you see this ?
[00:33:25] Mike Kaput: Yeah. I mean, we talk, we'll probably talk about it on a future episode, but I, two hours ago, sent the team an AI tool that is now. Creating podcasts from synthetic voices from scratch.
[00:33:37] So I got to listen to a completely machine generated, completely a manufactured interview between Joe Rogan and Steve Jobs. I mean, it's jaw dropping.
[00:33:48] Paul Roetzer: It's, And same this morning. So again, like twice of our wow moments in one day, I had characters.ai, which was two top engineers from Google left who were working on their lamb Lambda language model, which we talked about earlier.
[00:34:02] They went and built something that lets you have a chat conversation with any celebrity. So like if I wanted to have a conversation with, I dunno, Tom Cruise, I was think about Top Gun earlier. So I, I talk to Tom Cruise, I can interact chat bot wise with Tom Cruise, but it's actually just a language model that's, Trained on some data set to talk to me exactly how Tom Cruise would.
[00:34:23] I mean, it's, and, and you can just go in and create any character and it's completely synthetic and, but it's very, very convincing. And then if you think about the, again, the, the impact for marketers. Think about your chat bot right now and how bad it probably is at understanding the context of questions, answering in a, you know, very real way engaging people in a unique voice.
[00:34:44] Like what? I'm saying what if, like, I can guarantee this is coming very fast. You can actually, like, let's say I want a license to I don't know, James Earl Jones, like we just, he just did the the Vader one. Like I wanted that to be my brand voice on my, my site. Like I wanted my chat bot to interact with somebody, like in his tone and style, and he could, you could license that from him.
[00:35:05] Like he could, you know, give that away and now I can actually like go. Interact with it. Like this is the kind of stuff that's going to happen. Like anything is possible. We can, we can create synthetic versions of anyone with today's tech. And as long as you license those rights away, like, or you create your own, Like you think about all this money in the last like, couple decades that went into like sonic branding.
[00:35:27] Is that the, I think I'm using the right term around like you wanted, like a noise that people think about or a tone that people think about when they hear, you know, your brand. What about. Personalities within the conversational agents. Like, are you going to create that? Like, are you going to from the ground up, build one that interacts with people in, in ways that you never imagine possible?
[00:35:45] Like you, you could totally do it. Like there's no obstacle to the tech enabling that right now. Or in the, in your ceo, like, should you replicate the CEO and, and train it on a language model on your CEO's tone and voice. And so every interaction on your site is actually with a synthetic version of your ceo.
[00:36:02] Zero technical obstacles to doing that today. Right, right. It's just, whatever you can imagine is basically where we're at. Marketing. It's so crazy.
[00:36:10] Mike Kaput: And any, any one of these things which could happen tomorrow has the potential to just upend business as usual, which is why it's so important. No matter what area of business you're working in, to start getting a handle on how this technology affects whatever your particular use cases or role looks like in a given day.
[00:36:32] Paul Roetzer: And so our, our plan moving forward, again, like Mike and I talk about this stuff all the time, but again, it used to be like, yeah, once a month you catch up, whatever. Like here's the two things that happened last month. And it literally is just like daily where we're just seeing stuff and I'm just, She got a running list of things we should probably talk about.
[00:36:49] Yeah. And so what the plan now is like just each week Mike and I are going to kind of get together and try and pick. Things that caught our attention in a given week that we think are really interesting for people to hear. So, you know, if, if this kind of stuff's interesting to you, make sure to, you know, subscribe and stay in tune with what's going on.
[00:37:05] We publish summaries on the blog post each week, so you can check out that and we'll have show notes in there and timestamps. And then Mike's actually going to start taking each of the topics and spining off a, a post to kind of going a little deep on each of the topics. So, yeah, I mean, if this kind of stuff is interesting to you, We hope you you know, join us on this journey cuz it's, it's getting wild.
[00:37:25] really is anything else you want to cover today,
[00:37:27] Mike Kaput: Mike? No, I think that's it. I mean, I'm sure three or four other things have happened while we've been recording this, so we'll, we'll have to get caught up on that.
[00:37:37] Paul Roetzer: Yeah. All right. So thank you for being with us. Like I said, be sure to subscribe. It's on any of your favorite podcast apps and networks.
[00:37:44] And also head over to marketing ai institute.com. We've got a weekly newsletter. We've got free webinars, online courses, AI courses and obviously the marketing a book. If you haven't grabbed that, grab a copy of that and we'd love to hear your feedback once you've had a chance to read it. But other than that, until next time, stay curious and explore ai.
[00:38:01] 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.
[00:38:23] Until next time, stay curious and explore ai.