If you’re in content marketing, marketing, artificial intelligence, or an active LinkedIn member, there’s no doubt OpenAI’s ChatGPT has made its way into a conversation or your social feed.
This week’s episode takes a deep dive into ChatGPT, but not before touching on a few other topics in the AI world - Runway’s funding, the hype and controversy around Lensa’s app, Meta’s CICERO unveiling, character.ai’s launch, and more.
It’s been a busy few weeks since the last Marketing AI Show:
Nov. 22, 2022: Meta announces CICERO, “a breakthrough toward building AI that has mastered these skills. We’ve built an agent – CICERO – that is the first AI to achieve human-level performance in the popular strategy game Diplomacy.”
Nov. 30, 2022: OpenAI releases ChatGPT, “which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests. ChatGPT is a sibling model to InstructGPT, which is trained to follow an instruction in a prompt and provide a detailed response.”
Dec. 1, 2022: DeepMind announces DeepNash, “an AI agent that learned the game from scratch to a human expert level by playing against itself. DeepNash uses a novel approach, based on game theory and model-free deep reinforcement learning.”
Dec. 5, 2022: Runway announces a $50M Series C, and a new AI training tool, allowing the company to, “double down on building the next generation of creative tools.”
December 2022: The Lensa AI app focuses on their Magic Avatars capabilities, taking over social media, however, controversy ensues.
Dec. 5, 2022: The Character.ai team has launched their product after spending 13 months, “building scalable and extremely efficient infrastructure to train and serve our first large language models, and launching our first public beta that enables anyone to create and interact with conversational AI agents (“Characters”).”
The bulk of the podcast is about ChatGPT. On Dec. 2, 2022, Paul Roetzer published the article, “A Marketer’s First Experience With ChatGPT From OpenAI, then shared his findings and started a conversation on LinkedIn:
<iframe src="https://www.linkedin.com/embed/feed/update/urn:li:share:7004448641100775424" height="1342" width="504" frameborder="0" allowfullscreen="" title="Embedded post"></iframe>
Paul and Mike discuss the timeline, the buzz, what this means for the future of marketing and business, and more. Listen to the podcast below or in your favorite podcast player.
00:05:12 Meta announces Cicero
00:09:28 OpenAI releases ChatGPT
00:12:48 Mastering Stratego with DeepNash
00:15:54 Runway announces $50M Series C
00:17:24 Lensa’s AI app
00:19:12 Character.ai officially launches
00:23:34 Paul and Mike spend the remainder of the podcast talking more about ChatGPT
Links referenced in the show
- Meta announces CICERO
- OpenAI releases ChatGPT
- Mastering Stratego, the classic game of imperfect information
- A Marketer’s First Experience with ChatGPT from OpenAI
- Runway Announces $50M Series C
- What You Need To Know about Lensa, the AI Photo App
- Introducing Character
- The Future of Business Is AI, or Obsolete
- AI training costs are declining 60% YoY
- Users will soon build the tools, software, and code they need using AI
- AI is the next 100X productivity boost
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: The real key here to me, the barriers to entry, think about access to the tech, anybody could try it now. Technical skills required to use it, no code. You don't need to know anything to use it. Computer power, compute power, energy costs, which are basically going to zero in the next decade, all of that is falling. The barrier to entry is almost gone to use this technology. So that to me is what matters here and why it's so significant and the fact that it's just the beginning.
[00:00:30] 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.
[00:00:50] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host. Welcome to episode 25 of the Marketing AI Show. I'm your host, Paul Roetzer, along with my co-host Mike Kaput. What's up Mike? How's it going? Good. Back, after, I think we had a three week break. Unintentional. Yeah. We were sort of planning an impromptu Thanksgiving break week and then it sort of rolled into Mike and I working on a major release that I'll talk about in a moment.
[00:01:30] Paul Roetzer: And it just like, alright man, we have to punt this. And then amazingly thank you to the people who reached out and actually asked us, where is the podcast? I think sometimes as a podcaster, you just create stuff, and you don't really know if people are relying on it or listening to it at regular intervals every week at their gym as one person reached out to us.
[00:01:53] Paul Roetzer: So I apologize for our regular listeners I wasn't sure we had that were letting us know, and wanted to make sure everything was okay and that our podcast did not release on time. So, we are back. We are back to our weekly show and hopefully, outside of maybe another holiday week here or there, we will be hitting the regular Wednesday release schedule.
[00:02:15] Paul Roetzer: So, today's episode is brought to you by the new Piloting AI for Marketers online course series, which is the thing that Mike and I have been madly working to finalize, to release, in the very near future here. So the Piloting AI for Marketers series is brought to us by Marketing AI Institute's Online Academy.
[00:02:36] Paul Roetzer: It's designed as a step by step learning journey for marketers to guide them through understanding and adopting AI to advance their companies and careers. So this is something we've been working on for a really long time and thinking about, but we finally put the plans in place a couple months ago to produce this series.
[00:02:54] Paul Roetzer: So it's 17 on-demand courses that Mike and I are creating, original content we're creating for this, sort of inspired by the book, but really going far beyond what's in between the covers of the book. So much has happened since the book came out in June, and this series gave us a chance to really dive into a lot of the new things and the implications it has on marketers and on business leaders.
[00:03:19] Paul Roetzer: So there's dozens of new use cases and technologies offered within the course, collection of templates and frameworks that'll help people get started. So, like I said, we just put everything we've learned since we started researching AI in 2011 and then hundreds of hours of planning and production in recent months to bring this thing to life.
[00:03:37] Paul Roetzer: And so we hope it really helps move the industry forward. You know, as we were creating it ChatGPT came out and new advancements in image generation, video generation, I feel like the world has changed so much, that hopefully the course takes on even greater importance now than I I thought it was going to have when we first started building it.
[00:03:57] Paul Roetzer: So you can learn more at pilotingai.com. Again, it's part of the AI Academy for Marketers, which is from Marketing AI Institute. So, yeah, check that out. And then before I turn it over, so if you're new to this format, our weekly format, Mike basically curates what we think are the three most important topics in AI for a given week.
[00:04:19] Paul Roetzer: And then he and I talk about it. This week was particularly challenging because each thing that has happened seemed worthy of its own episode. And so today we're going to focus on ChatGPT and generative AI and language in particular, or in general. But before we do that, I wanted to just try and set the stage for how insane the last few weeks have been since Mike and I were on podcast leave producing the Piloting AI series.
[00:04:52] Paul Roetzer: So we're not going to stop and do our usual conversation around each of these events, but I just want to give you a sense of how much has changed since right before we went off air and November 22nd we'll start with. So we're just going to kind of read through some of this and try and process how crazy this is.
[00:05:12] Paul Roetzer: So, November 22nd, meta announces CICERO. This is from Meta. I'm just going to read this. So games have long been a proven ground for new AI advancements from Deep Blue's victory over Chess Grand Master Gary Kasparov to AlphaGo's Mastery of Go. The Pluribus out-bluffing the best humans in poker. But truly useful, versatile agents will need to go beyond just moving pieces on a board.
[00:05:38] Paul Roetzer: Can we build more effective and flexible agents that can use language to negotiate, persuade, and work with people to achieve strategic goals similar to the way humans do? Now as I'm reading this stuff, think about the dots to connect in business. So, I'll re-read that line, and we build more effective and flexible agents that can use language to negotiate, persuade, and work with people to achieve strategic goals.
[00:06:07] Paul Roetzer: So the rest of this I'm going to read here is not marketing and business specific, but keep that line in mind as I'm reading this. Think about the applications and business strategy, sales, negotiations, hr, finance, all of these things if these kinds of AI agents can exist. Okay. Today we're announcing a breakthrough toward building AI that has mastered these skills.
[00:06:31] Paul Roetzer: We've built an agent, CICERO, that is the first AI to achieve human level performance in the popular strategy game Diplomacy. Have you ever played Diplomacy, Mike? I have. Yeah. Have you really? That doesn't surprise me. I feel like you would geek out over Diplomacy. We should play that sometime.
[00:06:48] Paul Roetzer: It's a long game. Is it? Yeah. We got time. We got the holidays coming up. All right, so then the last part here, Diplomacy has been viewed for decades as a near impossible, grand challenge in AI because it requires players to master the art of understanding other people's motivations and perspectives, make complex plans and adjust strategies, and then use natural language to reach agreements with other people, convince them to form partnerships and alliances and more. CICERO is so effective at using natural language generation to negotiate with people in Diplomacy that they often favored working with CICERO over other human participants. Okay. I know I said we weren't going to stop and talk about each of these, but damn.
[00:07:30] Paul Roetzer: Okay. So just like 30 seconds, like what was your reaction to having someone having played Diplomacy of this sort of breakthrough in AI?
[00:07:38] Mike Kaput: Honestly, when I read this and knowing what I do, though, it's been a long, long while since I played Diplomacy, but knowing what I know of the game and how much it relies exactly on that type of informal negotiation and relationships and arrangements between players, I am pretty stunned.
[00:07:56] Mike Kaput: This feels like one version of passing a Turing Test in some ways. So it is wild that an AI assistant can actually do this. And especially given, and I'm sure we'll talk about this, how early we still are that we're able. This feels just as big to me in some ways for the natural language space and motivational space as the game of Go was to the overall deep learning advancements, I would say.
[00:08:27] Paul Roetzer: It felt that way when I read it. And now it's so funny, like again, so much has changed. I really forgot about this one. So right before this announcement on CICERO, Meta also unveiled Galactica, a large language model that they let people test, and it got shut down after like three days because people were abusing it and showing all the ways that it could be misused.
[00:08:52] Paul Roetzer: And so Meta had to make the decision to shut down Galactica. That happened in November as well. Just major, major stuff. And so it almost seems like the CICERO thing... So I follow a lot of AI influencers on Twitter in particular is where I get most of my news and curate most of my information.
[00:09:14] Paul Roetzer: This was like a blip, like it was the hot thing for a couple days and it was being talked about at the very high AI researcher levels, but it didn't make mainstream. And then I think whatever attention it should have gotten, we ran into November 30th when OpenAI releases ChatGPT. Rather unceremoniously, by the way, like they did not make a big push about this.
[00:09:39] Paul Roetzer: I don't even know that there was a blog post about it, or there may have been a blog post on this one, but not the DaVinci model. Like it was very silly, just like, oh, here it is, like just on Twitter, like you can check it out.
[00:09:51] Mike Kaput: Yeah, in doing some of the research for the podcast brief, I don't even know if I even saw at the time the blog post or the webpage.
[00:09:59] Mike Kaput: It's not extensive . No. It's got like three paragraphs of content about what it is and what it does. And I think I just heard about it because the internet blew up shortly after it was released. Yep. And so the crazy thing, and again, we'll get more into ChatGPT in in the rest of this episode. This is just the summary upfront of other major things.
[00:10:21] Mike Kaput: But if I'm not mistaken, I went back in my timeline, I couldn't find when this one happened, but I think it was either the day or two before ChatGPT, OpenAI announced that there was an updated version of GPT-3 called Da Vinci 003 model. I don't know if that's how they actually say it. I can't find a formal announcement of that.
[00:10:42] Mike Kaput: But you can go in to OpenAI's playground and test the new version of GPT-3 that's in there, this kind of advanced model. I used it. So I went into OpenAI's playground and I created a 440 word blog post. And all I said was: write a blog post on how marketers can get started with AI. And it wrote it in about 10 seconds, cost me a penny.
[00:11:02] Mike Kaput: So they charge by tokens. So you can get a thousand tokens for 2 cents, I think is the model maybe? The equivalent is a thousand tokens will generate about 750 words on the DaVinci model. I was paying the 2 cents. So the DaVinci is the most expensive, most powerful version of GPT-3. So for about a penny, I wrote a blog post that was 440 words long.
[00:11:25] Mike Kaput: And I can tell you, Mike and I worked together at my agency for 10 years. I owned an agency for 16 years. We were a content agency. So the content this created in draft form, not that I would've published it or passed it off to a client, but the draft form it created was better than I would've expected from a marketing writer with three to five years experience who already understood AI. Like if I gave this to a writer three, five years in, never written about AI before and said, write me how to get started, it would take 10 to 20 hours because they would have to go research what is AI? I don't understand it. I have to figure it out. So I started thinking about, like, Mike and I used to have clients in like chemical engineering as an example. We're not chemical engineers, we're writers.
[00:12:07] Mike Kaput: So we would have to like dive in and learn chemical engineering to write a single 440 word post about the chemical engineering process. And so just like it illuminated to me how transformative this could be. And that was before ChatGPT. That was like a day or two before I was playing around with the Da Vinci model.
[00:12:27] Mike Kaput: Then they launched ChatGPT and its like, oh my gosh. So again, we're going to spend the rest of the episode on that one, so I won't linger on this one too long. Okay, so that's November 30th. So, okay, the next day, December 1st, comes DeepMind into the ballgame.
[00:12:45] Mike Kaput: Don't forget about us. Kind of hold my beer moment. Okay. So, Mastering Stratego, have you ever played that one? Stratego. I don't even know if I'm saying it
[00:12:53] Mike Kaput: right. Maybe as a kid I don't recall that one as much. But yeah, I've heard of it. It's a popular game. Okay,
[00:13:00] Paul Roetzer: so classic game of imperfect information is the headline.
[00:13:03] Paul Roetzer: Okay, so here's what DeepMind said. Again, they're not talking about marketing and business, but put yourself in that mindset of what this means to marketing and business. Game- playing AI systems have advanced to a new frontier. Stratego. I'll just go with that one. Stratego, the classic board game that's more complex than chess and Go, and craftier than poker, this sounds a lot like CICERO, has now been mastered. Published in science, we present DeepNash, an AI agent that learned the game from scratch to a human expert level by playing against itself. DeepNash uses a novel approach based on game theory and model-free deep reinforcement learning. Its play style converges to a Nash equilibrium.
[00:13:46] Paul Roetzer: I'd have to ask OpenAI's ChatGPT what a Nash equilibrium is. Which means its play is very hard for an opponent to exploit. So hard, in fact, that DeepNash has reached an all time top three ranking among human experts on the world's biggest online Stratego platform, Gravon.
[00:14:07] Paul Roetzer: The value of mastering Stratego goes beyond gaming. I bold faced this, so it must be important. In pursuit of our mission of solving intelligence to advance science and benefit humanity, we need to build advanced AI systems that can operate in complex real world situations with limited information of other agents and people.
[00:14:27] Paul Roetzer: Our paper shows how DeepNash can be applied in situations of uncertainty and successfully balance outcomes to help solve complex problems. My initial reaction when I read the "building advanced AI systems that can operate in complex real world situations with limited information" was the whole reason I started pursuing artificial intelligence back in 2011 was I had this vision that someday there would exist the marketing intelligence engine that intelligently built marketing strategy, allocated budgets, did predictions around where you should spend your money and resources. And I asked myself, are we there? Like, have we actually arrived at the point where AI will enable the building of an intelligence engine?
[00:15:10] Paul Roetzer: I have no idea the answer to that, but it sure sounds like when you read the CICERO thing and the DeepMind thing that maybe that ideal that human strategy will be inseparable from AI is probably in the very near future and probably already existing within the major tech companies, so...
[00:15:32] Mike Kaput: wow. Yeah. The audience should thank us because we took a three week break and we changed the world in terms of the AI developments that happened.
[00:15:45] Mike Kaput: We go offline, everything happens, right? Okay. I'll run through these last three. Again, this is all in the last two weeks.
[00:15:53] Mike Kaput: Okay? So then yesterday, December 5th, Runway announces $50 million series C and a new AI training tool. So RunwayML, runwayml.com, we'll put the link in the notes. So they raised 35 million series B in December of 21, so a year later coming with 50 million series C to, as they would say, "build the next creative suite, the generative suite."
[00:16:19] Mike Kaput: "This funding will allow us to double down on building the next generation of creative tools. Our existing research and the open source release of Stable Diffusion, which is critical in the image generation movement, has forever transformed the generative AI landscape, and we are thrilled to continue building state of the art multimodal AI systems," multimodal meaning language, you know, text, voice, images, anything. Like video multimodal components.
[00:16:46] Mike Kaput: Runway also, again, the craziest thing, they just drop, "oh, by the way, here's an AI training capability while we're announcing the funding". So separate tweet, here's AI training. So now if you go in, and they're not a sponsor, I pay $28 a month for Runway to play around with it. I actually feature it in our Piloting AI for Marketers series.
[00:17:05] Mike Kaput: It's amazing technology. So they have an AI magic tool suite within theirs. That again, is part of the $28 a month package. And they have a custom generator and a portrait generator where you can upload portraits or objects and then create custom images using your images as the training set. And, along those lines, you have probably seen Lensa AI app, I think is how you say that one.
[00:17:29] Mike Kaput: I know it blew up in my social feeds in the last few. Where people are uploading photos of themselves and then Lensa trains it, and then they can create all these variations of themselves. So our own Cathy McPhillips has a blog post on that that we'll link to where she tested it. But it comes from, I think Prisma AI is the company that makes Lensa, an all in one image editing app with facial retouching capabilities.
[00:17:52] Mike Kaput: It sort of took social media over, as I said. Since 2018, Lensa, owned by Prisma Labs, has offered photo editing technologies, background blurring, filtering, face touching, borders, and more. They recently made the news with these AI powered magic avatars. Now kind of comes with a disclaimer: anytime you are uploading your information, images, videos, whatever it is, you are likely giving the rights to the creator to have access to your data and use it to train stuff. So this is a massive play to train an AI on people's faces, in essence, and whatever else they're using it for. So Cathy goes into that in her post where she kind of highlights some of the bias inherent.
[00:18:32] Mike Kaput: I've seen a lot of women online posting how it's sort of sexualizing women. So they're uploading normal photos, headshots, and all of a sudden, it's a lot of inappropriate things being created by the AI. And again, it's if you understand how this stuff works and the bias that that is inherent within it, It's kind of to be expected.
[00:18:52] Mike Kaput: This is sort of how these things play out. So that might be something you want to play around with. It sounds like you can do the same thing now in RunwayML. It seems like the same two technologies, which you also notice people keep releasing the same kind of stuff around the same time. It's almost like one person's release, someone else says, "oh sh*t, we have to get ours outta the labs and get it live."
[00:19:12] Mike Kaput: Which sort of seems to be the case with the final one I'm going to touch on here, which is character.ai, which seems to have a lot of things that ChatGPT is enabling and it was almost like, "hey, let's release." Maybe they were timing the announcement already. I don't know. But another company that we have talked about before, character.ai, they tweeted on December 5th that they had officially launched.
[00:19:36] Mike Kaput: Now I thought they'd already launched. I wasn't aware that they hadn't launched. That's why I thought it was odd when I saw that post come up, like we've officially launched. But again, in their case, imagine everything it could do for you from being your own personal teacher. It's a, it's a chatbot, basically, a very intelligent chatbot. "Being your own personal teacher assistant, or even a friend."
[00:19:54] Mike Kaput: Two months after watching in September, their beta generated over a billion words per day. Get a glimpse of the future character.ai, some of their stuff, and then from the blog post announcing it. So I just want to give you the context of why character.ai matters. Because you could look at and think, this is kind of quirky.
[00:20:11] Mike Kaput: This site doesn't look super impressive when you go there. But listen to who's behind it. So this is quoting from a founder. "As an early engineer at Google, I've built AI systems used by billions of people. From Google's spell checker, "What did you mean?" To the first targeting system for AdSense, I also co-invented Transformers.
[00:20:29] Mike Kaput: Transformers are the fundamental basis for GPT. That's what enabled GPT to be built. So transformers are a massive, innovation within our lifetime. Scaled them to supercomputers for the first time and pioneered large scale pre-training, all of which are foundational to recent AI progress. That is true.
[00:20:49] Mike Kaput: He's not overstating that. "Why I'm proud of launching the large language model's revolution," That's quite a claim to be able to drop in a blog post, isn't it? "I launched the large language model revolution. I'm even more excited about what comes next. My co-founder and I witnessed the technologies and possibilities firsthand as we invented and led the development of MEENA and LaMDA," LaMDA's, Google's major large language model.
[00:21:12] Mike Kaput: "With LaMDA, we explored whether large language models could be engaged in open ended conversation and quickly recognized their potential to help people and putting users at the center of the interaction, conversational aI can handle dynamic tasks that require the exchange of ideas, clarification, and feedback in ways that cater to the user's individual needs and interests." Goes on to say "we started character to lead the AI revolution and bring joy and value to billions of people at each step along the way.
[00:21:39] Mike Kaput: We were joined by a founding team of AI leaders from Google Brain and Meta AI who are experts in deep learning, large language models and dialogue." Long story short, legit. Like, pay attention to what this company is doing and what these founders are saying. It is often the way that we stay kind of at the forefront of what's going on.
[00:21:59] Mike Kaput: I watch very, very closely what the founders of the leading AI companies are doing, and the beauty is many of them are active online on Twitter and share their thoughts on things. And so you can learn a lot about AI by just staying close to what these people are talking about. And I'm telling you right now, they didn't leave Google without a really good reason and a belief that they can build something significant.
[00:22:23] Mike Kaput: So, With that, that sort of leads us into chat and conversation. And our main topic for the day of ChatGPT. Mike, I'm going to turn it over to you and get us back on track for the rest of this.
[00:22:38] Mike Kaput: No, I think that that's awesome. And there's so many things that have happened and I think they are, as we're going to see, as we pull these threads together, that they're all related and they have some pretty serious implications for marketers, business people, and honestly, probably society at large.
[00:22:56] Paul Roetzer: I mean, just like quick example with, with the character one and the ChatGPT, like my immediate question is, is every conversational platform previously obsoleted all of a sudden? Like if I have a conversational agent on my site, does it matter? Like I'm probably on the phone with my conversational AI platform company saying, What is going on?
[00:23:16] Paul Roetzer: Like how does this affect what you do? I don't know if they have answers for that, but that was my immediate thing is like, this seems transformative. This seems like it may obsolete some existing tech. So we're trying to figure it out just as you are. In some cases we're just asking the questions.
[00:23:33] Paul Roetzer: So really the thing that kind of kicked off all of our speculation, kind of interest in this, is November 30th, which will probably go down in history as a big day because OpenAI launches ChatGPT, and like we've kind of talked about, it's a chatbot that anyone can converse with by typing in text prompts.
[00:23:58] Paul Roetzer: Now all you do is type in a question or a command and ChatGPT produces a response. Now that sounds really simple on the surface, but everyone in the AI world, and I think the general public is now starting to understand this, they're positively stunned by the result. And the reason for this is, you know, Kevin Roose, a tech reporter at the New York Times put it really well.
[00:24:23] Paul Roetzer: He said, "ChatGPT is quite simply the best artificial intelligence chatbot ever released to the general public and what we mean by that is that it's actually displaying already, it's been literally five, six days. It's displaying a breathtaking range of capabilities. So we've kind of been geeking out back and forth about how Twitter users are posting experiments using it to do things like answer really complex questions like math problems, highly technical science questions, and it's doing a pretty decent job it seems in certain tests of actually answering questions that you and I certainly wouldn't be really able to answer at a very technical level about certain complex subjects.
[00:25:09] Paul Roetzer: People are actually using ChatGPT to say, build me a business plan for a software startup and break down this big project into concrete steps. It's able to do that. ChatGPT has also been used to generate poems, fiction, and what's really interesting, I even was playing around with this this morning, it's even been used to solve coding problems that you typically reserve for a programmer.
[00:25:34] Paul Roetzer: So you could tell it, Hey, generate me code that does abc and it will take a stab at doing that. And some programmers who have programming knowledge, unlike me, appear to be pretty wowed by the output. What's really funny is, Paul, you actually just wrote a recent post where you did a Q&A with ChatGPT, asking a number of questions, you know about itself, its impact on marketing, business, society.
[00:26:03] Paul Roetzer: And when we ask ChatGPT, you know, what is a definition of ChatGPT, can you give us a definition of yourself? Here's what it came up with on the first try, no edits. It said ChatGPT is a chatbot that uses GPT-3 generative pre-trained transformer three language model to generate human like responses to user inputs in natural language.
[00:26:28] Paul Roetzer: It can be trained on specific topics or domains to provide more personalized and accurate responses. ChatGPT is designed to improve the natural language processing capabilities of chatbots and enhance the user experience of conversational AI. Now that's a pretty awesome response. I just wanted to kick this off with Paul.
[00:26:50] Paul Roetzer: It's pretty clear that ChatGPT is a big
[00:26:53] Paul Roetzer: deal. How big is it? So my first reaction, and this is, you know, I think I told you, I was Zoom chatting with Mike and I messaged him, I'm like, dude, this is crazy. I think insane was my word I used. And so what happened was I don't remember what day, whatever day the post came out.
[00:27:13] Paul Roetzer: It was like four days ago, like December 2nd, December 1st, whatever, the day after it came out. The night before I'm laying in bed and I thought, yeah, it'd be interesting to interview, ChatGPT and see what it says about itself. And so that was my first experience in the platform is I just went in and started having this conversation about it to try and understand it. I figured I'll just ask it. And so when I did that, what we ended up publishing was just as Mike said, like no edits, it was just straight up, just kept asking those questions and I was asking it things like, What is it? How are marketers going to use it? How are writers going to use it? I started testing it a little bit because one of the keys is it "remembers" previous questions.
[00:27:57] Paul Roetzer: So you'll note in some of my questions where I'll say, I'll refer to ChatGPT as it, or like the technology is it to see if it knew what it was and if it could figure that out within the context of the question. And it did. So after I wrote the blog post, and so I'll read kind of the open to this. I said, my immediate reaction five minutes was the marketing profession, business World Insider, not even close to ready for what is about to happen as a result of rapid advancements in ai.
[00:28:26] Paul Roetzer: Everything we think we know about communications, creativity, and strategy is going to be redefined and reimagined in the months and years ahead. The closest comparison I can find in my mind to understand the magnitude of the. Would be to rewind, rewind to June, 2007 when the iPhone was introduced. What is about to be built in the generative AI space is basically the equivalent of the emergence of the app ecosystem, which Apple says facilitated 643 billion in billings in sales worldwide in 2020, which was the most recent study I could find with generative ai.
[00:28:58] Paul Roetzer: We're talking about trillions of dollars in value and wealth creation over the next decade, and the real key here to me, the barriers to entry. Think about access to the tech. Anybody could try it now. Technical skills required to use it. No code. You don't need to know anything to use it. Computer power, compute power, energy costs, which are basically going to zero in the next decade.
[00:29:18] Paul Roetzer: All of that is falling. The barrier to entry is almost gone to use this technology. So that to me, is What matters here and why it's so significant and the fact that it's just the beginning. Like I've seen some people go in who maybe don't understand AI very well and they'll test it and be like, oh, I got this wrong.
[00:29:39] Paul Roetzer: Like, it's stupid. It doesn't even know this or that. And it's like, you don't, you don't get it. Like you're basically playing with a beta. That's getting smart really fast because Sam Altman tweeted just yesterday, I think there are over a million users in the first four days. Think about the training data that they're getting and some are theorizing that GPT or ChatGPT is just the prelude to GPT four, that what they're doing is using ChatGPT as, as the training data to build and release GPT four, potentially at the, you know, early part of 2023.
[00:30:19] Paul Roetzer: So when you think of that, all we're seeing in as powerful and as insane as it is. It's just the leading edge of where this is going to go. We're not, the innovation isn't stopping, it's probably going to keep accelerating in 2023, and that's the part that has me excited and terrified At the same time, , I think
[00:30:37] Mike Kaput: I have equal parts, excitement and terror, and I think some of, a little bit of both.
[00:30:43] Mike Kaput: You referred to a little bit before, just to kind of bring it to our background. You're absolutely right. I've gotten, I've created sample blog posts using this technology since it came out that are better than what one might hire, you know, a content agency. To create and does that, again, we've talked about writing AI a lot on this, podcast.
[00:31:10] Mike Kaput: I, this just feels different in the sense that I'm seeing stuff where I'm like, I don't know. As a business owner who needed content or answers to some basic levels, say marketing questions, why I would need to ask someone else again, .
[00:31:26] Paul Roetzer: Yeah. And I think this, it kind of gets into, you know, if, if anyone's played with Dolly or an image generation tool, cause it's all this generative aIt's this, give it a prompt, it generates a thing is, is basically what's happening here.
[00:31:38] Paul Roetzer: And what we learned pretty quickly with Dolly was if you were a designer or an illustrator and like you knew different styles and different artists and different genres, and you could infuse that into the prompts, you would give the machine, you could get magic. Where if you're a non-designer like me and you're not, like, you can't really explain things visually very well of what you want, you're going to be impressed, but you're not going to be blown away.
[00:32:05] Paul Roetzer: And I think right now we're at that phase with language generation where you could go in and play around with ChatGPT or GPT three in their playground, or you know, a writing tool like a Jasper, hyper writer, writer, any of these, right? And you could be like, nah, it's okay. Like I asked it to write a blog post for me.
[00:32:21] Paul Roetzer: The blog post wasn't that good. So it's not going to replace my writing. And so you try it once and then you just think you've figured it all out and you, you move on with your life. That's the problem is one, you may just not know how to prompt it very well. You might not understand it's full capabilities.
[00:32:37] Paul Roetzer: And so the one prompt you tried or the three prompts you, prompts, you tried, you didn't get the results you wanted, you just give up because you don't think it's that powerful. But it is. And it comes down to understanding what it is capable of and how it does it. So I saw a crazy example online of, Someone who tweeted, using ChatGPT to create, I think it was like a lease agreement in South Carolina that had specific elements of law in that state that tied to the least contract and it asked it to write it, and it wrote it probably better than most young attorneys.
[00:33:15] Paul Roetzer: And so to your point, it's like, it's not just a writing tool. Mm-hmm. , it is, it is meant to help, figure things out at scale in business, in society, in marketing. Yes. And in writing. Yes. But it's understanding what it's capable of doing, and then how to ask it for that information that enables you to do it.
[00:33:36] Paul Roetzer: Like, I mean, I've seen some insane things around like simple things like. I think it was Doug Kessler, you know, a good friend of ours wrote, wrote a country song with it. I've seen other people that'll write it in specific styles, like they want a poem by, you know, written in the style of a specific poet, or they want a movie script in the style of Quentin Tarantino.
[00:33:55] Paul Roetzer: Mm-hmm. and like to know that it's capable of going beyond, just write me a blog post about this. Writee me a blog post about this in the style of this person written during this time period. And you start to like really push the limits of what it can do. And that's when I think people will realize we are not playing with just simple blog post writing tools.
[00:34:16] Paul Roetzer: If you're thinking about this movement as that you are thinking about it the wrong way. It is not just going to be an assistant to write things faster. Mm-hmm. it is going to change the. We think about creativity, communications, strategy, business, you know, structure. Everything is, it's, it's really hard to comprehend how, how transformative this is going to be.
[00:34:40] Paul Roetzer: And it, it's now like 2023, it's going to happen.
[00:34:44] Mike Kaput: It's interesting to me. The more I think through it, and these are still kinda really first draft thoughts given how quickly, it's been since it came out. I think in some ways the writing piece of it almost misses the point in that it is not only a writing tool, it's an answer engine.
[00:35:02] Mike Kaput: And like you can quibble all day with the. Validity of the responses in terms of accuracy. And I think that's a really valid critique that what it produces can sound really good, but isn't accurate. It hasn't even been trained on recent data on the internet yet. And you figure here, like think of how many careers, jobs and functions are essentially being a human answer engine, right?
[00:35:24] Mike Kaput: Another way to say that is consultant or professional service is in a lot of cases. So that's where it gets really exciting and scary to me, where I'm thinking like, well, okay, if the answers could get even better, which they will. What does that mean for anyone who provides expert knowledge to a client, to a customer, to a
[00:35:47] Paul Roetzer: partner?
[00:35:48] Paul Roetzer: Yeah. And I mean, if, so, if you're thinking of this and you're just like, mind blown, you're in good company. Like after I wrote the blog post, I sent it to Mike and our, our team. And then I texted like as soon as Mike got the post live, I texted it to some of my friends in the space and I was like, we have to talk.
[00:36:06] Paul Roetzer: Like I don't, I need to, once we get through the potting, I launch and like get it all. Like I really need to take a few days away and just think about this because like you said, it's happening so fast and you're seeing so many use cases emerge as more people were really smart, experiment with the tech and then they make connections about what it can do and you're like, damn, that was smart.
[00:36:28] Paul Roetzer: I was talking to my, my kids in fifth and fourth grade and I happened to be, mentoring an entrepreneurship class at their school. Beginning of the week. I think I mentioned this to you, Mike and I got talking with, the president of the school about impact on education and as soon as I explained it to him, his mind started lighting up with like, oh my gosh.
[00:36:48] Paul Roetzer: And he's rattling off ideas and things and I'm like, I hadn't even thought about that yet. Like there's so much opportunity moving forward to help. Really smart people understand what it is and then let their minds go to work on what does it mean to them, because we're not going to figure all that out, like, stuff to worry about.
[00:37:06] Paul Roetzer: But I just want to take a few days away by myself first to think about this, but then I want to start having like, think tank style conversations with really smart people in different areas because I'm so curious about where it goes. Yes. In marketing, but bigger than that to me. I, you and I talked like, I worry a lot about higher education.
[00:37:28] Paul Roetzer: Mm-hmm. and, and even into high school, like I do, I had concerns about higher education before this kind of stuff. But the one point I think I made on, on one of the posts was you, you could write a high school or college essay without a, without a problem with this technology. Today's technology, OpenAI playground, you know, you using DaVinci, the GPT model, ChatGPT mixed in and feel like you could a hundred percent write, it's going to pass any plagiarism check.
[00:37:55] Paul Roetzer: And most professors and even college administrators have no idea the tech exists. So what I was saying is like, if you're in higher education or in high school, like you have to establish a point of view on this at minimum. And then like, are you teaching? Like I had, I'll give you an example, that I was like having this existential crisis internal, personally.
[00:38:16] Paul Roetzer: My daughter had to do a project where they, for music class, had to take a song. They were learning how to write, basically how to rhyme. They had to take a song and rewrite the, opening verse and the refrain, and then write it to the style of that, that that musician and that song. So she had picked one.
[00:38:31] Paul Roetzer: I was like, well, I can't even figure this out. Like I can't figure out the rhyming sequence here. Like, we're going to have to pick something else out. So we picked, I think it was Mermaid by train and we rewrote it to, I helped her rewrite it, taught her how to like rhyme words and the sequence and how many syllables, earnings line, all this stuff.
[00:38:47] Paul Roetzer: I'm teaching her this. And we wrote it. Now, this was a couple days before chat GT came out. Hmm. And I asked myself after the fact, I was like, oh, damn. I probably could have given it the lyrics to that song and said, write a, a song in the, in the style of mermaid by train, about unicorns. And my guess is it would do it.
[00:39:08] Paul Roetzer: I didn't try it, but then I asked myself like, oh, would that be cheating? Like, would I, did I even teach my daughter that, that that is possible. Like, would I have, I mean, I've stopped doing piloting eye courses for 45 minutes to help her figure out how to write this . And I'm like, I probably could have done that in 45 seconds.
[00:39:25] Paul Roetzer: Yeah. But, but should I, and I actually don't know. Like, but these are the kinds of things like, I'm starting to wonder, will I, knowing what we know, would I give my kids the advantage of. Let me just show you, like, trust me, when you're in high school, college, like you're not going to need to be doing this. Like the AI is going to do these things.
[00:39:46] Paul Roetzer: Mm-hmm. . So let me show you now to give you a head start of how to use AI in your own education. Is that being a bad parent? I have no idea. Like, I haven't had time to grapple with these questions. But that's, that's what, like your question, why is it such a big deal? It's a big freaking deal. Cause it changes everything.
[00:40:06] Mike Kaput: you, you're telling me you haven't figured it all out in the five days since it's , it's come out.
[00:40:12] Paul Roetzer: This is the most, like, I've even really thought deeply about. Now I'm just getting more overwhelmed by the whole thing. .
[00:40:17] Mike Kaput: Well, I couldn't agree more. I feel, I feel like I need to. 72 hours and go to like a cabin or something and just with scotch,
[00:40:26] Paul Roetzer: we do it with scotch.
[00:40:27] Paul Roetzer: I'll pay for the cabin.
[00:40:30] Mike Kaput: Done. I like it . Now I think this is really interesting too to kind of bring this obviously back down to kind of what we're doing every day, which is some form of marketing. And you know, I look at this as a bot that can essentially answer any question, write thoughtful content on any topic, create basic plans and strategies.
[00:40:51] Mike Kaput: This is obviously going to have a massive impact on marketers, and we asked ChatGPT how it will change marketing. I won't read the full response, but it, it started honing in on customer service and engagement. Basically saying, you know, this is going to upend conversational AI or at least highly augment it and improve it, which I agree.
[00:41:14] Mike Kaput: And we can talk about that because I think your point's really well made. If I'm a conversational company right now, I probably should be having an all hands meeting figuring this out, right? But to me, that's also just the tip of the iceberg. Like what are the implications here across other areas of marketing?
[00:41:34] Mike Kaput: I mean, I just see this, I can't right now think of a single big area of marketing that you couldn't functionally start using this technology to at least
[00:41:44] Paul Roetzer: augment. Yeah. You know, I go through the list of the categories we teach about that we write about things like advertising and communications and PR and seo and email and content marketing.
[00:41:58] Paul Roetzer: Analytics, like, and I can't come up with any where it wouldn't affect them. Customer service and experience obviously. But even strategy, like I like, as you were talking, I was thinking like, oh, I should ask it. So like, let's say the new piloting AI for marketer series, like, what messaging should I use to convince a writer that they should, research and understand ai?
[00:42:24] Paul Roetzer: Like right now, if we were thinking about the ads we were going to create or the social media shares or the emails we were going to write, how would we do it? Like, you're trying to get yourself into the, per the, into the persona that you're writing. So you're trying to pick an individual to write. So you're trying to think about their pain points.
[00:42:39] Paul Roetzer: You're trying to think about what is the problem they're aware of? What is the solution you offer, what's the value exchange Like? We do all this as marketers, as writers, instinctually, sometimes we don't even think about it. You are just sit down and you just write an email, like, and you just know what to write or you write a blog post.
[00:42:55] Paul Roetzer: But like you said, it's almost like this. You almost have this like research assistant where we'll say, well, lemme see what chat GPD says. Like, maybe it has an idea I'm not thinking of. So what messaging should we use, to convince, I'm just going to do this again. just care about ai. So I just asked it. I see this is going to be good.
[00:43:16] Paul Roetzer: What messaging should we use to convince writers to care about ai? To convince writers care about ai, you should focus on the potential benefits AI offers their work. For example, AI can write, AI can help writers by providing them with suggestions for words and phrases, organizing their notes and research, and even assisting with the creative process itself.
[00:43:31] Paul Roetzer: Additionally, you could highlight the potential AI revolutionize the way stories are told and consumed, as well as the potential for AI or help writers reach new audiences through the use of personalized content recommendations and other tools. Overall, the key is to focus on the ways in which AI can improve the writing process and help writers to achieve their goals more effectively.
[00:43:47] Paul Roetzer: Damn, that's pretty damn good. . It is not bad. Kinda first go. Then my file would probably be like, okay, what do the pain points writers have? Like again, the things that the, a good writer, a good marketer, a good strategist is going to mentally do themselves to arrive at an output. Mm-hmm. , what if, as you're asking yourself those questions, you're just also asking the eye, maybe ausable, maybe it's not, but now we're talking about, like you said, a consultant, a strategy assistant that's different.
[00:44:17] Paul Roetzer: That is well beyond AI writing. It's, and that's, I think, like the mistake people could make is that it is, they think of it as too small. They don't think of the bigger implications. But yeah, I mean, there's so many, and I think the key is if you're an advertiser by trade, that's your main thing. Or if you're a communications professional or a consultant or email marketer, whatever it is, whatever you identify as in terms of the marketing realm, take this information and take a step back and think about what could it mean because.
[00:44:53] Paul Roetzer: Previously, it used to be we would ha go through these thought exercises. Well, what, what could the impact of aib? And we were thinking into some unknown timeframe in the future. Well, if they figure this out and this out, then it could change everything. Mm-hmm. , I think we're now at that point in the future, we were looking towards that.
[00:45:09] Paul Roetzer: Now when we say, well, what could it mean? We're talking about in the next 30 to 90 days, there's, there's very few AI innovations that I couldn't see happening in the very near future that wouldn't have a transformative effect on marketing. Like the tech is already there. We're seeing it. Now you're going to have really smart entrepreneurs and business people that come up with like, mind blowing applications of it, and that's where the innovation's going to just take off.
[00:45:39] Paul Roetzer: Like 2023. Again, I think I've said it a few times, like it's just going to be weird. Like, I don't know how to explain it. It's. Because we know there's, there's more coming. Like, yeah, this is going to improve in the next three months. So you may be listening to this in, you know, early 2023 and the stuff we're talking about is already like, I can't believe they were only talking about that in December.
[00:45:59] Paul Roetzer: Like, look what we have now, . It's going to happen.
[00:46:03] Mike Kaput: Absolutely. And I know we have kind of touched on the issue in the past of marketing agencies, which I consider, you know, a really good example because they're such, they're ubiquitous in the industry and they're in high demand, often being at the cutting edge of technology.
[00:46:19] Mike Kaput: If you are an agency, not getting everyone to drop maybe their other professional development at the moment and figure this out, I would say that's a horrible
[00:46:30] Paul Roetzer: idea. Yeah. Their job might not exist in 12 months, so. Right. I'm, I'm with you. Like I have no, you know, I don't have. There's a horse in the race is the right word, , right?
[00:46:40] Paul Roetzer: Like I don't run my agency anymore. I sold the agency, last year. I don't spend my time thinking deeply about how to infuse AI into an agency, but I ran one for 16 years. And there are definitely those moments where you're just kind of driving in the car and you're thinking like, my gosh, if I was still running an agency or if I have a lot of friends who still run agencies mm-hmm.
[00:47:01] Paul Roetzer: and, you know, we're still close with the people running our agency. It is so ripe that that business model, I want to put a positive spin on it. I would say there are massive opportunities to reinvent the agency model. . Yep. And yes, you're right. I mean, you get paid to figure things out and to build plans and to create content and to drive outputs.
[00:47:28] Paul Roetzer: And I don't know how you do that, in the very near future without AI being infused into every aspect of what you do. Couldn't, couldn't agree more. again, saying it in the most positive light, I can figure this stuff out fast. If you are an agency person, the
[00:47:46] Mike Kaput: positive part of this is you have no excuse to go use it right this second.
[00:47:51] Mike Kaput: It's free. It's pretty. As long as you know how to type a sentence, you're okay. I think. And you don't even have to use proper punctuation. It's pretty good at figuring out what you mean. So that's the positive to me is this doesn't require you to sit down and, you know, think about your whole existence necessarily.
[00:48:09] Mike Kaput: It go, start
[00:48:10] Paul Roetzer: experimenting with it. There's no learning curve. You, you experience the tech and then you apply your expertise to how to use it. And I think that's the opportunity in the future is no one's going to come from your specific perspective, your set of experiences and knowledge, and look at AI the same as you.
[00:48:31] Paul Roetzer: And that's what we need is more leaders at agencies, at brands, at nonprofits, at universities, whatever it is, we need people applying their own knowledge and experience to what AI enables. Mm-hmm. , because you and I aren't going to figure it out for them, we're going to try, like, we're going to try and go our best to educate people in specific areas, specific personas.
[00:48:51] Paul Roetzer: At the end of the day, the opportunity for you is you have a unique perspective on how to use this tool, but you have to understand it first. So, one
[00:48:58] Mike Kaput: other area of marketing I did want to touch on and kind of get our first draft thoughts on before we move to kind of how this is going to affect business at large is, I keep coming back to SEO in the sense that if there is tomorrow, just as a hypothetical, an answer engine that generates accurate data and information, and maybe you could click into it and say, okay, where exactly did that come from?
[00:49:25] Mike Kaput: Doesn't that get rid of the need for Google and for the content that is created for, for the search engines? I just wonder, I mean, obviously no one's going to stop using Google tomorrow, but over a medium term, I am really trying to like, twist my brain around that piece of it to begin with, because I don't know how that doesn't invalidate the entire premise behind seo if consumer behavior were to switch to exclusively using these
[00:49:56] Paul Roetzer: types of tools.
[00:49:58] Paul Roetzer: Yeah, so this is one, again, I've had a lot of passing thoughts and I haven't had the thoughts that go deep, deep on this and I also don't know that I'm the expert to do it. Like I know I have friends who are far more knowledge about SEO and search and the business model at Google than I am. But I'll, I'll share a couple of things that I immediately thought about.
[00:50:18] Paul Roetzer: So when I wrote the blog post, the first one, I actually asked that question because it was top of mind for me. It was like, How could ChatGPT change the way we search for information? Could it replace Google search? So chat GT's answer, it that ChatGPT could potentially change the way we search for information by providing more personalized and conversational responses to our queries.
[00:50:39] Paul Roetzer: Instead of typing out keywords and phrases, we could simply asked, ask ChatGPT A question in natural language and receive a detailed and accurate response. However, it is unlikely that ChatGPT could fully replace Google search as it is, as is. It does not have the same level of access to vast amounts of information and data.
[00:51:00] Paul Roetzer: Additionally, chat g PT is designed for more conversational interactions while Google searches primarily geared towards providing relevant webpages and results. Okay, so couple of the thoughts, and I I didn't write down an outline for this. I'm just kind of off the top of my head. My first thought is there is no way.
[00:51:21] Paul Roetzer: That Google didn't know this tech was coming right? There is also very low probability that they don't have something more advanced already within the walls of Google. Now, Google has far greater barriers to releasing an innovation than an OpenAI does, or a stability AI or hugging face or runway, or any of these major players right now in the general space.
[00:51:49] Paul Roetzer: So Google has to be more cautious about what they release and when and how. That being said, on the surface, I cannot imagine a near future where we are searching and scrolling through links to click on a bunch of things past a bunch of ads to arrive at an answer to a question, that seems in retrospect of what we know and insanely, An insane process full of friction when, you know, you can maybe just get a better answer by just asking the question.
[00:52:25] Paul Roetzer: Now it seems like what the promise of Siri and Alexa and Google Assistant were now, those haven't delivered to date on the promise of being like the perfect, immediate answer. So I do think that there is massive disruption in the near future for search as a whole. I have to imagine Google knows that and has a play in mind, but I don't, I'm not smart enough to figure out how they pivot their entire revenue model.
[00:52:54] Paul Roetzer: Mm-hmm. , which is dependent upon ads in search to do that. How you cannibalize yourself there and, and come out ahead. So that's one thought is I think it has a major impact. I just don't know what it is, but I think Google probably does have a plan. The second thing I think about is, are we on the cusp of one of the greatest comeback stories?
[00:53:16] Paul Roetzer: In business history being the Microsoft Bing search engine. Mm. So when Microsoft did the billion dollar exclusive license with OpenAI for GPT technology, and then I believed it, a follow on investment in not too distant past, I can't imagine that Satya Nadella would've done that without a Right refer, a first refusal to acquire OpenAI if they sold.
[00:53:44] Paul Roetzer: So there was that immediate thought to me of like, shit, does Microsoft just buy OpenAI? Could they even buy them? Like, I think, if I'm not mistaken, OpenAI has like a $40 billion valuation right now on a 500 x revenue multiple, like wild . Would they even want to sell? Like, I have no idea, but it would seem that Microsoft has first preference.
[00:54:05] Paul Roetzer: I'm just, again, making assumptions here that if they wanted to, they could go hard at OpenAI and potentially acquire them. And own the future of search. So if you think about how Google just came to just dominate search and everybody else who played, including Microsoft with Bing, just like, sorry, like game over.
[00:54:26] Paul Roetzer: And wouldn't that be wild if this reinvented search? And it was Microsoft that ended up with the keys to the kingdom. And again, I'm, this is total theory, total, just like thinking out loud, but I'm wildly fascinated to watch where it goes. But again, I would, I would not sell my Google stock. I would not assume Google is not smart enough to solve this.
[00:54:52] Paul Roetzer: They bought deep mine in 2014. Microsoft wanted deep mine then. Yeah. They, they have, they have the benefit of, some amazing talent and billions of dollars in AI research. So I think Google has another play to be made in the near future. I don't know what it is though.
[00:55:12] Mike Kaput: I think something you said there really jumped out at me is that, so I'm interested, really interested too in kind of the Google versus Microsoft versus whoever.
[00:55:21] Mike Kaput: But I think especially too, if you're trying to wrap your head around this, it's not always just about who's going to win. You said something that I have a hard time envisioning a future where we're clicking on links and scrolling through websites to try to find information when this technology exists.
[00:55:38] Mike Kaput: That statement alone should have a lot of people really sitting down and thinking about what that means. And yeah, like we said, I don't, I don't know what that means yet. Right. Until, and what about website
[00:55:51] Paul Roetzer: traffic? Right? Like, I have no idea. I haven't, again, I would love to get some brains together in this one, like grill them, because I wonder like, If there is this brilliant AI that knows everything and it can just ask it a question and you never have to come to our website or any,
[00:56:09] Mike Kaput: there's such thing as organic traffic,
[00:56:11] Paul Roetzer: what is, what do blogs exist for other than to train the ai How to answer the question, right?
[00:56:16] Paul Roetzer: Yeah. We, we, we have no traffic anymore for organic search. And is that the fu I don't know. Like I, yeah, I don't know. It's making me want to like set up a think tank and like really figure this stuff out that,
[00:56:31] Mike Kaput: that's awesome. Yeah. So I think in kind of the. Remaining time left. I think there's just so many and you know, it's so much, such a big topic for a single podcast episode.
[00:56:42] Mike Kaput: So I'm sure we will dive further into this in the coming like weeks and months as the world gets crazier. But I do kind of want to maybe riff a little bit more on kind of how this is going to change business just in general as we know it, to really just, I guess, hammer home for the audience, like the magnitude of this.
[00:57:00] Mike Kaput: And a couple things online in my research jumped out. Now these are, you know, disparate and kind of more random, tweets and conversations online, but I think there's an interesting set of dots to connect here. And one of them, and you've referenced this, is so first. I saw a tweet from Will Summerland, who is someone at ARC Invest, which is a major investment firm tweeting that AI training costs in general are declining 60% year on year.
[00:57:31] Mike Kaput: And if you really gamed out that math, it would be about, it took about 5 million to train GPT three in 2020. So this wondrous technology we're talking about 5 million probably sounds like a lot to me and you, but like to a big company is nothing. So that same level of performance they estimate would cost $500 to train in 2030.
[00:57:55] Mike Kaput: So he calls this AI training is Moore's Law on Steroids. So keeping that point in mind, we also have some tweets from, various people about the idea that these complex technical tasks like coding are now getting opened up to non-technical people because we can now prompt a machine. With simply writing, Hey, could you build me an app or a code block that does A, B, or C?
[00:58:23] Mike Kaput: And one of the guys who's really at the forefront of a lot of ai, Amjad Maad, is the founder, CEO of relet, which is kinda a, a developer framework, predicted that there's going to be a hundred x productivity boost thanks to ai, especially among programmers. So when I hear these stats and I see some of these conversations and start connecting the dot, Honestly, I kind of see the stage set for both monumental disruption of the existing order of business as usual and the economy and jobs.
[00:58:58] Mike Kaput: And I kind of also at the same time see maybe the greatest economic opportunity of many of our lifetimes. So how are you thinking about, at least, you know, again, five days out, the bigger impact of ChatGPT and just the overall acceleration of AI development on the economy jobs kind of business, as we know.
[00:59:19] Mike Kaput: I know we've gotten into various use cases, but really where do you see this having the kind of monumental impacts on society?
[00:59:29] Paul Roetzer: We, in May, I published that post the Future of Businesses AI Obsolete, which we've talked about on the show before. And the basic theory was that no matter what your business, no matter what your industry, matter what your profess.
[00:59:41] Paul Roetzer: That in the near future. You're that be AI native, you build a company from the ground up infused with ai, that's just smarter, more efficient than the peers competitors. AI emergent, you're an existing company. You figure it out. You have a visionary leader. You infuse AI across the company, marketing, sales, service, operations, finance, hr.
[00:59:58] Paul Roetzer: You build a smarter company, smarter version of your business and you do it real fast, or obsolete. You just become irrelevant and then you dial overnight. But your your ability to create value. Think about marketing agencies as the example. Agency A goes and infuses AI into all of its services, builds more efficiency, creates more value for clients for the same cost, but reduces its costs internally.
[01:00:18] Paul Roetzer: Increases profits from 15% to 35%. Like done. You can do it next year. How do you, how do you stay competitive if you're not them? Like, if you're, if you're the one that doesn't figure this out, you're done. And it might not happen in a year, but give it two, three years natural turnover process. Your clients are all gone.
[01:00:35] Paul Roetzer: I think when I wrote that post, I thought it was realistic that we were looking at a five to 10 year window that by the end of the, you know, the 10 years, you'd probably be at the stage where it was probably like 80, 90% of the companies were AI emergent and everybody else had just gone outta business that they just, they couldn't compete anymore.
[01:00:54] Paul Roetzer: Hmm. I think what happened with Chad GPT, some of the stuff we talked beginning with CICERO and Deep Minds Innovation, I almost wonder if that timeline didn't dramatically accelerate if, if we're not now looking at a three to five year window for a lot of these industries. So again, like let's take, when you think is safe, let's assume lawyers are safe, that, you know, attorneys could wait a while.
[01:01:21] Paul Roetzer: Law firms in general. I could have heard an argument that that was true back in May when I wrote the Post that maybe they had five to seven to 10 years because they got legacy relationships and all this stuff and, and now I look, I'm like, Yeah, I don't know about that. Like if, so let's say like, let's say I'm, I'm going to raise a round of funding that I want to, raise a Series A, but I'm a SaaS company, whatever.
[01:01:49] Paul Roetzer: I'm using a series A and I'm doing it through convertible notes, or maybe I'm going to raise through traditional vehicles. I'm going to do a valued round. I got all these questions. So I went through this raising a seed round. I had no idea what a convertible note was. I didn't know what evaluation cap was. I didn't know any of this stuff, but I raised a million dollar seed round for the institute.
[01:02:06] Paul Roetzer: Now I have a great attorney and they were phenomenal and I couldn't have done it without them. But in retrospect, 70% or more of what I spent to do that seed round was being educated on all of these things. Like what is, you know, what is a template contract for this? Okay, we'll use a KISS template. We'll use this.
[01:02:26] Paul Roetzer: Okay, what is that? And I'm like, so I'm Google searching. They're spending mobile hours doing things. And so I wonder. In the future, couldn't I just say like, I'm, I'm a media company. I want to raise a, a seed round, a million dollar seed round. What should I set my valuation cap at? What is evaluation cap?
[01:02:45] Paul Roetzer: What should I set it at? And I start going through, okay, can you spin me up a sample contract now? Mm-hmm. , I still need my attorney to make sure it's all factual. because the AI gets stuff wrong all the time. It doesn't know facts. It just like, it makes stuff up. So I still need the human in the loop. I still need the attorney in the loop to do these things.
[01:03:03] Paul Roetzer: But I start to look at that profession and say, wow, there's a lot of knowledge that you're paying for with attorneys. The same with agencies and consultants. You're paying for knowledge, knowledge that could be at your fingertips and maybe better from the people you're paying it for. So I think from knowledge work standpoint, we are, we are accelerating the disruption is the simplest Right?
[01:03:26] Paul Roetzer: Thing was if If AI, native AI merges or obsolete meant in the next five to 10 years. I think in a lot of industries that may have moved up to three to five with not only what happened in the last few weeks, but what's about to happen in early 2023.
[01:03:40] Mike Kaput: Yeah, that's a really good point. And I think it's also worth reiterating for our audience, because I often fall into this trap of thinking like innovation is just this clean thing.
[01:03:53] Mike Kaput: Like you're either disrupted or you're not. And like that's true, but also not every single person needs to be replaced, or AI doesn't have to suddenly be superhuman at being a lawyer tomorrow to radically change the economics of that industry. Right? You might still have the top 15% of lawyers on the planet may get paid.
[01:04:15] Mike Kaput: They might have, they might 10 x their income, but. There be negative effects on the other 85%. So I think it's worth, there's no way to predict that, but it's worth people taking this seriously now, even if the technology is imperfect, because you don't need to hit a hundred percent disruption or displacement to truly transform an industry.
[01:04:36] Mike Kaput: I mean, it's often rarely how
[01:04:37] Paul Roetzer: it works. Yeah. And the thing I'll say there, and again, you know, Stanley, the attorney example, what, what I'm saying is not that attorneys won't matter or doctors won't matter, or writers won't matter, or consultants won't matter. But what I wrote in one of the recent, my LinkedIn post, so I used a quote from Curtis Lang.
[01:04:58] Paul Roetzer: Lotz was a professor radiology and biomedical informa informatics at Stanford University School of Medicine. And he had a quote in 2018 in a paper about AI that said AI won't replace radiologists, but radiologists who use AI will replace radiologists who don't. I think that is replaced radiologist with X.
[01:05:19] Paul Roetzer: So whether it's, marketers, writers, designers, lawyers, accountants, brokers, developers, engineers, whatever. So I think the future is that I love my attorney. Like I wouldn't trade 'em for anything. The insane, insane Knowledge, amazing network does incredible value add stuff for me all the time. If they figure out how to infuse ai, so rather than them having to spend 10 hours on something, they can spend two mm-hmm.
[01:05:43] Paul Roetzer: and they can still charge me a value based rate for it. But now they can do that 10 more times than they would've done. So they're 10 xing their output. So that's what's going to happen is lawyers, accountants, marketers, consultants, people who figure out how to infuse AI into their business models are going to have such a disproportionate advantage over people who don't.
[01:06:02] Paul Roetzer: And I think ChatGPT accelerated the ability for the average business person, average knowledge worker to infuse ai. Hmm. So until now, it wasn't. That easy. The barriers weren't down, the barriers are gone. And so now anybody, again, I give my kids school president, he was on ChatGPT, like then showing him Dolly, like, here, go get Dolly.
[01:06:26] Paul Roetzer: Like you can start at any level, in any profession, you can start experiencing it and as soon as you do, your mind lights up with what can happen. And so again, I don't, I wouldn't put it past like great developers and brokers and attorneys, like they're going to figure this stuff out, but they're going to be a lot who don't because they don't think it, it matters and they think it's just a trend and there's no motivation there, but it's going to happen.
[01:06:52] Mike Kaput: That's such a good point that I think is worth reiterating is this idea that why this matters so much is the democratization of getting AI in the hands of all these different domain experts. Like you said, we can't solve for every use case. We don't know everyone else's industry. Every single person out there represents an opportunity for someone to understand both the technology of ai, which is the newer thing we're talking about with marrying their 30 years of experience in some industry.
[01:07:24] Mike Kaput: None of us have even thought of using AI in. That to me, is really exciting and I think every single one of those people represents the true disruption of, of this technology. So, you know, I understand people's skepticism out there, but that point would give me some pause and say, wow, now that everyone and anyone can experiment with this stuff, you just unleashed.
[01:07:48] Mike Kaput: All the smartest people in every industry on the planet, figuring out if they so choose how this could work in their domain.
[01:07:56] Paul Roetzer: Yep. Yeah, I think that was, I mean, I've said it before, I think the acts, the accessibility of AI is what changed everything. Dolly was the start to me in April when it came out, whenever, may, June.
[01:08:07] Paul Roetzer: You could get it, get off the wait list and use it. And once you could use it, you could understand it. And once you understand it, you connect the dots to the impact it'll have on you. And so that's the genius of the OpenAI model is like, make the stuff free at first. Like put it out there. Not only do they learn from how people use it in training data mm-hmm.
[01:08:26] Paul Roetzer: it creates a wider understanding of what's possible. And so, yeah, I mean it's, we are, we are in for a, a wild ride and I don't think there's a profession that isn't changed by it. It is exciting. It is terrifying. It is, unfamiliar and. And abstract, in terms of the impact it'll really have. But I think you're just going to see this level of innovation that we've been experiencing the last few weeks.
[01:08:54] Paul Roetzer: I think it's going to continue. I mean, you're going to be seeing weekly new stuff coming out, new companies, new technology, new use cases and industries you couldn't even imagine that come up with creative ways to do it. I think that's the part that's really going to really accelerate the exponential curve of AI adoption is all these innovative use cases for it now that everybody can use it.
[01:09:15] Paul Roetzer: Well,
[01:09:16] Mike Kaput: given that, I mean maybe I'll have to manifest this into reality. Maybe we have to go to two times a week on the podcast
[01:09:26] Mike Kaput: I know that's happy hubris coming out a break of
[01:09:30] Paul Roetzer: three weeks, but I mean it, we honestly could do it daily on this at this point. We really, especially if we start interviewing people who can figure this stuff out, like if we start interviewing people who are building this stuff, so, yeah. Yeah. I don know, maybe when we retire from actually building the institute, we'll just do the podcast every day.
[01:09:44] Paul Roetzer: There you go. .
[01:09:46] Mike Kaput: Well, I know we're getting to time here, but any final kind of closing thoughts? Any, I mean, we've covered a lot of ground and I feel like we've still only scratched the surface. I'm super excited, but I'm also like, whoa, okay. have to move. have to go
[01:09:58] Paul Roetzer: figure this out. . No, I've said it I think on the podcast before, I know I've sent it on stage giving talks like we have spent the last, you know, I've spent the last 11 years researching this, last, eight, writing about it and speaking about it, the last six running the institute, you know, with Mike's help and building this and trying to create an audience around this.
[01:10:19] Paul Roetzer: And we have been very cautious. To not create fear. We have made every effort in our messaging, in our content to, make ai in a way aspirational, to make it inspirational, to, to drive curiosity and let people explore it. And we have really, intensely tried to avoid the tone that would create fear.
[01:10:42] Paul Roetzer: What I would say now is, if you haven't picked up on it, we have an increased sense of urgency. It is very, very important. And this is not overstating the, the state at which we are, you have to solve for this. You, you have to make understanding AI and figuring out how it's going to impact you and your company, a top priority in Q1 of 2023.
[01:11:07] Paul Roetzer: Like there's no waiting anymore. And I think that's the biggest change with what's happened in the last few months. You cannot wait around. You have to solve for it. You have to figure out how it's going to impact your company. You have to figure out how you can take advantage of it in your career.
[01:11:24] Paul Roetzer: That's the piloting AI from marketer series I mentioned being. That's why we built it. It's like, okay, everyone has to solve for this. What do they do? We can't just give 'em 50 courses to choose from. Can't just say, go read 900 blog posts. We had to create a learning journey. And so that was, that was where it came from, was this need of like, we have to solve for this.
[01:11:40] Paul Roetzer: We have to help people figure this out. And so maybe that's the path for you, or maybe it's just doing your own re whatever it is, commit to yourself that you, you have to do this. Whether you're a marketer, a business leader, an educator, a student, whatever it is, this is the future and it's happening right now.
[01:11:57] Paul Roetzer: And I just don't want you to waste another minute without seeking this information and continuing to explore it. I love it.
[01:12:08] Mike Kaput: That is,
[01:12:08] Paul Roetzer: That probably we can transcribe that, because that was probably the best way I've
[01:12:11] Mike Kaput: said it, right? Yeah. We need that should be, that'll probably be at the top of this.
[01:12:17] Mike Kaput: Yeah. Right. No, that's awesome advice. And as always, thanks for the time. Thanks for the insight. I think we've got a ton more to talk about. I can't even imagine, you know, next week we'll have another year's worth of developments to discuss, I would
[01:12:32] Paul Roetzer: imagine or we'll just go back to the beginning and pick one of the seven things that happened that we didn't talk about.
[01:12:38] Paul Roetzer: Yeah. So thank you everyone for joining us. Again, to the listeners who reached out in our short absence and let us know that they missed us, that means a lot. We appreciate that. If you are a regular listener, we'd love to hear from you. Connect with us on LinkedIn. Reach out, join the AI community we're building, through the institute.
[01:12:55] Paul Roetzer: And yeah, be sure to give a like, and, you know, leave a comment on the blog or a blog, the podcast. As we keep trying to build this community and, and spread the understanding of AI so people can figure it out and take advantage of it. So until next time, we'll talk to you then. Thanks, Mike. Thanks Bob.
[01:13:11] 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.
[01:13:33] Paul Roetzer: Until next time, stay curious and explore ai.
Cathy McPhillips is the Chief Growth Officer at Marketing AI Institute.