This week’s episode of The Marketing AI Show touches on generative AI, and you guessed it, ChatGPT. But it’s not more of the same. APIs and HubSpot take ChatGPT to the next level. Tune in!
ChatSpot…the latest in ChatGPT
The week is starting off with a big development. Just yesterday, Monday, March 6, HubSpot co-founder and CTO Dharmesh Shah released ChatSpot, an AI tool that combines the power of ChatGPT, image generation AI, and HubSpot’s CRM. The tool lets you ask questions of your HubSpot portal and provide instructions in natural language through a chat interface. For example, you can use ChatSpot to give you a summary of data in your portal, create a report of companies added last quarter summarized by country, or generate an image of an orange rocket ship. Mike and Paul break down this latest development and what it means for HubSpot customers and agencies.
The biggest winners generative AI tech stack…so far
Legendary venture capital firm Andreessen Horowitz published a deep dive into the generative AI market: “Who Owns the Generative AI Platform?” To create this, the firm met with dozens of startup founders and operators who deal directly with generative AI to better understand where the value in this market will accrue. Andreessen breaks down the generative AI tech stack into three main categories:
- Infrastructure – the cloud platforms and chips used to train models
- Models – the foundational models like GPT-3 that power generative AI tools
- Apps – the actual products like Jasper that customers use
Andreessen observed that infrastructure vendors are likely the biggest winners in this market so far, capturing the majority of dollars flowing through the stack. Application companies are growing topline revenues very quickly but often struggle with retention, product differentiation, and gross margins. And most model providers, though responsible for the very existence of this market, haven’t yet achieved a large commercial scale.
Bottom line: the companies creating the most value — i.e. training generative AI models and applying them in new apps — haven’t captured most of it.
APIs are available for ChatGPT and Whisper
We knew it would happen soon: developers can now integrate ChatGPT and Whisper, OpenAI’s human-level speech recognition system, into apps and products through the company’s API. Since December, OpenAI says it has reduced the cost of ChatGPT by 90%—savings that API users will now receive when they use it, making it much easier and cheaper for companies to incorporate the capabilities of ChatGPT and Whisper into their businesses.
However, this doesn’t just mean every business can have its own instance of ChatGPT. It means they can use these capabilities to build innovative new products.
And tech and e-commerce companies are here for it. Already, Snap, the creator of Snapchat, introduced My AI, a customizable on-platform chatbot that is built on the ChatGPT API. Instacart is using the ChatGPT API to pair ChatGPT with its own data so that customers can ask open-ended natural language questions. And Speak is an AI language learning app and the fastest-growing English app in South Korea. They’re using the Whisper API to power an AI-speaking companion product. It’s impressive to see the API in action.
These advancements and developments—happening at lightning speed—have an immediate impact on the marketing world. Paul and Mike help us uncover new opportunities and possibilities.
Listen to this week’s episode on your favorite podcast player, and be sure to explore the links below for more thoughts and perspectives on these important topics.
00:03:05 — HubSpot introduces ChatSpot
00:17:13 — A deep dive into generative AI
00:30:15 — ChatGPT and Whisper APIs
00:39:05 — Rapid fire topics – AI regulation, US/UK lawmakers struggle, Figure’s big announcement, WIRED’s generative AI policy,
Links referenced in the show
Watch the Video
Read the Interview Transcription
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: 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] Paul Roetzer: My name is Paul Roetzer. I’m the founder of Marketing AI Institute, and I’m your host.
[00:00:28] Paul Roetzer: Welcome to episode 37 of the Marketing AI Show. We have a lot to cover today. We had to bump some topics like last minute. There’s so much going on. So I am your host, Paul Roetzer, along with my co-host as always, Mike Kaput. Good morning, Mike.
[00:00:42] Mike Kaput: Good morning, Paul. I feel like every podcast, like I’ve aged another year because it feels like things are moving so
[00:00:48] Paul Roetzer: fast.
[00:00:50] Paul Roetzer: probably la the end of last week, like slowed down a little. I like everybody apparently was waiting till Monday morning to drop some stuff on us. , which is great. I mean, I’m glad we do this. So if you, if you’re new to the show, we record this on Monday mornings, so today is Monday, March 6th. This will release on Tuesday, March 7th.
[00:01:08] Paul Roetzer: And some of the things we say today may be out, out of date by the time this hits tomorrow. Who knows? . So today’s episode is brought to you by the AI for Writer Summit. It’s ai writer summit.com. This is a virtual event that we’re launching that’s come, that’s happening on March 30th this year. So just in a couple weeks now.
[00:01:27] Paul Roetzer: It’s a free event. There’s a free option at least. And as you’re going to discover in today’s session, there is a lot happening in the world of AI writing and language AI and ChatGPT. And all these other language models and applications. And the reality is nobody has a clue what’s happening. And so we thought, let’s create a summit and bring some people together.
[00:01:47] Paul Roetzer: And let’s talk about this. What’s the impact on you as a writer or a marketer, or a business person or a technologist who’s building tools, whatever your role may be? What is going on? What does it mean to us? How is it going to impact careers and talent and, and companies and marketing strategies? And so it’s a half day virtual summit on the 30th from 12 to four Eastern.
[00:02:11] Paul Roetzer: We have a, a collection of amazing topics. Talking about what’s going on today, state of, we’re going to take a look at the future and figure out where do we think this is going to go, you know, over the next six to 12 months. I don’t think it’s realistic for anyone to try and predict beyond six months at this point.
[00:02:26] Paul Roetzer: Sometimes I think six days is hard to predict , but, lot’s happening. So again, For Writer Summit is brought to you by Marketing AI Institute where Mike and I are both employed. Marketing AI Institute is the creator of the event as well as this podcast, so check that out. Again, there’s a free registration option, ai writer summit.com.
[00:02:46] Paul Roetzer: And with that I’m going to turn over to Mike again, if you’re new. Mike and I pick three topics each week, and then we have a little rapid fire session at the end. The three topics today changed about 30 minutes ago. Thanks to our friend Dharma Shaw at HubSpot. And with that I’m going to turn it over to Mike to talk to us about what’s going on.
[00:03:04] Paul Roetzer: Thanks,
[00:03:05] Mike Kaput: Paul. Yeah, this first topic is so hot off the presses that I think it’s still warm, honestly, like I can feel the heat coming off the Google Doc here. So literally just this morning, HubSpot co-founder and C T O Dharma Shaw released something called Chats Spott, which is an AI tool that combines the power of chat, G B T Image, generation AI and HubSpot cr.
[00:03:29] Mike Kaput: And c m s platform. So this tool lets you actually ask questions of your HubSpot portal and provide instructions in natural language through a chat interface. So it’s almost like think ChatGPT for your specific HubSpot instance. For example, you could use chats spott to do things like. Give you a summary of data in your portal.
[00:03:50] Mike Kaput: Create a report of companies that were added last quarter, summarized by country, or generate an image of an orange rocket ship. For any of our audience members who don’t know, there’s some really important context here. Not only is HubSpot a leading marketing platform, but Paul actually founded and sold a marketing agency that I also used to work for.
[00:04:10] Mike Kaput: That was HubSpot’s first ever agency partner. So we followed HubSpot closely. Very closely since its very early days, and have a lot of relationships at the company. On the heels of this announcement, HubSpot also introduced kind of with less fanfare, A G P T powered content assistant that generates content in the platform.
[00:04:32] Mike Kaput: So we’ve alluded to this feature. That we would’ve loved to have this feature for a while in HubSpot, and it looks like they have now incorporated it. So Paul, I want to turn this over to you to just unpack your thoughts for us on this release on chats Spott on HubSpot at large, and their play in AI and what this all means for marketers.
[00:04:53] Paul Roetzer: Yeah, so the, as you alluded to, Ihave a long history with HubSpot. We started working with them back in 2007. Dharmesh wrote the forward for my second book in 2014, which is the first time I publicly talked about artificial intelligence. In that book, I theorized something called a marketing intelligence engine that would use C R M data and automation system data to build predictive models for strategy and budget allocation.
[00:05:18] Paul Roetzer: And it was sort of like the reason I got into AI was this, this use case of an intelligence engine. I was about a decade ahead of my time, apparently because , it wasn’t not only possible, but no one was working on it back then. So that being said, I have had many conversations about artificial intelligence with Dharmesh, in particular and with people at HubSpot through the years and at times I’ve honestly been a little hard on HubSpot in recent years about the lack of a AI in their platform.
[00:05:48] Paul Roetzer: When I saw that there was a massive opportunity for them to be building it with all the proprietary data and everything. And, and so Ithink that we have to give kudos where it’s due. I mean, I, what what he showed in the, it was like a 19 minute demo where Dharmesh kind of walks through it. One, Ilove that it’s Dharmesh doing it.
[00:06:05] Paul Roetzer: Like, you know, I always say the companies that you want to bet on are the ones where the leadership. Are are out with a point of view on ai. There’s still so few SaaS companies where the C-suite don’t even have a public point of view on ai. They’re not talking about it. And here we have Dharmesh actually not only building, this with his team, but he’s out in front as the face of what they’re doing.
[00:06:27] Paul Roetzer: And obviously Dharmesh has invested a lot of time and energy in figuring out where HubSpot might be going in this area. And so, you know, I think it’s, it’s awesome and Ihad seen some of this. Ihad a, a sense of kind of like some of the stuff that, you know, d Darmesh was working on. But I hadn’t seen what he, you know, showed today in terms of the full capabilities of this.
[00:06:47] Paul Roetzer: So I was really excited to see it. And I love the practicality of the use cases. He showed some sales CRM use cases, he showed some marketing use cases, he showed some reporting use cases. So you can’t, as a HubSpot customer or marketer watch that and not realize the impact AI is going to have on you. I also thought it was interesting because it sort of brought home the world of bits stuff that you and I talked about on episode 35 of the marketing AI show.
[00:07:16] Paul Roetzer: So if you listened to episode 35, we sort of put out this world of bit’s idea that, you know, really where people were going was action transformers, where the AI would be able to take actions on your behalf, not just, you know, create things, but you. The example we used in the post was a 21 click sequence to send an email in HubSpot.
[00:07:33] Paul Roetzer: So like today, if you wanted to send an email in HubSpot, you have to click at least 21 times to do it. And Darmesh actually referenced, that kind of example within what they’re building with chats spott, where it reduces the need for all these clicks. So it’s sort of like a lot of this stuff coming together simultaneously.
[00:07:50] Paul Roetzer: And Ithink it’s, it’s really exciting. You know, in terms of the impact it could have on customers. The impact it can have on the broader market. So
[00:08:01] Mike Kaput: I think it’s also important. To layer in a little context here, given that, you know, we both spend most of our days in HubSpot. We’re very familiar with the partner ecosystem and services agencies offer around HubSpot.
[00:08:15] Mike Kaput: I wouldn’t sleep on how powerful this idea is. There is still a lot of manual work that goes into actually executing anything in HubSpot. Not to mention as anyone who. Struggles sometimes to get value out of their c r m knows it’s not always easy to get at all the data that we have in there. I mean, how many times over the past, you know, 15 years have we sometimes struggled to say, Hey, like, how do we pull a report on this data set?
[00:08:43] Mike Kaput: How do we get more value out of the data we’re collecting on contacts? So all of these things for people that. Aren’t as familiar with HubSpot. These are all really transformative. So how do you see this? What does this mean for the wider market that HubSpot is now integrating AI into its platform like this?
[00:09:03] Paul Roetzer: The, to start, this sort of tees up what we’re going to talk about in our second topic. But, but basically I think it create, it’s going to create a lot of confusion. . So if you have been an early adopter of AI technology and you’ve been adding different applications, different third party writing tools, different, different intelligent automation tools, Once this moves outta alpha, you know, Dharmesh is very clear.
[00:09:25] Paul Roetzer: It’s an alpha, launch, so it’s very early. But it’s going to get better fast, for a lot of different reasons. And so I think you immediately are like, well, do we need. The other writing tools, like what does this replace the need to have a third party writing tool? And I actually don’t know the answer to this.
[00:09:42] Paul Roetzer: I’m just posing this as, as soon as I saw it, that was my immediate reaction was people are going to be very confused and we are already dealing with an undereducated industry at large. . So generally speaking, people are buying AI technology that they don’t understand. They’re finding some, some basic use cases that they do understand though.
[00:10:00] Paul Roetzer: Okay. It can help me write blog posts, it can help me do social media shares, it can help me write ads or whatever it is. So they understand the use case, but they don’t understand the underlying technology. So they certainly can’t step back and understand do, do I need this in HubSpot and and these other third party tools, or is this just it?
[00:10:17] Paul Roetzer: Because they might not even understand language models and how the APIs work and all this different stuff. So I think that there is going to be some confusion. There’s going to be a greater need for educat. On the positive side, I think HubSpot, jumping in to AI in a bigger way helps accelerate that, the understanding and adoption, because now that it’s baked in, and again, it wasn’t like HubSpot didn’t have some AI tools, but realistically, and again, this what I was referring to, I’ve been a little hard on them at times.
[00:10:44] Paul Roetzer: Their artificial intelligence features were the same, a week ago as they were two, three years ago. . , like they have not innovated at a, at a rapid. And so a lot of the, like, there was like 10 different features and there were, there was interesting stuff, but there wasn’t anything groundbreaking by any means within what they were doing.
[00:11:02] Paul Roetzer: And they knew that, and Ithink they would be the first to probably admit that they, you know, maybe weren’t innovating at a very rapid pace on the AI front. I think that’s about the change. And so by infusing it in, let, I mean, let’s think about the context or if you’re not familiar with HubSpot, it’s a, a, as of this morning, a 20.2 billion market cap company, their total revenue in 2022 was 1.7 billion, up 33% versus 2021 in a down economy, and they grew their customer base to 167,000 customers by the end of 2020.
[00:11:34] Paul Roetzer: Which was up 24% over 2021. So, I mean, this is a, a very important company in the overall marketing sales c r m ecosystem. And for them to go all in again with a co-founder, C t O in Dharmesh doing this and leading a charge, I think that’s going to wake a lot of people up to what a, that AI matters and it matters right now, and they better make some moves to figure it.
[00:12:00] Mike Kaput: Yeah, so maybe talk to us about that a little more on a practical level, now that these AI powered capabilities are part of the platform, and presumably we’ll see more of them coming down the line as well. What effect will this have on the work that marketers are doing in HubSpot?
[00:12:19] Paul Roetzer: Once you understand what it’s capable of, and again, just go watch the demo video.
[00:12:23] Paul Roetzer: You know, if you, if you run sales or involved in sales enablement, or if you’re a marketer and you create content or you run reports, whatever, you can go yourself and look at the use cases and think what’s going to be possible. Now, again, you may not realize like massive productivity increases in efficiency right away, because again, it’s an alpha solution.
[00:12:42] Paul Roetzer: But you can connect the dots and realize, wow, this is going to make a major. So I think productivity for HubSpot customers will almost immediately rise once you have access to the tools and it’s just going to keep going up. And I’m not sure that there’s a cap on that. So if you, again, like take, take a practical example of, drafting emails or writing blog posts or pulling reports on number of new contacts added last month and how many were in the SaaS industry and all these things that you do all the time.
[00:13:11] Paul Roetzer: And you can just do that through not only typing it, but he has a voice function in there. So I can just say, How many re, how many contacts did we had last month? How much was it over the month prior? How many of them are in the SaaS industry? How many of your companies that are a hundred million or more in revenue?
[00:13:24] Paul Roetzer: Like I can just ask these things that historically you all get from Zoom messages from me. Can someone please go into HubSpot and pull for me X, Y, or Z? Now, in theory, I’m going to be able to just pull my HubSpot app while I’m sitting at home and say, Hey, how many contacts have we added this month? Like, whatever pops in my head.
[00:13:41] Paul Roetzer: And so the productivity is going to be massive. Now that said, It’s going to be very disruptive. So one, you have to, you have to have people who understand this technology and the capabilities you have to upskill your team to even know what to do with this technology. But at a bigger level, it starts to lead into this the ecosystem.
[00:14:00] Paul Roetzer: So HubSpot is built through a massive ecosystem. And again, you mentioned upfront, we were the first HubSpot partner back in 2007, so we were the origin of the partner program that. Last I had heard publicly again, this is an inside knowledge, like 45% of HubSpot’s revenue came through their partner ecosystem, and that includes service providers, like marketing agencies and technical providers who build on top of the HubSpot APIs within, you know, the solutions marketplace.
[00:14:26] Paul Roetzer: So it’s a massive part of their business. If, if I’m an agency right now that gets paid to create content and pull reports and do the things he shows in. I am, I am really starting to wonder what is the future of our service model with HubSpot? Like what is Hub? What is Chats Spott not going to be able to do is probably the question I’m asking myself right now.
[00:14:48] Paul Roetzer: Because you have to start moving your service model to where they’re not going to be able to just ask a question and get the thing they used to pay you a thousand dollars for, and that’s where it’s going. Now, chats Spott may not be there, and that’s not the intention of what they’re trying to do. I mean, knowing HubSpot, they are not trying to obsolete their own service.
[00:15:04] Paul Roetzer: But it’s a, a potential byproduct of it. Itat least in parts. So I think that, and then I think from an a technological standpoint, if I’m a, a developer and I’m building tools into HubSpot, I am starting to wonder what won’t HubSpot be able to do on its own with chats spott? So I’ll give you an example.
[00:15:22] Paul Roetzer: I’ll throw this out there now because it’s an obsolete idea. One of the products I had considered building, so we’ve played around at different times with building some AI technology, was a c r m writer. Iwas thinking about building an, an AI writing tool that had access to CRM and could write anything you want based on the CRM data.
[00:15:40] Paul Roetzer: You see Microsoft doing this with Viva, you know, they’ve made a play into this space, and I’ve thought for a while that HubSpot needed. And so I was like, yeah, maybe we’ll build that. Maybe we build a, like, like a cool little app that, you know, can write based on CRM data. We, it’s done. Like if I, if I had built that product, it would be obsoleted right now, and I would be emailing my investor saying, Hey, sorry, darash just built this into chats spott, like, we’re, we’re done.
[00:16:01] Paul Roetzer: Thanks for the million. We just burned. So Ido think there is a very, and again, this leads into our second topic very well. There is a massive amount of uncertainty right now around what should be. And, and what the moats are to defend what you build. And I think this is a perfect example, that there’s a number of tools that I’ve thought about building on top of HubSpot the last couple years that I could now see having been obsoleted by this one Alpha play.
[00:16:31] Paul Roetzer: And, you know, as Dharma shows, he’s, you know, he shows his little grid of like, this is just the very beginning and he was this a much bigger idea. And knowing Dharmesh and having spent a lot of time with him through the years, you know, personally and professionally, Dharmesh is an insanely smart person and a very big thinker who can actually build the stuff he envisions.
[00:16:50] Paul Roetzer: I would not bet against Dharmesh. Like, you know, as a, as someone who, who has invested heavily in HubSpot through the years. I am, I am very happy to know that Dharmesh is, very focused on AI in HubSpot. I think it will bode very well for the platform and the stock moving forward.
[00:17:08] Mike Kaput: So you bring up a great point there.
[00:17:11] Mike Kaput: About defensible moats. And as our second topic shows, even the smartest people in the generative AI space are having a hard time figuring this out. So legendary VC firm, Andreesen Horowitz, they just published a deep dive into the generative AI market, and it’s called Who Owns the Generative AI platform.
[00:17:34] Mike Kaput: So they basically have met with dozens of founders, operators, people are working. In generative AI to kind of understand where the value in this market will actually accrue, and they break down the generative AI tech stack into three main categories. First is infrastructure, so think the cloud platforms and the hardware and chips used to host and train models.
[00:17:56] Mike Kaput: Then second is the models themselves. These foundational language models like GPT-3, that power generative ai. And third are the apps, the actual products like the Jaspers of the world that customers use and. The full post is well worth a read and we’ll link to it in the show notes, but Andreessen’s conclusion is essentially as follows.
[00:18:18] Mike Kaput: They say that we’ve observed that infrastructure vendors are likely the biggest winners in this market so far, capturing the majority of dollars flowing through the stack. Application. Companies are growing top line revenue very quickly, but often struggle with retention, product differentiation, and gross margins.
[00:18:35] Mike Kaput: And most model providers, though responsible for the very existence of this market, haven’t yet achieved large commercial scale. In other words, the company’s creating the most value i e training generative AI models and applying them in new apps haven’t captured most of it. They also strongly note that today they don’t appear to see many competitive moats at all in this market, but the market.
[00:19:00] Mike Kaput: Is so important to understand because it’s so vast. They say, quote, the potential size of this market is hard to grasp somewhere between all software and all human endeavors. So we expect many, many players and healthy competition at all levels of the stack. Paul, let’s start off by talking about what this means specifically for marketers before we dive into the overall market dynamics.
[00:19:25] Mike Kaput: So if I’m a marketer, what should I be taking away from this?
[00:19:29] Paul Roetzer: Yeah. First of all, Imean, just kudos to the three authors of the piece. I think it came out in January. They don’t put, this drives me. They don’t put dates on it. They’re posts. I actually don’t know exactly when it came out, but it came on my radar last week.
[00:19:42] Paul Roetzer: So it’s possible this has been out since like late January. , but it is, it is a really well written piece. Like it does a great job of connecting the dots and explaining some relatively complex concepts. So gr great job. It is an 18 minute read though, so like you’re, if you’re in, you’re, you’re committing to, to going deep on this topic.
[00:20:01] Paul Roetzer: My first reaction was for marketers and business leaders, the main consideration is at that application level. The software companies that are being. to enable you to do things more efficiently, to be more creative, to improve your decision making, to personalize, you know, whatever it is. The key to me here is like, how in the world are we supposed to bet on the winners?
[00:20:24] Paul Roetzer: . Like if I’m building my tech stack and I need an AI writing tool at the core of that tech stack, and I go get Jasper or writer or hyper write or word tuner or, or I just go direct to OpenAI APIs and I build my own stuff. It’s really, really hard if the VCs who do this for a living can’t figure out who the winners are going to be and what their mode is going to be.
[00:20:49] Paul Roetzer: As a marketer who doesn’t spend your entire day analyzing this, how do you figure out which SaaS companies to bet on? And this leads me back to what I was saying at the beginning about the confusion in the marketplace. It’s way safer if I’m a marketer to. I’m just going to bet on HubSpot. Like I, they’re, they are core to my tech stack.
[00:21:09] Paul Roetzer: It’s what we’ve said for years for people. How do you get started with ai? I have said it over, in, over inpe. It talks you and I wrote it in our book, start with your existing tech stack. the key way to infuse AI in your business right now is go ask your existing core tech companies, do you have smarter features?
[00:21:28] Paul Roetzer: We’re not. Because if the answer is yes, we actually have an AI writing tool baked into our platform, or we have a generative AI tool that creates images in our platform, great. I don’t have to go looking for one that I don’t know the company behind them. I’ve been using HubSpot for seven years. I trust you all to do this, right?
[00:21:44] Paul Roetzer: I’m going to just ride with HubSpot and just see this out. So, That to me is the biggest challenge and, and like we’ve been saying, these big tech companies, were going to figure it out. It was only a matter of time until the platform companies realized that they needed to be building this technology in.
[00:21:59] Paul Roetzer: And as soon as they do it totally shakes up the ecosystem now of, well, who do I go for point solutions? I just need a writing tool or I just need an email tool, or I just need this. And it’s only logical that the platforms are going to build those capabilities in. Now they might do it through ecosystem partner.
[00:22:16] Paul Roetzer: But in a case like this, with the innovation happening with ChatGPT and the opening up of that a p I, which we’re going to talk about as the next topic, it just like resets the ecosystem in my opinion. So that was my, my first takeaway away is that you need to focus on the applications layer and there is a, a massive amount of uncertainty in that layer.
[00:22:37] Paul Roetzer: And then the other thing that jumped out to me, They just don’t know. Like they’re, they’re fairly confident the infrastructure layer where the clouds and the GPUs, like Nvidia, and I’ve been bullish on NVIDIA for years. Like if anybody’s listened to me, give talks, . And so Ithink that the infrastructure layer seems really obvious, but that’s more of a like personal investing thing.
[00:23:00] Paul Roetzer: Like if you’re a marketer or a CEO. The infrastructure layer doesn’t matter that much to you. I mean, other than the cloud you choose to, to participate in, like whether it’s AWS, Microsoft Azure, or Google Cloud. I mean, those are your three main players. So that decision, you know, is something you can affect.
[00:23:17] Paul Roetzer: But those. The things that are being built are being built on GPUs and TPUs from Nvidia and Google. . . And that’s really not something that affects you. So then the next layer up where you’re actually getting into the models that that is starting to come into play. Like, I’ve been doing this myself.
[00:23:32] Paul Roetzer: It’s like, well, we have OpenAI’s playground, we have Coheres playground, we can, you know, you have Thropic, you have the Stability AI. Like you have all these people that are now building models. And the question becomes, well, if I wanted to build my own thing, do I just go to those? But again, the most obvious short term thing you have to solve for as a marketer or business leader is which applications are we using?
[00:23:53] Paul Roetzer: Which SaaS companies are we partnering with? . And I don’t think it’s very obvious right now.
[00:24:00] Mike Kaput: What about the flip side of this with the vendors themselves? I mean, this market went from us seeing a couple, you know, big giants raising a ton of. And having really unique products to suddenly, oh my gosh, like there’s so many more competitors.
[00:24:16] Mike Kaput: There’s a lack of differentiation. What should the vendors themselves be thinking about?
[00:24:21] Paul Roetzer: I think they’ve been losing sleep over this since chat BT came out, honestly. There was, there was toward the end, there was a really succinct kind of conclusion that I thought was very, well said. And I, I’ll just kind of read this.
[00:24:36] Paul Roetzer: So they ask where, where will the value accrue? And this gets into like the vendors. So as there don’t appear today to be any systematic modes and generative AI as a first order of approximation. Applications lack strong product differentiation because they use similar models, aka a, they’re all building on top of CO here or OpenAI or you know, other language models.
[00:24:54] Paul Roetzer: Maybe it’s a mix. Maybe they’re developing their own, but generally they’re using the same infrastructure, applications, lack, Product differentiation because of similar models. Models face unclear long-term differentiation because they’re trained on similar data sets with similar architectures. Meaning, oh here, OpenAI.
[00:25:12] Paul Roetzer: The language model companies, they’re all largely training on the same data sets. They need proprietary data sets to really differentiate the models and give them, you know, power that they don’t have otherwise. They’re all kind of commoditized. Cloud providers lack deep technical differentiation because they run the same GPUs, which are hard to get.
[00:25:29] Paul Roetzer: So there’s a scarcity of supply in that. And even the hardware companies manufacture their chips at the same fabs. . . It’s just like, this is why there’s uncertainty. So the next one Ithought was brilliant. There are of course the standard moats scale, moats, quote I have or can raise more money than you.
[00:25:46] Paul Roetzer: That’s, you know, the place. Some are making supply chain moats. I have the GPUs. You don’t. Ecosystem moats. Everyone uses my software already. That could be like a HubSpot play. Like I, Ibuild these tools and I already have 167,000 customers, so we’re going to win, distribution moats. I already have a sales team and more customers than you.
[00:26:08] Paul Roetzer: Data pipeline moats. I’ve crawled more of the internet than you, but none of these moats tend to be durable over the long term, and it’s too early to tell if strong, direct, network effects are taking hold in any layer of the. So that to me is like, again, if I was building c r m writer and HubSpot just launched chats spott, I’m done.
[00:26:29] Paul Roetzer: Like, I had the vision, I had the idea, and that’s irrelevant because they’re, they have a bigger moat than me and a couple of key areas and I didn’t get there first. And sorry, like, you know, good idea. Go on to your next one. So that’s kind of the space we’re in is this like massive uncertainty around what is going to happen.
[00:26:49] Paul Roetzer: In this place, and again, the infrastructure vendors seem like the logical play, but that’s not going to help you as a marketer, . It doesn’t get you anywhere. So
[00:26:59] Mike Kaput: before we move on to our third main topic today, I want to end by talking a bit more. I love that quote in the, in the article about the potential market size, where they say it’s somewhere between all software and all human endeavors.
[00:27:13] Mike Kaput: So if some in our audience might be a little confused by that statement, it makes generative AI sound a lot. Than just writing and art tools. And we’ve talked about this. Can you kind of outline here why generative AI is so much more than just AI assisted writing?
[00:27:30] Paul Roetzer: Yeah, we talked in a previous episode about that there won’t be software companies that aren’t AI powered like in three to five years.
[00:27:37] Paul Roetzer: Like my, my belief, like if we look at the marketing technology landscape, you know, Scott Brinker has that MarTech landscape. It’s got like 9,700 companies on it. The, any company that isn’t infused with AI within three years is done like that is, and I’ve said this publicly to, SaaS companies, I’ve told venture capital firms this, that only invest in SaaS companies.
[00:27:58] Paul Roetzer: Like you have to audit your portfolio. . , if you only invest in SaaS, there has to be someone on that team that deeply understands AI and can build the AI roadmap for that company or else that investment is going to get burned. So I think. We are in a space where all software will become AI powered software.
[00:28:16] Paul Roetzer: Like there will be a day in the not too distant future where you don’t have a software product or company that isn’t an AI company. And Idon’t even think that’s debatable. Like I, Ithink it’s, it isvi, it’s been obvious to some people for years. I think it is now obvious to everyone. All the venture capital firms I’ve talked to, all the SaaS companies are talking.
[00:28:36] Paul Roetzer: G P T ChatGPT woke everyone up to the fact that this was real and it was happening right now. All of human endeavor. Kind of starts to go more in the realm of the world of bit stuff we talked about and knowledge work and creativity, certainly labor. We, the, I, we didn’t put this on the agenda for today.
[00:28:54] Paul Roetzer: Maybe we’ll put it for next week. This company called Fusion that just released, or no figure that just released their, humanoid robot, the version one. And they talked about every human on earth having. Their own robot, like one robot for every human on earth. And they’re talking about it in at-home care.
[00:29:13] Paul Roetzer: They’re talking about off world adventures. They’re talking about, labor and, and solving for job short shortages. Those robots, whether you want them to exist or not, ex mak style robots, are powered by ai. They’re made, they’re being made possible by these leaps in AI language, understanding, language generation, computer vision.
[00:29:34] Paul Roetzer: So yes, like it. As crazy as it sounds to say the potential size of this market is somewhere between all software and all human endeavors. It’s not actually crazy when you understand how this technology works and what it’s going to be capable of. So, Yeah. I mean, it’s kind of a funny way to end it. And that’s what I said, like I just, I loved the article.
[00:29:54] Paul Roetzer: I thought it was written with a little bit of humor. Yeah. But a lot of understandable facts, and I think it’s important for people to read it. If you’re building technology or buying technology, or leading a company or investing in technology, Ithink you have to read the article.
[00:30:11] Mike Kaput: It’s definitely good to pair that article with the next topic because it just became a lot easier, I think, for a lot of developers and companies to start building these types of applications because developers can now integrate ChatGPT, and Whisper OpenAI’s human level speech recognition system.
[00:30:30] Mike Kaput: Into apps and products through the company’s api. I, since December, OpenAI says it has reduced the cost of ChatGPT by 90% savings that a P I users will now receive when they use the api. So this makes it much easier and cheaper. To incorporate the capabilities of ChatGPT and whisper into different products, apps, business models.
[00:30:54] Mike Kaput: And it doesn’t just mean that every business can have their own instance of ChatGPT, it means they can actually use these capabilities to build innovative new products. So as part of the release, OpenAI provided a few examples and I just want to run through those quickly so people understand what’s possible.
[00:31:13] Mike Kaput: So one early example is Snap, the creator of Snapchat, they just introduced something called My ai, which is a customizable on platform chat bot that you can speak with like you would a friend or an online acquaintance. And that’s built on the ChatGPT A P I. Instacart is also using the ChatGPT API to pair ChatGPT with its own data.
[00:31:37] Mike Kaput: So customers can ask open-ended natural language questions. So instead of having a recipe in front of you and searching Instacart for what you want to buy from the store, you could literally just ask something like, how do I make great fish tacos? And you can start not only getting an answer, but also.
[00:31:53] Mike Kaput: Having Instacart’s products and recommendations come up as part of that. And last but not least, speak is an AI language learning app, and it’s apparently the fastest growing English app in South Korea, and they’re using the Whisper API to create an AI speaking companion. So Paul, what do you think of the effects?
[00:32:13] Mike Kaput: What do you think the effects will be of releasing these APIs, on the overall market, given everything we just talked about?
[00:32:21] Paul Roetzer: Even more rapid innovation. You know, we’ve been saying for months, like, you, you have no idea what’s coming. Like the pace of innovation is going to be hard to comprehend. It. It just accelerated 10 x like the, that’s basically what it did.
[00:32:33] Paul Roetzer: it reduced the cost 10 x to build things on ChatGPT. So for. Developers and tech companies that are building, it just got way cheaper to build what you are building, which means that as consumers or users of it, you are going to see rapid innovation in what you can now do with this technology. So now you may not personally care, like it might not make sense to you like that they dropped the 10 x, but give you a really practical example in our piloting AI for Marketers course, and in some of the talks I give, I think even our Intro to AI class I teach, we show OpenAI play.
[00:33:05] Paul Roetzer: Where we write a blog post, 440 words that we write in 40 seconds. Most of that was writing the prompt and that it cost one penny to do so. Previously, if you wanted to access the OpenAI APIs to generate something, it was 2 cents for roughly 750 words. Just simple math here. . For 2 cents, you can now write 7,500 words basically, so that that like, so rather than 1 750 word post for two pennies, I can now write 10, 750 word posts for two pennies.
[00:33:38] Paul Roetzer: So if I’m HubSpot and I’m enabling you to write content, it now costs me 10 times less to let you do that than it did a. That that’s the . That’s basically what it comes down to. And what’s going to happen is it’s going to force the rest of the market to make all of their models more affordable. And it’s going to rapidly in increase innovation.
[00:34:00] Paul Roetzer: And I’m like, give tangible examples of innovation verticalized solutions, like people building things very specific to your industry. If you’re in the legal profession or dental or whatever it is, insurance, it’s now 10 times cheaper and faster to build verticalized solutions for. So the tech you use, your cortex stack is going to get smarter.
[00:34:20] Paul Roetzer: People are going to be launching companies and tools left and right. It’s going to be hard to even fathom the amount of innovation that’ll be spurred by this technology getting cheaper and faster.
[00:34:33] Mike Kaput: So just to add to that, I saw this tweet a few days ago from Jim Fan, who is an AI researcher at Nvidia and is a great follow on Twitter.
[00:34:43] Mike Kaput: He said that now with the new ChatGPT pricing, it costs $4 and 30 cents to process as in process and entirely learn the entire Harry Potter. And I think he did, he’s done a couple other high profile examples where like, I think the Bible took $2. So people really do need to understand whether you’re building on top of this or not, you need to understand that there were a huge amount of critiques around chat J P T when it first came out, that that it was unsustainable from a cost perspective.
[00:35:17] Mike Kaput: And a lot of people in this space were pretty clearly saying, well, Give it a few months, I guarantee you the cost will drop, the innovation will scale up accordingly. And that’s exactly
[00:35:27] Paul Roetzer: what we’re seeing. Yep. And the models are, again, there, there’s more advanced models sitting behind the walls at these research labs.
[00:35:35] Paul Roetzer: So when you combine the cost of doing this with more advanced AI capabilities, it’s, it’s just going to be crazy like Imean, I’m, so for our March 30th AI for Writer Summit, I’m doing a state of AI and writing. I’m not even sure I can create that keynote until March 29th. , like I’ve been debating what am I even going to present?
[00:35:56] Paul Roetzer: And Ithink I’m going to like create the framework of it, but I’m going to have to finalize that thing like 24 hours before the event because it’s moving so quick.
[00:36:05] Mike Kaput: I hope our event team is not listening to next podcast. You
[00:36:08] Paul Roetzer: gimme the deck when .
[00:36:11] Mike Kaput: So before we jump into rapid fire, I want to end with asking you.
[00:36:15] Mike Kaput: how should companies, vendors, developers, how should we be thinking about building on top of these APIs? I mean it, what, what kind of considerations do we need to have, as we’re building our own products?
[00:36:29] Paul Roetzer: I, it goes back to the previous topic of what’s defensible, like what moat can you create around these things because the technology is there to build all kinds of fascinating tools.
[00:36:39] Paul Roetzer: It’s just a question of whether or not those tools will be around in six months or if Salesforce or HubSpot or Adobe or Oracle or you know, any other list of marketing, sales technology companies are just going to build it themselves because now it’s 10 times cheaper for them to build it too. And they have development teams and they have AI research labs.
[00:36:57] Paul Roetzer: So that to me, is it’s, it’s the billion or trillion dollar question is what gets built and what can you grow. And I think, in my opinion, I haven’t, I haven’t thought like deeply about this. Well, I’ve thought very deeply about this, but I haven’t really come to conclusions. I think distribution matters.
[00:37:16] Paul Roetzer: I think your existing audience size is critical. So if you’re HubSpot and you have 167,000 customers, that matters. And I think proprietary data matters to train unique models. And so distribution and data, with all else being equal, you know, we all have access to the same models. We all have access to the same infrastructure.
[00:37:37] Paul Roetzer: Basical. I think those two things matter, which means the big players may get bigger. . And plus they can buy up all the interesting tools. So it may just contin, like, there’s going to be, I think a, a period of explosion of tools and companies built in the generative AI space. And then I could see over time a rapid consolidation because those, a lot of those tools are just going to get churned out and obsoleted over and over and over again, like every few.
[00:38:08] Paul Roetzer: But the best ones are going to rise. The ones that are interesting are the ones that built a media company first. Going to our friends, Joe Pulitz and Robert Rose at Content Marketing Institute. Been preaching, you know, build a media company yourself. Yeah. That’s the play we did with the institute. It’s like our play was always, let’s build an audience of people that are interested in AI and want to learn about this stuff.
[00:38:27] Paul Roetzer: And then someday maybe, you know, we do something else with it. And so I think, I think distribution matters. I think you might see more software companies buying more media companies again to get access to those people. And I don’t know. I mean, I’ve, I’m not confident enough to make some bets. I’ve been making bets personally from, you know, from a stock standpoint, but I’ve been doing that for years.
[00:38:46] Paul Roetzer: Yeah. Like I’ve, I’ve believed that AI companies were undervalued in the market. I don’t think Wall Street understood ai and Ithink that actually is still the case. I think there’s still a lot of value in the market for AI companies. But yeah, I would, I would think long and hard about what is defensible before I start building.
[00:39:05] Paul Roetzer: All right,
[00:39:05] Mike Kaput: let’s jump into several rapid fire topics. Like we said, a lot has been going on in the last week, so we’ve got a lot to cover, so we’ll move fast. First up is the US and the European Union, had their first meeting as part of a joint AI research initiative. So this is a collaborative research partnership that’s meant to speed up AI development in both the US and Europe, as well as determine what Regulat.
[00:39:31] Mike Kaput: If any are needed. So in this first meeting, the US and EU teams gave the agencies all involved on their ends 60 days to compile guidance on how AI can be ethically applied in different areas. They focused on five areas to start. First is extreme weather and climate forecasting. Emergency response management, health and medicine improvements, electric grid optimization, and agriculture optimization.
[00:39:58] Mike Kaput: So Paul, I wanted to get your sense of what is the significance of a partnership like this, and does it mean that regulations are coming.
[00:40:05] Paul Roetzer: I don’t think it means regulations are coming. I think it’s important that they are collaborating. Again, the data is what makes things, valuable here, and to be able to share data and share learning is going to benefit humanity.
[00:40:19] Paul Roetzer: So overall, I think it’s good. I think it’s necessary. I don’t necessarily think it has any real implications to regulations being accelerated because of some of the obstacles we’ve talked about in
[00:40:32] Mike Kaput: previous episode. Gotcha. And kind of related to that, our second topic is this New York Times article that came out recently highlighting how US lawmakers are struggling to understand artificial intelligence, and that’s even despite some of them raising alarm over AI.
[00:40:49] Mike Kaput: Tools like ChatGPT. In this quote from the times they say, but even as lawmakers put a spotlight on the technology, few are taking action on it. No. Bill has been proposed to protect individuals or thwart the development of AI’s, potentially dangerous aspects, and they go on to quote, A congressman from California who said, the problem is that most lawmakers do not even know what AI is.
[00:41:13] Mike Kaput: And this is from Jay Uber nte, if I’m pronouncing that right, who is the only member of Congress with a master’s degree in ai. So Paul, you called this out on LinkedIn and highlighted what I thought were two really important points. The first being. You said, by the way, you can replace lawmakers with CEOs and we face the same issues in business.
[00:41:34] Mike Kaput: The lack of understanding begets a lack of urgency to pursue AI responsibly. You also, mentioned that Cade Mets a New York Times AI reporter and a friend of the institute confirmed for you on Twitter. That this is the first time that the New York Times is reporting that OpenAI. C CEO. Sam Altman actually did a demo of GPT-4 for Congress.
[00:41:59] Mike Kaput: So can you tell us a bit more about why this article and the topic caught your attention?
[00:42:04] Paul Roetzer: Yeah. It goes back to what we talked about in the previous one. Like we’re not any closer to legal regulations because nobody understands ai. . And that is the fundamental problem I have seen over and over again in business is you have CEOs and boards and venture capitalists driving some of this who don’t understand ai.
[00:42:21] Paul Roetzer: And so we’ve said over and over and over again that understanding is critical. We have to start there like, Trying chat Q B T and finding ai. Writing use cases is fantastic, but leadership has to actually understand the technology that they’re using because it’s going to be infused throughout your business.
[00:42:36] Paul Roetzer: This is not a simple marketing tool or two. It’s marketing, sales, service, ops, product, hr, finance, legal, it like it is going to be in every part of your business. So that was the first thing that jumped out to me is one. Shows we’re not getting any closer to any kind of legal regulations here. Governmental, guidelines.
[00:42:55] Paul Roetzer: Two, it is representative of where we are in business and three buried at the bottom of that article from, Ithink I said it was ca but it was Cecilia, Kang and, and Adam set to Reno. Was that Sam Altman demoed g PT four. Now, we knew Sam Altman had met with Congress that came out a few weeks ago, but I had not seen anything where they.
[00:43:15] Paul Roetzer: He did GPT-4, which is basically, Iassume then meaning that they are worried about what comes next and he was probably there trying to like say, Hey, here’s what it is. Here’s the guardrails we’re putting in place. We won’t release it. It makes, it actually puts in context some of the other stuff they’ve been doing that we felt was a PR move recently.
[00:43:33] Paul Roetzer: Yeah. About AGI I and stuff like that. I feel like maybe. There’s a lot of pressure on them from government to, to take some immediate actions to per prevent, misuse and, and damage in society. And so it feels like they’re playing a bit of a PR game and a bit of a product game here. Yeah. So yeah. I just thought that was, that was interesting.
[00:43:57] Paul Roetzer: It was overall a really good article and, and, a lot of things about it caught my a. So you alluded to our third
[00:44:05] Mike Kaput: topic here. Oh, I figured. Which is, this is really crazy and, you know, it’s going to sound futuristic to people, but I would, I would personally start taking it pretty seriously given what we’ve discussed, but a major startup exited stealth mode saying it is ready to build autonomous humanoid robots.
[00:44:22] Mike Kaput: So on Twitter, Entrepreneur, Brett Adcock announced that his new company called Figure was exiting stealth mode and they’re building this autonomous humanoid robot, called The first model is called the Figure zero one. And in the announcement he said, quote, today we’re unveiling the most important technology for the future of humankind.
[00:44:42] Mike Kaput: Advanced AI meets robotics figure is designed to solve what they see as this huge problem. So Adcock says that as a result of an aging population, our labor force is shrinking and he cites a stat that claims there will be 85 million job shortages by 2030. He goes on to say, if we want growth, we need more productivity, and this means more automation.
[00:45:06] Mike Kaput: So the solution is, figure believes robots built for a human environment will maximize our impact on labor shortages and improve lives. And kind of the core thesis here is that the physical world is already designed for human forms. So it makes humanoid robots kind of a natural choice to. The world.
[00:45:26] Mike Kaput: This isn’t a new goal or a new realization, but Adcock does think this time is different due to some enormous advances over the last 10 years in things like compute and G P U energy and power density, bipedal lo com, locomotion control and autonomy. He actually showed some mockups of the robot as well.
[00:45:45] Mike Kaput: It’s five feet, six inches tall. It’s a humanoid robot with a human form and a screen for a. And you know, lastly and importantly, the team they’ve assembled has some really heavy hitters. They’ve got people that used to work at Boston Dynamics, Tesla, apple, and Google Apps. What did you make of this company and kind of the overall mission and challenge they’re trying to
[00:46:09] Paul Roetzer: solve here?
[00:46:10] Paul Roetzer: I mean, it definitely comes across as like overly sci-fi at first. And if, if I didn’t look into like the people behind it, I would’ve kind of dismissed it a little bit. Yeah. But when you combine that, how much money they’ve raised, Iwas just trying to find, I think they’ve raised over a billion already.
[00:46:29] Paul Roetzer: Like, I mean, they, they’re backed by some major players. And the fact that this is the exact market it would appear that Elon Musk is planning to go after with Tesla bot, aka optimist. It certainly adds legitimacy. Legitimacy to some major players. Think that this is going to be a massive market. And I think that was kind of what I threw up when I put some commentary on LinkedIn about this was, they’re all going after the same obvious markets, the labor force, the consumer household, dealing with aging population exploration.
[00:47:00] Paul Roetzer: I’ve said before, I dunno if you said on this podcast, but I don’t think a human’s, the first thing on Mars. I’m, I’m pretty confident that optimist is intended to be the first thing that goes to Mars from a humanoid standpoint. So Ithink that. This is a space, there’s going to be a lot of speculative investments made into, robotics has been kind of quietly behind the scenes for most people developing like Boston Dynamics.
[00:47:24] Paul Roetzer: We see all the crazy robots doing the fun things on YouTube. . But this is real tech. The government’s been working on this stuff through the Defense Advanced Research Projects Agency at DARPA for decades. This isn’t new. And I think that’s for, for many people might be the first time they look at someone like, oh my gosh, they’re actually building this stuff from the movies.
[00:47:42] Paul Roetzer: Yes, they, they are. And they have been for decades. I think we might be, nearing the commercialization of that technology though in the next, you know, five to 10 years. So something to keep an eye on. It’s not going to, you know, you’re not be buying robots tomorrow to take care of your parents. Your aging parents.
[00:48:00] Paul Roetzer: It would certainly appear that within this decade we may, may, may see some humanoid robots walking around. Wow. Yeah. What a world. . . Yeah. I had to do with like, anyways, have a great weekend. Yeah, , right? Put that up on Friday. It was just so like, wow. This is
[00:48:15] Mike Kaput: deep as if you didn’t have enough to think about starting your week too.
[00:48:20] Mike Kaput: Yeah. All right. Our last topic for today is courtesy of Wired Magazine. So Wired is a major tech publication and they just published their own ground rules for how they will and won’t use generative AI moving forward. So they break down these rules by different types of generative ai. So for things like text generators, they say they will not publish stories with text generated by ai, except when the fact that a story is AI generated is the whole point of the story.
[00:48:51] Mike Kaput: So literally, no AI generated content outside of writing, like a think piece on, Hey, this article is written by chat C b T. They also will not publish text edited by ai. They say they may use AI to suggest headlines or texts for social media. They may use it to generate story ideas, and they may experiment with it as a research tool.
[00:49:12] Mike Kaput: For image generators, they say they will not publish AI generated images or video. They do say that this will be in effect at least until the legal issues around artists and photographer copyright is settled. So it’s possible they’ll
[00:49:26] Paul Roetzer: indefinitely ,
[00:49:28] Mike Kaput: right? It’s in theory they would be open to it, but yeah, indefinitely in practice, they specifically will not use AI generated images over overstock photography, stating that stock photography is an important source of income for photographers and they may use AI to generate idea.
[00:49:46] Mike Kaput: For images. So we’re seeing this more and more, companies and brands taking public positions on how they’re going to use AI tools. What do you make of this announcement and the guidelines themselves?
[00:50:00] Paul Roetzer: I love that they are being proactive. Ihope we see a lot more media companies do this exact same thing.
[00:50:08] Paul Roetzer: That’s just very clear. We’ve talked about it before. You don’t necessarily, you don’t want to be like reading every article and wondering, did AI write this? Like, what did you know? You don’t know. I, I’ve been a belief, like, I don’t think you’re going to have to disclaim it on every post, like stamping that this was an ai, or not an ai.
[00:50:50] Paul Roetzer: Like yeah, Iactually, you know, we’ve, we’ve been pretty clear to date on ours. We use it to do transcriptions and we’re, we’re transparent. Like we, we transcribe audio. I don’t have any problem with that. I think that is a very natural use case that you just otherwise wouldn’t want otherwise do. And I think it is user-centric.
[00:51:06] Paul Roetzer: It’s to give the transcript. We’re not going to sit there and type it. We use it for content summarization of transcripts. Again, very clear use case. I don’t, I don’t really have a problem with. But I know we’re going to get flooded with prep content. We talked about this in an episode or, or two ago about, you know, the need for more human content.
[00:51:25] Paul Roetzer: So Idon’t hate media companies and brands taking a stance that we’re not using AI generated content. Like, I, Ireally don’t. I I think that there may be a movement toward that. Honestly, the one about not published text edited by AI is very, Aggressive in my opinion. Yeah. Idon’t, I don’t get that one.
[00:51:43] Paul Roetzer: I would be interested to hear more perspective on that. because does that mean they, they won’t even use Grammarly or Microsoft Word that has AI doing N L P on your, like that’s really extreme in my opinion. . Idon’t, I wouldn’t agree with that one and nor would I endorse our company doing that. The image generation one, Irespect, Ithink.
[00:52:07] Paul Roetzer: There is a spectrum of how you feel about the impact AI is having on artists, whether they’re photographers or, you know, illustrators or designers or whatever, and whether or not their work is being stolen in the training of these image generation tools. . There is certainly an argument that it is.
[00:52:26] Paul Roetzer: I don’t think we’re going to have any legal precedent for a time, that at least that’s sustainable, like Supreme Court ruling kind of precedent. So I think that’s a very aggressive stance as well. I don’t have a problem with using AI generated images, especially if we’re clear that that’s what it is.
[00:52:44] Paul Roetzer: But I could certainly listen to an argument that we should consider that stance down the road. So I think that those key areas of like how are, we’re using it for writing, how are we’re using it for editing? How are we using it for image generation and video? They didn’t get into video, but that’s going to be something you’re going to have to address in 2023.
[00:53:00] Paul Roetzer: Are we going to use it for video generation? I’m guessing they will say. I do think that heads of marketing and businesses, sh should be taking a stand on this. And I think that to me is the main takeaway is it’s a really good model. You know, we released our responsibility, AI manifesto to try and accelerate the conversation around some of these issues in the industry.
[00:53:18] Paul Roetzer: And Ilike that they’re doing this, and I think people should start pondering what their stance should be on these.
[00:53:26] Mike Kaput: Yeah, and we should also mention it’s not just kind of a nice to have a stance. It’s kind of a need to have in the sense of actually figuring out how you’re going to treat these tools and avoid potential pitfalls down the line, whatever those look like to you.
[00:53:39] Mike Kaput: But also, like you mentioned, differentiating your brand. I mean, you can’t just sit here as a credible content strategy today and say, great chat. PT can write everything for us. It’s like they can do it for
[00:53:50] Paul Roetzer: anybody. Yeah. Yeah. I, Ireally. That’s why I’m interested to see what I say on March 30th for our writer’s song,
[00:53:57] Paul Roetzer: Idon’t know what I’m going to say. Like I really, there’s some issues that I haven’t personally resolved in my mind yet, like how I really feel about some of this stuff and, you know, taking a stance. I think so much of ours is presenting information in both sides and sometimes not having a very specific point of view, because in part it just hasn’t been formed yet.
[00:54:18] Paul Roetzer: It’s not like I’m trying to. Saying, this is what I believe. Right. And I think there’s a lot of unknowns right now in particular in writing and the impact it’s going to have. And it’s just a really fascinating time to be having these conversations. And
[00:54:33] Mike Kaput: with the amount of topics we’ve covered today, we have to keep having the conversations because man, there is a lot going on.
[00:54:40] Mike Kaput: So, you know, Paul, as always, thank you so much for breaking everything down for us and for the audience. I think it’s
[00:54:46] Paul Roetzer: incredibly. Yeah, and we appreciate everyone listening. We’ll be back, next week, so we’ll talk to you then. Thanks a lot.
[00:54:52] Paul Roetzer: Thanks for listening to the Marketing AI Show. If you like what you heard, you can subscribe on your favorite podcast app, and if you’re ready to continue your learning, head over to marketing ai institute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.
[00:55:14] Paul Roetzer: Until next time, stay curious and explore ai.