Loading summary
Transcript82 lines
- [00:00]
Paul Raitzer
So I choose to be optimistic. I choose to believe that the future can be abundant. It can be incredible. We can get time back, we can invent new career paths. We can go through a renaissance and creativity and entrepreneurship. I think all of those things are possible.
- [00:16]
Mike Caput
Welcome to the Artificial Intelligence show, the podcast that helps your business grow smarter by making AI approachable and actionable. My name is Paul Raitzer. I'm the founder and CEO of Marketing AI Institute and I'm your host. Each week I'm joined by my co host and Marketing AI Institute Chief Content Officer Mike Caput, as we break down all the AI news that matters and give you insights and perspectives that you can use to advance your company and your career. Join us as we accelerate AI literacy for all.
- [00:53]
Paul Raitzer
Welcome to episode 128 of the Artificial Intelligence Show. I'm your host Paul Raetzer, along with my co host as always, Mike Caput. This is a special episode, special edition of the Artificial Intelligence show podcast. It is our final episode of 2024. What we're doing here is it's not our normal weekly format where we go through main topics and rapid fire items. Instead, we are going to be answering 25 user submitted AI questions for 2025. And so that is the focus here for your regular listener for a few weeks. In what month are we in November? We talked about the fact we were going to do this episode. We had an online form from Google people could fill out and that's where the questions are going to come from. So Mike's actually going to walk us through a little bit of the methodology before we dive into the first question. We're going to move pretty quick through these. Some of these, I mean, they're honestly like incredible questions. I scanned through them this morning before we started recording. We're recording on December 17th and they're, they're, they're amazing questions that we could easily spend five to 10 minutes or more on each of them. But we're going to move like one to two minutes is the goal for each of these. So we'll try and get through as many as we can, hopefully in under an hour. Okay, so this episode is brought to us by the AI Mastery membership program. So this is, I've been talking about this recently on the podcast. This is our annual membership program that gives people access to exclusive educational opportunities, insights and experiences. It's. Mike and I largely run the programming, so we do a trends briefing every quarter. Kind of the 10 key things you need to know about that quarter. We do a quarterly gen AI Mastery series. We do demonstrations of AI technology and we do a quarterly ask me anything session. So every month there's at least one unique programming opportunity for members. In addition to ungated access to the on demand webinars, ungated access to blueprints, and then we're actually working on a bunch of plans for 2025 to add even more value to this membership program, including new courses, new certifications, new experiences. So as you're thinking about 2025, this is a great chance to get in. And we always think about everything in like a learning journey. And I think about intro piloting scaling as kind of like the main steps. But throughout that process and beyond, you always want to be pursuing this idea of mastery in your education and capabilities. And so that's our goal here, is like this ongoing program in addition to the podcast, kind of a more deeper dive into helping people really drive their own AI transformation in their careers and in their companies. So you can go to SmartRx AI and click on education and it's under there or SmartRx AI AI mastery. And we will put that link in the show notes. You can use Pod150 and that will get you $150 off of the membership. So again, that's Pod, Pod1fit150150. Okay, Mike, we. We've got a lot of, like I said, incredible questions, very thoughtful questions. We appreciate everybody who took the time to submit these questions. So I will turn it over to you. You can kind of explain to us the process we went through, how to use a little bit of AI in the process and we'll jump into the questions from there.
- [04:21]
Mike Caput
Sounds great, Paul. First up, I just want to. I'm not even saying this to be nice to everybody. Our audience just deserves, like, gold medal for the quality of these questions. Like, I was just blown away. It was really hard to curate them.
- [04:33]
Paul Raitzer
They're very thoughtful questions.
- [04:34]
Mike Caput
You're incredible. Yeah. And you'll hear in a second just how in depth some of these are. So what we did is over the last few weeks on the podcast, we solicited questions through just a quick Google form, had people submit anything on their mind related to AI. We got dozens and dozens of questions. So I went through and kind of manually reviewed them all, basically curated from, in my opinion, kind of the top 25 questions, and organized them thematically. You know, people have a lot of the same types of themes on their mind. So I wanted to also be sure, you know, we pick a question that's also representative of several others that people ask, try to cover as much ground as possible. What was cool is it was really helpful to use Notebook LM during the prep for this because we have what, 127 other podcast episodes at this point. So most of those are in a Notebook now in NotebookLM via YouTube video links. And I was able to quickly like query for many of these questions what we've talked about in the past. Kind of just pull some interesting threads that we've been hitting on, which I found to be really helpful in terms of prep. And then we kind of synthesized our answers and kind of talking points what we want to discuss and also just like kind of off the cuff conversation related to all of these questions. So that's kind of how we got to where we are. So Paul, if it's good with you, I'm just going to kind of dive in and start reading out some questions for you.
- [06:02]
Paul Raitzer
Yeah. And for context. So I mean, Mike sent me this link a few days ago and honestly I looked at it this morning before the co CEO webinar. So I just kind of went through and the way I actually approach this because generally speaking, when I do Q and A, whether it's a podcast or intro, I actually don't prep for it. Like I prefer to not even know what the questions are going to be. But for this one, knowing we were doing these 25, I thought, well, I'll at least go through and like read through these things. So I actually haven't read through Mike, your notes on what we previously talked about. I just went through and jotted a couple of bullet points and I, I thought that I would actually almost prefer to take like this fresh perspective on these things. So what I'm going to do, Mike, is that when you asked me the question, I'm going to kind of give this more off the cuff, fresh answer and then if there's things we've previously said about it on an episode, you know, feel free to add that stuff in it's good context.
- [06:55]
Mike Caput
Cool. All right. Yeah, I feel like you're in the hot seat now. You ready? All right, so question number one. I live somewhere where AI skills will not be taught in schools. I want to prepare my kids and also help guide them to the best possible future employment. Do you have any thoughts or tips?
- [07:18]
Paul Raitzer
So the couple of thoughts I have here is I've obviously talked on the show numerous times about my own kids. I have a 12 year old and an 11 year old and they go to a wonderful school. And I wouldn't say that School is proactively teaching them about AI, like they're open to it, like it's not outlawed for use in the school. And I think I've seen that in a lot of schools where they're allowing it within the confines of how individual teachers often determine that they're going to use it. And so even though you know this person's saying they live in a place where it's not going to be taught, even people like me who live in a place where you would assume it's going to be taught, it's still, you can't rely on the schools for this. And so I guess my point here is I'm trying to be very proactive with my kids to make them very aware of AI technology and very savvy with AI technology. Because the way I, I think about this is whatever their career path is. So they're in sixth and seventh grade, as they move into high school, as they move into college, like I've got a little time, they've got time. But I know some of these listeners may have kids who are in college or, you know, very soon going to be. And so you're starting to think about like career paths and things like that. And so my feeling is that they're going to be at an advantage because if I look forward to the future of work, I am very confident that AI is going to be infused into every profession. And so I want them to be confident with the technology that they will use in their career. And so I'm trying to experiment with them in fun ways like storytelling, image generation. We'll definitely play with like the video generation. Play around with voice mode on the way to school in the mornings, we'll use it. When they ask me a question about stuff I don't really know, I'll pull up advanced voice mode and we'll have a conversation about it. I'll let them ask follow on questions. So like I'm just trying to let it be a part of their life without allowing them to use it as a crutch to avoid learning. So I'm trying to teach responsible use and then depending on how old they are, would dictate how proactive I would be about their future and education, career choices. So like, know if they were seniors in high school and they were making college decision today, I might be a little bit more proactive about how I would guide them than I am when they're in sixth and seventh grade. Because I feel like it's not time yet to make any changes or anything like that or say, oh, you're not. That job's not going to exist when you grow up. Yeah, so I'm not doing that. So I guess my, my short answer to this would be just be proactive. Bring the knowledge to them. Take the responsibility yourself to try and engage them in the technology in a responsible way.
- [10:08]
Mike Caput
So question number two, somewhat related. What should universities be doing in 2025 to better prepare college students for AI? Or what do you see as the future for higher ed as it relates to emerging AI capabilities?
- [10:23]
Paul Raitzer
Yeah, this one I'm very adamant about. We have to teach the teachers. You know, you can put whatever policy guidelines you want in place in a univers university or even in a high school, but until the teachers themselves are enabled, you know, until they have a strong confidence level and understanding the technology themselves, what it's capable of, how students could be using it already, how the students could benefit from it, we have to teach the teachers. And, and I. The other thing that I think about here is like, you know, Mike, you and I both have done a ton of workshops and speaking. We've spent a lot of time with, you know, brands, enterprise leaders, and the organizations who are racing ahead have CEOs who are bought in. When the CEO and the C suite are presenting roadblocks to this or not encouraging responsible adoption, then those are the organizations that are falling behind. And so I just feel like the same is going to be true in education. We need the leaders, the provost, the deans, the directors. They've got to be involved in doing this and not trying to shy away from it or ignore it. They've got to embrace it.
- [11:34]
Mike Caput
All right, so question number three. People always ask about what will change about AI or what are the biggest changes from AI. But, and I think we've cited this interview before, Jeff Bezos once said in an interview, the most important question I'm never asked is, what hasn't changed? And he says in the context of Amazon, you know, people are always concerned about price and delivery, no matter how much everything else changes. So this listener wants to use the same logic on AI. What has AI not changed? Or what will AI not change?
- [12:09]
Paul Raitzer
Yeah, this is a challenging one. So I often think about what remains uniquely human and what isn't the AI going to be able to do or to simulate. And honestly, that list keeps getting shorter. You know, the things I thought a year ago were pretty uniquely human. I'm not so sure anymore. So when I think about this question, like, the first thing that comes to mind is human connection. Like that's not changing. Like, we still want to be around other people, we want to interact with them. I want to see people in the office. I want to meet our customers, I want to hear their stories. Like, that doesn't change when I think about the customer side. And again, I come back to, like, our specific business, just for context. People want to feel needed, valued and fulfilled in their careers, and they need education and training to do that. So, like, when I think about the future for us, I think about that's not going to change. It actually might become more important and harder to figure out, like, what is fulfilling for the human when the AI is doing more and more of the work. So. But how we deliver that information, how we, you know, help and delight people in providing that education and training, that changes because AI enables whole new ways to do that. And I'm constantly thinking about how can we improve the way we deliver this? But those fundamental needs aren't. And so that's, I don't know, like, I don't know how, like, universal that is to different industries, different business models. But human connection and the human need to feel needed, valued, fulfilled, like, those don't change. The tech just enables whole new ways to deliver that, I guess. But, yeah, it's hard when you look at human traits and skills, to know what doesn't the AI at least be able to simulate.
- [14:05]
Mike Caput
One trend that came out of some of our past episodes is, if I had put it in one word, broadly this idea of curiosity. Because, like, right now, at least if we had to bet, I would say we're still always going to have questions, we're still going to be asking questions, and maybe we can actually ask better ones or more complicated ones that AI can help us answer. Now, obviously, I don't know if that's like, a skill you're putting on your resume, but I suspect we'll still need to be asking really smart questions moving forward.
- [14:37]
Paul Raitzer
Definitely, yeah. The machine itself is not curious by nature. It just solves things you ask it to do, it does things you ask it to do, but it doesn't seek out things to do.
- [14:47]
Mike Caput
Yeah.
- [14:48]
Paul Raitzer
Or to learn.
- [14:49]
Mike Caput
All right, question number four. Is this race with all these AI companies to be first to market with new models, new features, new solutions, whether these things are ready or not. Is this all setting a dangerous precedent?
- [15:04]
Paul Raitzer
Well, a great timing on this question coming off of episode 127, when we talked about the craziest week of AI updates I can recall, yes, it is. Like, there's no real way around this we are in unchartered territory and none of the companies know where they're taking us. That is indisputable. People like, they don't understand the ramifications of what they're doing nor the potential, like long term implications if they succeed in building the AGI or the superintelligence they intend to build. So yeah, I mean, we don't, it's not clear. Like just last week we had a new research report on scheming and how these models are able to scheme, to manipulate humans. We know that they're highly persuasive, they're likely superhuman at persuasion, but we also know that these frontier model companies extract that capability. They try and keep them from being overly persuasive. We don't know what happens if and maybe when who you talk to. These things become aware of their own thoughts, like metacognition, like they're like a human, like I'm aware, I'm thinking right now. Machines we don't think are aware that they're thinking, but when they do, then they become self aware. And that's a really weird thing to start considering the ramifications of. I think Ilya Sutskeva talked about that concept in his recent talk that we covered on episode 127. And then there's this whole idea of once we build agents that can really do things and plan on their own and take actions, they likely become self improving. They develop the ability to just replicate themselves or to improve themselves at a very rapid pace. And we don't know what that means. So yeah, the reality is, is like we're racing really, really fast to build and release smarter models for competitive reasons, for financial reasons, for egotistical reasons. Like there's probably a lot of reasons why different companies are doing this, but safety and alignment is not moving as fast. And we are basically depending upon these companies to, to halt or slow down when they think it's gotten too dangerous. But I'm not convinced that will happen because anthropic is supposed to be, at least, you know, from a PR perspective, they position themselves as being the ones who are supposed to be championing this and yet they're putting out computer use before anybody else, which is as dangerous as anything right now. So yeah, it's a, it's a big area of concern and I hope more people moving into 2025 spend some brain power on, on that topic.
- [17:42]
Mike Caput
All right, we have a bunch of questions that would fall kind of under the banner of AI career in business advice. And first up, question number five in our 25 questions is what are your thoughts on learning code in the age of AI? If I'm trying to further my career in AI, what are the most valuable education or tools that I should work towards achieving or figuring out?
- [18:06]
Paul Raitzer
So I'll go back to the, the context of like if one of my kids was a senior in high school and they wanted to go into computer science, would I guide them not to I at this moment, I would not steer them away from that. So if my daughter's son said I'm going to go wherever and I'm going to, you know, I'm going to do code coding, I want to build software. I think in the near term people who can code are going to 10-100x their capacity, their capability to build things, it's just going to become that much better. And the people who know how to build will get more value from these tools than people who don't. That being said, there is definitely a clear direction among these companies to make coding the co the language of code. Human language, like you don't need code, you, you will just say what you want. Like imagine how we build GPTs. Like you just give it instructions and you build it. You'll be able to go into places like Replit and Microsoft and Google and you just as a non developer, a non technical person, you're gonna just be able to build apps and websites and products and companies with your words. And that's not far off in my opinion. So I think coders are still going to matter. I think they're going to be way, they're going to have superpowers. But I also think the average person is going to be able to do what coders, you used to need coders to do. And I'm not sure what that means to the future of that profession. But again, I would not currently steer someone away from pursuing that profession if that's what they chose.
- [19:50]
Mike Caput
Question number six. A lot of the value that AI brings is not always easily measured. How can we gather the ROI of the AI tech stack as we go into 2025? Like how can we get better at measuring ROI?
- [20:05]
Paul Raitzer
I don't think this is very different than just traditional software tools. You would buy things like that. I think anytime you're going to make an investment, you have a goal in mind of what we are going to achieve with this investment, this project, this campaign. You benchmark prior performance and then you measure whether or not it's improving. So you know, we always talk Mike, about like these pilot projects where we're gonna, let's just say podcasting. We're gonna go get an AI product for podcasting. It's gonna do these three things. Okay. Our goal is to increase say 30 efficiency. You know, let's just pick a number. Then we're gonna take time and benchmark. If we have existing time data, we'll, we'll use that. If not, we'll do it the way we've been doing it for a week or two, benchmark it, and then we will measure whether or not these tools improve over a 90 day period. If they don't achieve what we wanted them to do, create the value we wanted, we move on. If they do, we double down or we buy that annual license at that point. So I think you just have to approach these things with goals in mind and then know what the metrics are you're going to measure to tell you whether or not it's working.
- [21:14]
Mike Caput
All right, question number seven. What unintended consequences of widespread AI adoption should we be watching for and preparing to address?
- [21:24]
Paul Raitzer
I'm really glad we didn't make this one the last question, because this is like, it's like just shove it in the middle.
- [21:29]
Mike Caput
I structured the mood of these questions end on a positive note, let's say that.
- [21:34]
Paul Raitzer
So I'll, this is definitely one that we could spend the whole time talking about. So I'll be brief here. Near term, it's jobs, new jobs, jobs. Like that is like the, the main thing I'm worried about. I know that concern is not widely shared among economists and government leaders. At least they're not saying it publicly. If it is, I, I think they are misguided. I hope I'm wrong on this. I really do. I hope two years from now we look back and say, yeah, you were, you know, overstating it. It really wasn't that big of a deal. Increasingly, signs are pointing to it's going to become a problem in the next one to three years, that we are going to need fewer doing the work that we currently do now. There are always a bunch of industries who can't hire enough people. It's not going to be universal. I'm not saying like everywhere across the economy, this is going to be the case. Some industries need AI to fill gaps, I get it. But there's a lot of knowledge workers and a lot of them are going to be directly impacted by AI's capabilities in the next one to three years. Midterm. You know, we're starting to look at what happens with this next generation when they're raised having AI assistance on demand when they can just ask questions of Siri or, you know, Google Voice or OpenAI ChatGPT Voice. And three to seven years from now, like, what if they forget how to learn? Like, what if we don't teach this the right way? And kids, it does become a crutch and they never learned the fundamentals that we all had to learn coming up in our professions that enabled us to become, you know, domain experts or gain that experience, gain that knowledge and understanding. I worry that if we don't do this right, they, they won't learn those things. And then seven to 10 years out, now we're looking at societal change, government unstable instability, educational, massive educational system shifts, massive impacts on the economy, which I think can be mostly positive, but they're going to be an impact nonetheless. And so that's kind of near term jobs, midterm, you know, impact on students as they grow, and then long term macro, like everything. That's the stuff I think about.
- [23:42]
Mike Caput
All right, number eight, what is the most impactful piece of advice or literature that you came across on AI in 2024?
- [23:52]
Paul Raitzer
This is an interesting one. Like, I'll be interested to see if you have any that don't make my list. I know it said just a piece I had trouble really picking. And I could explain why each one of these was so impactful. But first was Intro to Large language models or LLMs by Andres Karpathy. So that YouTube video from January, it had a profound impact on me because he laid out what all the major AI research labs were working on and that validated a lot of what I thought was happening. But to hear him say it, it was like, okay, good. Like, I definitely feel. And that worked into like every keynote I gave after that and eventually informed part of the exposure key I built for Jobs GPT. The, the, you know, the custom GPT we created. The other two that I would mention is my last five years of work by Avital Balwit, who's the chief of staff at Anthropic for the CEO, and then situational awareness by Leopold Aschenbrenner. We'll put links to all these in. But we, we talked about all of these on the podcast and in many cases multiple times, we referenced all three of those. Is there any that jump out to you, Mike, as something that was impactful for us?
- [24:56]
Mike Caput
Well, all of those would definitely be on my list. I would just add to that in aggregate the work of Ethan Malik, just because I think the biggest piece of advice he has that recurs through his posts and you know, his engagement online is this idea of like, look, even the labs themselves don't know how to apply this stuff to your job. And it's very positive in a lot of ways because it's really easy to sit here and say, someone I'm sure knows more about this than I do. But he kind of says, look, you're supposed to go play, experiment. It's messy. Go figure it out for your own job, your own work. Which is something we talk about all the time as well.
- [25:34]
Paul Raitzer
Yep.
- [25:36]
Mike Caput
All right, number nine. So you know, we talked about this past year, Sam Altman's kind of quote about AI automating 95% of like marketing work, creative advertising agency work. So as I think about this as a marketer, someone asked and remaining indispensable. Like in your view, which uniquely human traits are most critical for marketers to upskill while they focus on building their AI skills and capabilities?
- [26:06]
Paul Raitzer
Yeah, so I, I kind of alluded to this one earlier. I, I do struggle with this. What remains uniquely human? Which traits should we really be pushing? I, I don't know that the 95% by 2027 is anything I would put any money on. Like the probability of us getting to that by 2027 is, is probably pretty low. It's not zero, but it's, it's low at least across like the general populace. Maybe in some industries it could be. So that being said, like some of the things I've highlighted in talks, I know my opening keynote at Macon, I talked about this is common sense is uniquely human. And, and that one I'm, I'm fairly confident probably stays curiosity. You had called out earlier experience. That's an interesting one because can that be simulated through just a bunch of training or simulations? Probably so I'm not sure. Imagination. So thinking of creative things to solve creative problems, to solve creative approaches. That's I mentioned on episode 127. I'm not so sure that the AI lab, specifically Google, don't think they have a way to do that. Like I think they think that they are unlocking the ability to give these machines imagination, create original ideas and thoughts. That's to be determined instinct, which comes from experience. A lot of times intuition, love, like actually caring about the people you're creating solutions for that is uniquely human. I don't think machines are going to have that. And then self awareness we sort of talked about. So I mean at the end of the day, like what I think the future of work is going to Be is telling the AI what to predict, what to create, what to plan, and then knowing what to do with it. So an example here is like the O reasoning model from OpenAI. It's now able to go through chain of thought and do this reasoning, but it doesn't do anything. It sits there as a blank screen until you think of what to ask it to solve. And so the human has to come up with the things to solve and, you know, develop a creative plan around that. And then once you get the output, you, you have to decide if it's any good. And so those are the things I would really focus on is like logic, like logical thinking, strategic planning. Again, there's like common sense intuition, Like, I don't know how you teach those traits per se, other than going through experience and trial and error in a lot of, a lot of ways. But for us to judge the output of the machine, we, we have to have those things. So I don't know. I mean, the people I think who excel in those areas stand to do really well. If, if all you do in your career is like the repetitive tasks that they could take someone tomorrow and plug them in and they could do the same thing you're doing, that's, that's not going to bode well. You have to have these unique abilities that isn't going to be easily replaced with some smarter version of ChatGPT.
- [29:11]
Mike Caput
All right, question number 10. It already feels impossible to keep up with the pace of AI changes no matter how much the technology improves. But the number of AI tools and companies, all the options, all this constant leapfrogging, it's impossible to keep up with this person asked as general counsel, I'm on the board with AI for my company, so is our president. But the pace it progresses at does not align with what seems like a reasonable pace to scale the company and to get the employees to embrace this. For example, just when the president agrees to get an enterprise account for a handful of employees, ChatGPT announces an expensive pro plan where a competing AI tool announces an advancement that makes it seem like that tool is better for our company. How does a company like ours, that he says they're a construction company, make decisions about the use of AI tools so that we're emerging urgent, but we avoid all this waste. We don't want to sit back and wait, but we also don't want to start down the wrong path.
- [30:12]
Paul Raitzer
Yeah, I will just say, first of all, no, you are not alone, that as you're reading this question, I would imagine a lot of our listeners might be thinking to themselves, well, I could have sent that question. Like the same problem. Yeah, Mike and I feel that. And we monitor this space for a living. Like just this past week, as I felt like I should have tested the video generation model more, I should have had a chance to go in and do this. And like, I just feel like AI world just like moved past me in the last three days because I had other stuff I had to do.
- [30:42]
Mike Caput
Yeah.
- [30:43]
Paul Raitzer
So a lot of people feel that same bit of overwhelm and helplessness in some ways. So the first thing I would say is have trusted voices. Hopefully Mike and I function in that role for many of you. You know, just knowing once a week we're going to try and unpack the things you should really care about. And if you miss the news on your own for a week, hopefully, you know, you feel like we can catch you up. And so that's, you know, whether it's us or somebody else, like, have the people you trust as your filters who actually are living and breathing this stuff every day and can, you know, help you see it through. The other thing I would say is, like, don't worry about trying to solve all this. Don't worry about always having the latest thing. Honestly, like, if people just took chatgpt4o or gemini 1.5 or however it may be, and you just went hard on that platform and you've just maximized the value and personalized the use cases and taught people how to do it and enabled them with these tools, and you ignored AI updates for the next 12 months, you're going to come out ahead of most of your peers. Like, I think just picking a platform and going and training people on it and building, you know, custom GPT or whatever it is that just enables people to get value, get them, you know, 1, 2, 3, use cases they use every day, that's going to be enough. Like, it's going to get you ahead. And then if you can, depending on your structure, company, create a little AI lab or a center of excellence or whatever and have like two or three or five people who want to be a part of being out on the frontier, let them test this stuff. You know, join our AI Mastery membership program and like, just follow along with what we're experimenting with, however you want to do it, Let them worry about that. But until they've built a business case and they can pilot this with a small group of people, don't even think about the new stuff. Just like, let them handle that and then they'll Bring it to you when they think they have a proven business case. So I don't know, Mike, you, you obviously deal with this all the time too. Do you have any other thoughts on that?
- [32:40]
Mike Caput
Yeah, I liked what you said about just go hard on one platform. Because I can guarantee you, if you actually do that and you get to the end of whatever that pilot looks like or that experimentation period and you say, shoot, we're using the old technology, we're ready to advance, you've probably saved so much time and money already that it will not be hard to get a little extra money for the next tool or for a few tools. So that's the key.
- [33:05]
Paul Raitzer
Yeah, and I think the other thing that comes to mind is like Sam Altman talking to startup founders and saying, like, don't build something. We're going to just like obsolete with our next model. If you build around today's capabilities and adopt it within your company, it's only going to get better when they improve the model. That's what I'm saying. Like if you just bet on Gemini or Chat GPT or Co Pilot or Agent Force or like whatever your platform of choice is, whatever your company of choice is, and you just invest in maximizing the value you can extract from that. And then if someone introduces a new model that then is put into Salesforce or judging or whatever, and it gets better at reasoning, it gets better at, you know, content creation, creative. Right, whatever. Great. It's not going to change your plan and adoption. It just may like level it up a bit. So I don't think you can go wrong with, with making a bet on a platform that you're confident in.
- [34:03]
Mike Caput
All right, number 11, what is your advice for AI enthusiasts who work in industries that are traditionally slower to adopt new technology and where leadership doesn't recognize the urgency we should be placing on AI adoption?
- [34:17]
Paul Raitzer
So I'm just going to hit this one fast with like three things. Each one of them again, we could unpack, you know, another time. But if your company isn't embracing it or even worse, isn't allowing it, like, you cannot use chat CBT within your company. You can't test custom GPT on your job, so you have no idea what we're talking about, then invest in your own personal life. Go get a paid $20 a month account, Go build a GPT to help you plan a trip to help you talk to your kids and Gen Z slang, like, whatever, whatever it is. Like, just go in your personal life and experiment. The second thing is push for change internally. Like with actionable and achievable plans and viable pilot projects. Like, bring ideas to the table that solve business problems. Not, hey, I want to try some AI. Let's go get Sora. It's like, no, nobody needs Sora right now in the company. Like, what do you actually need? We need to, like, figure out churn or we need to generate more leads or we need to be more efficient. Whatever it is, push for that change. And if all else fails and nobody wants your voice involved and nobody wants to listen to the opportunities here, then go find another job. Like, look in your industry and say, who is moving forward? Like, because I will tell you this as hard, and I trust me, like, I am. I am very lucky in my life that I run companies and I have for the last 20 years and I've never had to go look for a job, but I know plenty of friends and family and peers in my industry who have, and I understand how hard that process is. All I'm going to say here is, like, if you're in a company and they're not going to do this, and three years from now you're going to be doing the same thing you're doing using the same tech you're doing, you will have been obsoleted in your profession. So I just don't feel like people should be okay sitting still. If they personally feel that they have to do this, then. Then you have to find a company that is embracing this and go be a part of that. Because the payback for you if you spend the next three years figuring this out versus the next three years sitting still is going to be tremendous. And I think that eventually it's going to start to feel very, very helpless. For people who are in companies that aren't adopting AI, you're going to feel like you are now being left behind. That is not the case yet. You have not been left behind. If you're in that kind of role, it's still early, we still have time. But three years from now, I don't think we'll be saying the same thing.
- [36:52]
Mike Caput
Okay, we have a bunch of questions here about AI tools and use. Cases number 12. What is your favorite AI tool for work and why?
- [37:01]
Paul Raitzer
All right, so I'm definitely going to let you throw in on this one, Mike. I have a hard time with this one, but I have to say ChatGPT, because it is the thing I dominantly use. Like, if I stacked all my AI uses up, I would say like 80% of them is ChatGPT, specifically the co CEO, personal custom GPT I built. You can go check that out on SmartRx AI. We just launched that GPT in a webinar around it. I am though very, very intrigued by the voice and video capabilities that are now in ChatGPT as well. Perplexity I'm a huge fan of I do not use it nearly as much as I did earlier this year. I use ChatGPT more than I am using Perplexity and then the new ones, Google Deep Research and notebooklm. I'm very bullish on those. So chatgpt if I have to pick one is my answer. But I I'm a big fan of the others.
- [37:49]
Mike Caput
Yeah, I mean Chad, GPT obviously just has to be top of the list for me. I would just drill down even further into if you are not trying to push Advanced voice mode as far as it can go in areas of your life, I'd recommend it can be kind of a hack to do so. It's been the biggest thing for me lately. Maybe it's just how I prefer to work sometimes, but honestly like if I'm on a long drive or have like dead time, you know where otherwise I'd be just like listening to a podcast. I have gotten incredible amounts of work done of problem solved just using Advanced voice mode.
- [38:24]
Paul Raitzer
Agreed. Yeah, I don't use it enough, but when I have it is awesome and long drives are great example of that. And I think even just like the quick drives into the office, I'm starting to do it more and more.
- [38:34]
Mike Caput
All right, number 13, can you share an automation or workflow that's been the most effective one that you've implemented for the team, for yourself?
- [38:43]
Paul Raitzer
Yeah, the one I always come back to just because it's relevant to everyone listening to. This is our podcast that is probably like the campaign that we have done the best job of continually optimizing the use of AI within it. Our main platform is descript. That is the predominant tool we use to do it on episode 55, which we'll put in the show notes, Mike and I actually walked through that workflow. It's definitely evolved since then, but the same steps are there. The tools we're using have probably changed a little bit, but you can go hear a lot about that. So podcasts would be the number one for me. Do you have one mic different from the podcast?
- [39:21]
Mike Caput
Mine would just broadly be custom GPTs for everything for your role for different domain areas. Just building these almost mini workflows with custom GPTs has been transformative.
- [39:32]
Paul Raitzer
Yeah. And if you're not a ChatGPT user. You can do a similar thing with Google Gems. Hopefully they're going to keep improving Gems next year and offer more capabilities within them. And then anthropics Claude projects, I believe. I haven't built one there, but I think that functions in a similar way as well.
- [39:49]
Mike Caput
All right, question 14. What would you consider to be the most impactful or beneficial AI use case that you've seen or heard of in and out, in or out of marketing?
- [40:00]
Paul Raitzer
Yeah, I'm actually going to piggyback off of your previous answer. Custom GPTs, like our hands down. Like, you know, I, I think Notebook LM took off for a lot of people because they saw practical use case, I think a mass market or heading toward mass market. Actually, just this morning or yesterday, I saw that Google actually put Notebook LM on the home search page, which I did. That is holy ground at Google. So like to inject a project there, you know, that they, they believe deeply and they're seeing strong metrics. So I think the key is to make something so tangible and simple, to understand the value. And so I think that's what NotebookLM did and I think that's what custom GPTs do. It's like if you say, oh, you're an email marketer or you're an accountant, or you're an HR professional, or you're a CEO. Let me build a custom GPT that helps you do the things you do every day and they open it up and it just helps and it just works. That is how you create value. That is how you get adoption within enterprises. So to me, I agree, like, if you personalize GPTs for people in their role, that is the fastest way to have success when you're looking at adopting AI in any company.
- [41:16]
Mike Caput
All right, question number 15. We've got several questions here about our favorite topic of the day, AI agents. With the rise of AI agents that might be able to autonomously execute complex tasks and strategies, how do you see the role of marketers evolving in the next few years? What skills will become indispensable? How can organizations or employees prepare for this shift?
- [41:40]
Paul Raitzer
So, yeah, I know we have like a couple questions about agents, so I think it's good to just, you know, I'll give it specific to, you know, marketers as an example here, but we'll talk more, more broadly as well. So just for context, AI Agent we're currently defining as an AI system that can take actions to achieve a goal. So ChatGPT doesn't take actions. ChatGPT outputs something. It does it creates a plan, it writes an email, whatever. It doesn't go do something. It doesn't fill out forms, it doesn't complete workflows, it doesn't do those things. So what we're talking about is systems that can take actions. Agents are early. They are not autonomous, despite what you may hear. Um, we talk about this in episode 124. I think it was Mike, kind of, what is an AI agent? I went through this so you can go back and listen for some more context. But if we think about what goes into an agent, there's five fundamental steps I like to think about. So there's goal setting, telling it what it's going to do, what is it supposed to achieve. There's planning, so there's going through, setting up of this, the integration of the data, things like that. There's the executing part of doing the actions, going through the 10 steps, 20 steps. That is the part that can be autonomous in, in a way not necessarily fully, but that, that executing is the part that we're hearing about being autonomous. Then there's the improving the agent, and then there's the analyzing the agent, the performance. So to, to like, don't just listen to me, I'll do. Jensen Huang, the CEO of Nvidia, this is what he said in a presentation in November. The he likes to think of them as AI workers, like digital workers. These AI workers can understand, they can plan, and they can take action. We call them AI agents, he said, and just like digital employees, you have to train them, you have to create the data to welcome them to your company, you have to teach them about your company. You train them for particular skills, you evaluate them. After you're done training, you guardrail them to make sure that they perform the job they asked, they're asked to do. And of course you operate them, you deploy them. So whether it's a marketer or any other profession, I don't care what you do. The humans are needed to set the goals, plan and design the agents, connect the data sources, integrate the supporting applications and tools, oversee execution, provide the inputs and iterate and improve them, and analyze the performance so that like, if you say, what is a marketer going to do? What is anybody going to do? All of those things, like the agents we're going to have in 2025 aren't doing those things for themselves. They, they execute the actions in the middle part, they'll increasingly do some of the other things. But right now, every profession is going to still need to do all that work.
- [44:24]
Mike Caput
All Right. So this other question about agents kind of piggybacks on that. Like my question is how, this is number 16. How do you build agents? Like do you do it directly in chat GPT? Is there a particular platform or app that we would recommend or think about?
- [44:40]
Paul Raitzer
Yeah, so this is great follow on question because again this perception exists that these things just magically show up and start doing these things. That is not what happens. So if you're in Salesforce you would use their agents force platform, you would go through and define the role, set up the data, do all those things. I just walked through. If you're in Microsoft, they have a copilot studio where you go through and build agents. If you're a Google user, they have Vertex AI where if you're a more advanced person you can go build AI agents in there. It's not Google. Vertex is not for the non technical audience quite yet. They might go there, but you can go do it. Agent AI is another example where people, right now it's Dharmash Shah and his team that are building the agents for you, but they're going to enable people to build them. So there's, there are platforms that are starting to allow you to build agents and set the rules for them. But that's, that's, it's kind of like a platform by platform thing. And then I would imagine you could probably do like hugging face like if, if you're more technical, if you're a developer, there's probably a lot of other places you could do this.
- [45:44]
Mike Caput
All right, we have quite a few questions about AI and its impact on things like agencies or professional services. Question number 17. With AI potentially reducing the time spent on content creation and strategy, how should agencies think about reinvesting the time they spend on those things to drive greater value for clients?
- [46:06]
Paul Raitzer
Yeah. So if, if you have the opportunity, I think in the agency setting, professional services, education, training and change management are going to be, you know, a big part of where it's at. You know, you have to think about where is this going. So you might expand your services to be things like building agents. Like that's a very logical thing that people would turn to an agency for if they trust them already. And if like let's say you're helping manage Salesforce or HubSpot for somebody, it would be a very natural extension of that to offer agent building. Now it's going to obsolete some of your services, but who cares? Like the idea is that you're going to keep having new capabilities emerge because of these AI models that you're going to be able to build services on top of. So that's, that's kind of look. I would look at these immediate service areas like agent building, but I would also think deeply about education, training and workshops.
- [46:56]
Mike Caput
Number 18, any specific recommendations for language that agencies can use or should use in their client contracts regarding the use of AI.
- [47:08]
Paul Raitzer
You can play around with ChatGPT or Gemini or whatever and draft this stuff. I've had some success building policies and guidelines within those tools. I'm not a lawyer though, so I would always turn that over to my lawyer and have them review it, which is really my main thing is talk to your attorneys also. You know, again, I ran an agency for 16 years. Sometimes we signed our contracts, sometimes we signed theirs. More often than not, it was probably theirs that we were signing. And what I have seen from the brand side is language not allowing agencies to use Gen AI, where it's explicitly saying unless you get our permission, you are not to create anything with Gen AI. So that you need to consider. And then the other thing I would just think about from an agency perspective is, you know, pricing and fee models because you know those are going to have to change.
- [48:01]
Mike Caput
I'm glad you mentioned that because in a couple questions you're going to get to answer that question too.
- [48:07]
Paul Raitzer
That's a very popular question. It really is.
- [48:10]
Mike Caput
But first up, number 19. So this person says that they're suspecting AI tools will basically, you know, kind of come for what junior associates at their firm or any firm are doing. So think agencies, consulting, legal firms, et cetera. But we used to live in a reality when junior associates became senior associates, then managers, then principals, then partners. Where are professional service firms going to be getting their partners in an AI world?
- [48:39]
Paul Raitzer
So I actually got this exact question from a law firm. I was meeting with a bunch of partners and when I explained, went through state of AI where it's going, somebody said, oh, so we're not even going to need to hire associates? And I said, well, that's one way to think about it, but who are your future partners? And they just stared at me and I was like, that's on you. Like, I don't know, you probably all want to retire at some point and nobody can be left. So the answer is like, we don't know. Like, if I think about professional service firms, and again, having run an agency for 16 years before I sold it, I just think there's going to be fewer humans needed to do it. And that doesn't mean that we won't still hire people, it just means we won't need to hire as many people. So if I was like, if I was going to be done doing what we're doing and I was going to go start an agency right now, I would do it under the assumption that with 2025 models, this is even really looking into 2026 and beyond, yet we would need one employee for every three to four hires of what we would usually have done. So if I go back to like 2015 and I'm thinking about our production and the client demand and things like that, I think you could seriously build an agency of 20 people that does the work of what previously would have been a 60 to 80 person firm. And I honestly don't even think that's an exaggeration. That might actually be a conservative estimate. And you could apply that to client side as well. Like when you're thinking about your marketing team, your sales team, your customer service support team, I think we're very quickly, in the next one to two years entering a phase where one human can do the work of three to four humans, depending on what the profession is. And that's a very weird environment. And if, if so, if there's increased demand for your product and services, you're going to just keep hiring people and you're going to do more great. Like that company doesn't need to lose people. But if there isn't increasing demand or even worse, there is shrinking demand for your services, that company will reduce their workforce. And I don't, I've said this on the past, I don't understand the argument against that assumption. Like, I really don't comprehend how that isn't what happens. So yeah, I think that's how I would, I would look at the service firm side of it is I think fewer people are going to do the work and hopefully there's demand. You can still hire, but I don't know, you're going to need as many people.
- [51:10]
Mike Caput
All right, last question on agencies. Question 20. Do you think AI will change the typical agency price pricing model? If so, how should agencies adapt?
- [51:21]
Paul Raitzer
Okay, so I have, I've said this before. My first book in 2011, the Marketing Agency Blueprint Chapter 1 was Eliminate Billable hours. If you are a firm that is still charging billable hours 13 years later, yes, your model is cooked. Like you cannot charge unless it's just straight up advisory consulting work where they're just paying for your knowledge. And you say it's going to cost X per hour for you to talk to me. That's A different story probably. But if we're charging to produce an output and you're charging by the hour and they know you're using Gen AI to do it, there is no way you're going to be able to make the same amount of money doing what you're doing. So you're going to need a lot more work in the pipeline because you're going to have to be doing it for less. Unless you're on a value based model, whatever that looks like to you, some project fee, a set fee or you know, output or performance driven, however it is you need to do it. I would go listen to the episode 127 combo. We just had Mike yesterday, I guess this whenever it was this morning about SaaS pricing and I think the same principles are going to apply.
- [52:30]
Mike Caput
All right, our last handful of questions here are broadly under the category I'm calling what 2025 and or the future holds. Question number 21 as you consider the growth of AI and its impact on business and society and you think about the opportunities and challenges that 2025 may bring, what one opportunity and one challenge excite you the most or fill you with dread and why?
- [52:58]
Paul Raitzer
So the dread would be already said. It's jobs. I obviously have concerns about jobs in the economy. Thing I'm hopeful for the opportunity is that explosion of entrepreneurship. It's getting time back in our lives and it's to see what career paths people create. Because I do think like one to two years out we're going to have this impact on jobs, but we're also going to have these really cool new careers people are creating around AI. Like things you and I probably can't even sit here and conceive of yet. Right. And so I think that's exciting. But on the entrepreneurship side, like there's never been a better time to build a business from the start because you just don't need as many people. AI is going to help you do a lot of things. You won't need as much advisory work outside. You're not going to pay all the legal fees and accounting fees and things like that because you're going to be able to do a lot of that for yourself. You're still going to need those experts, just not as much. So I think building a business from the start with fewer people is an amazing opportunity. I think having to transform an existing business is going to be painful. Like I. For the reasons I already outlined. If you need fewer people, you're going to have to go through a bit of pain to get There.
- [54:08]
Mike Caput
All right, question 22 is a bit related. How do you stay optimistic about the future of humanity, especially the world of work, when the potential near term, potentially advent of artificial general intelligence threatens to disrupt work like never before?
- [54:24]
Paul Raitzer
It's definitely a choice. Like, it's a mindset. Like, if I was dreading it every day and I was negative about it, and I listened to all the negative voices in the AI industry who are like, trying to like, counter the optimism, then I would just, I wouldn't even do what we're doing, Mike. Like, if we, if we had to think about this every day and I just assumed it was going to go wrong. Yeah, I would. Like, what good would it do? So I choose to be optimistic. I choose to believe that the future can be abundant. It can be incredible. We can get time back, we can invent new, you know, career paths, we can go through a renaissance and creativity and entrepreneurship. Like, I think all of those things are possible. I am very much a realist, though. I know there's going to be bumps in the road. I know it's going to be hard at times. I know there's going to be pain in different sectors and different industries more than others. But I think net, it's going to be a positive for society and humanity.
- [55:19]
Mike Caput
Question number 23. What's an industry where you think AI is currently underutilized and an industry where you as a business owner and, or as a dad are personally the most excited to see AI adopted in earnest?
- [55:33]
Paul Raitzer
Yeah, so every industry, it's underutilized. Let's be straight on that. That is the whole purpose of Smarter X. The X stands for, like, pick an industry, pick a business model, whatever, you can build a smarter version of X, anything. And so take your business, you know, whatever it is, manufacturing, retail, e commerce, law firm, accounting firm, whatever, you can build a smarter version of that business. And so I'm excited to see what people create in different industries. But if I had to pick, like, personally, it's education, like, that's, you know, I'm balanced. They're like, I, I'm fearful of what it looks like. I don't know what college looks like when my kids are going to be there in five and six years. Like, I, I really don't know that it's going to be able to look like it currently does. So I worry about that. But I also think the ability to personalize learning, you know, bring learning to people who otherwise wouldn't have access to it, democratize educations and give everyone like a personalized Learning assistant, like, could be incredible, but, you know, we got to be intentional about getting there.
- [56:37]
Mike Caput
Question number 24. If you could have one wish granted for the future of AI in 2025, whether that's a breakthrough, a trend, a shift in how AI is used, what would it be and why?
- [56:49]
Paul Raitzer
So I'm going to steer out of the technology side of this one for a moment and say what I really want is in our Scaling AI course series, I taught a course on how to do AI impact assessments. And I think those are really, really important. I think that leaders in different companies, different industries need to proactively look out at the models one to two years out, because we can reasonably project the things they're going to be capable of doing one to two years out and say, how does that change the jobs in our company? How does that change our industry? How does it change our partner ecosystem and be proactive in planning for that change. Nobody's doing that yet. Like, I don't know of a company that is doing that. And so that's what I want is like proactively looking at the impact and proactively planning to do this responsibly. On the technology side, the things, I don't know if I'm like looking forward to them. I'm just going to tell you what's coming. Reasoning is going to get a lot better. So we'll go from 01 to 02 or whatever, oh, 1.5 or whatever OpenAI is going to call it. Gemini 2.0 is going to have reasoning baked in. They're all going to have reasoning. Memory is going to become much more significant and that's going to be impactful vision. As we've talked about built in Apple Intelligence has some elements of vision. Now you're going to, you know, Project Astro from Google has this vision, like to be able to see and understand the world around you. And you can test it right now in voice mode in OpenAI and chat GPT. You can go play with this ability to see the world around you. Voice interface is going to be huge and AI agents are going to keep getting better. So like those five areas, reasoning, memory, vision, voice, action, that's gonna happen next year in all these models and a lot of cool things are gonna come as a result of that and a lot of more uncertainty is gonna come.
- [58:40]
Mike Caput
Yeah, just piggybacking on that. I would love to just see truly useful, almost unlimited context Windows plus memory. And we've talked about how those factors can come together. Because if I could just have an Always on assistant that knew truly everything I chose to reveal about myself and all the content and context and books and information I've ever read, that would be pretty incredible.
- [59:07]
Paul Raitzer
Yep, agreed.
- [59:09]
Mike Caput
All right, our last question here. Number 25. What is the one most urgent piece of AI advice you would give to any business professional as they are entering the new year?
- [59:22]
Paul Raitzer
Get through the fear and uncertainty and anxiety and just get started like it is again. You can feel like it's passing you by. It's not. It is early. Most enterprises haven't solved for this yet. Most educational institutions haven't fully solved for this yet. You have an opportunity to lead, regardless of your age or experience level, and you have an opportunity to reimagine your career. And for anybody who's like, early on, you know, that's an amazing thing. I think I told this story maybe on the podcast once before, but I gave a talk, I think it was last year at Ohio University to a bunch of the professors, and a professor who was, you know, in his, like, late 70s, early 80s, came up to me after the talk, and he said, I'm so envious of you. And I said, why? He goes, I would give anything to be in my 40s again. Like, the stuff you're going to get to see and build and experience with what you just showed us is, like, incredible. And I. It weighed on me, like, I thought about that deeply on my ride home that day, because I do think it is, like, this incredible opportunity we have and this, like, amazing Runway ahead of us to just, like, reinvent everything, and that's. That's very exciting. So get started and be a part of that.
- [60:40]
Mike Caput
Well, that's an amazingly uplifting note to end on as we get to stack the questions. Well, well, I. I don't know that. I mean, I'm going into, like, Christmas, and now I'm like, no, no, I'm super excited.
- [60:53]
Paul Raitzer
I know this is. This is the last thing, like, last major thing I do this week. Again, we're on December 17th here, and I have, like, some big planning and vision things I got to do. But, like, I am looking forward to the holidays to actually step back and just, like, experiment with some stuff and, like, get started myself in some ways and try and think about how to, like, really transform stuff next year in a very positive way. Yeah. And, like, whether it's with my kids, with my own professional life and with the businesses, like, I do just think there's such cool things ahead and. And, you know, I'm inspired by our community, you know, like, 9,000 people or something in the Slack community now and thousands of people come to events and it's like I love hearing those stories and just seeing what people are doing. So yeah, I mean, as we said in episode 127, Michael, I'm just grateful that we have a voice that matters to people and is helping people. And I'm appreciative of being able to be part of the AI journeys of all of our listeners and viewers. And we're all figuring this out together. So it's going to be fun and overwhelming and exciting and terrifying and all those things in one. So we'll go through it together though.
- [62:05]
Mike Caput
Well Paul, as always, thanks for sharing all your insights and thanks for answering all these questions. I think with audience is going to get a ton of value out of this.
- [62:13]
Paul Raitzer
Yeah, thanks for pulling it all together Mike. And thank you again to everyone who's listening and who submitted all these incredible questions. We appreciate it.
- [62:21]
Mike Caput
Thanks for listening to the AI Show. Visit marketing aiinstitute.com to continue your AI learning journey and join more than 60,000 professionals and business leaders who have subscribed to the weekly newsletter, downloaded the AI blueprints, attended virtual and in person events, taken our online AI courses and engaged in the Slack community. Until next time, stay curious and explore AI.