Our Intro to AI courses offer foundational insights into the world of artificial intelligence. Having completed 32 sessions, we've encountered a diverse array of questions from our attendees. In this week's podcast episode, we're excited to delve into the most frequently posed questions and provide insightful answers. We hope you enjoy this episode, and join us for our first Intro to AI for Marketers class of 2024 on January 11!
Listen or watch below—and see below for show notes and the transcript.
This episode is brought to you by Algomarketing:
Algomarketing connects ambitious B2B enterprises to the competitive advantages of the autonomous. Their workforce solutions work to unlock the power of algorithmic marketing through innovation, big data, and optimal tech stack performance. Visit Algomarketing.com/aipod and find out how Algomarketing can help you deliver deeper insights, faster executions, and streamlined operations through the power of AI.
00:05:27 — How does an enterprise go from “one-off” approaches (individuals learning and using various AI tools for the needs of their role) to a more holistic AI strategy where tools and apps are chained together and are aligned to the ultimate goal of the company?
00:07:46 — How can I train AI on my company’s content?
00:10:38 — How much concern should I (my company) have about the content we feed into these public models? How do we keep data privacy top of mind?
00:12:24 — What is (or what will be) generative AI’s impact on search? How can marketers and businesses mitigate risk or stay ahead?
00:14:34 — To what extent does success with AI require good data? Should we be using the potential of AI as a trigger to elevate a culture of data hygiene across our organization?
00:15:56 — What are your best recommendations for agencies looking to take their teams (or their leadership team) on this journey?
00:18:25 — Once AI usage in marketing scales and matures, how will it impact the marketing team structure, and certain roles & responsibilities?
00:23:03 — In which marketing roles/functions are you seeing the greatest increase in efficiency or productivity? Which roles in our current agency/org setups need to be retrained/repositioned most quickly?
00:25:44 — Some companies don't permit employees to use ChatGPT as a tool because of data privacy concerns. What are the pros and cons of using ChatGPT in the workplace?
00:29:21 — What advice do you have for marketers eager to take your advice who are hamstrung by overly restrictive AI policies from risk-averse legal teams? Is there support for pushing back?
00:32:07 — Are there prompt engineering rules or best practices to help validate the accuracy of data included in the content and also avoid controversial topics/language?
00:36:32 —What suggestions do you have for AI development for small companies with a specialized market focus?
00:39:04 — What marketing processes/tasks do you feel can/will benefit most from GenAI? Which ones do you feel are least likely to benefit?
00:41:06 — We talked a little on the last Intro to AI class about AI for lead gen. You had some good thoughts about thinking bigger than generating leads, and more about the content needed, analysis being done, etc. Can we talk more about that?
00:43:59 — We get asked this a lot - what’s the best way, the best tool, the best use case to get started?
Similar to Episode 52, we have received a considerable number of questions from our Intro to AI classes. In that episode, due to the change of pace, we focused on Intro to AI classes 28 – 32, which were our post-MAICON 2023 classes.
This episode aims to address these latest developments and the prevalent questions from our attendees.
For episode 52, Cathy ran all 147 questions through ChatGPT in batches (30-40 at a time) due to input limitations and then consolidated them down to come up with 15 of the most representative questions.
This time, Cathy went through first and highlighted the questions she thought were the best questions to ask – that were either slightly different from ep. 52, or that had a newer, more thorough answer than previously answered.
ChatGPT was only used slightly this time around – to fine-tune the question for the podcast discussion, and also to have the questions re-ordered to better flow for a natural conversation.
Similar to last time, Paul reviewed the questions to ensure they were good questions for him and the audience, however, he didn’t prepare responses in advance.
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: So I don't think we can shortcut human knowledge. I don't think we can shortcut critical thinking and strategy. So there's things that are going to remain uniquely human, and I think that's part of the path forward is figuring out what those are for you and figuring out what they are for the industry.
[00:00:15] 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.
[00:00:35] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.
[00:00:45] Paul Roetzer: Welcome to episode 77 of the Marketing AI Show. We have a special edition today of Top AI Questions. Which hopefully will be really helpful to you heading into 2024. so I am joined by
[00:00:58] Paul Roetzer: Cathy McPhillips today, our [00:01:00] chief growth officer. Hello, Cathy. Hello. While we will not have our regular episode this week with, the main topics and rapid fire items, just a reminder, subscribe to the newsletter.
[00:01:11] Paul Roetzer: So marketinginstitute. com slash newsletter. There will still be a regular Tuesday edition with all the links for the most interesting topics of the week. there's a lot going on. So it's kind of like pains me not to be doing our episode this week. We have OpenAI doing a deal with Alex Springer, which publishes Politico, Business Insider, and a couple big European publications.
[00:01:33] Paul Roetzer: They're going to not only have access to their data for real time responses, but I think this is a prelude to some other bigger things that are going to happen between media companies and AI companies in 2024. So we will talk about that in the new year. Microsoft released a new model called, Phi-2, which is like a smaller model that has all kinds of impressive reasoning capabilities.
[00:01:56] Paul Roetzer: Google DeepMind released ImageGen 2, which is their image generation [00:02:00] technology. And Apple, I just downloaded this last night, Apple now has a new AI powered journal app. That comes with OS 17. 2. So if you haven't updated your iPhone update and there will be a new journal app automatically in there that like searches your music, your photos.
[00:02:21] Paul Roetzer: your conversations through text. So kind of, I guess, kind of creepy if you think about it, but it searches those and it recommends things to journal. So it's, it's kind of a fascinating play by Apple. I think it's going to be very under the radar, but it's maybe a prelude to where Apple goes with their stuff next year.
[00:02:39] Paul Roetzer: So again, check out the newsletter. I'll LinkedIn. I am going to be out of the office for the rest of the year. hopefully not working for the most part, but I will still keep some updates on LinkedIn. So, today's episode is brought to us by Algomarketing. Algomarketing connects ambitious B2B enterprises to the competitive advantages of the autonomous, [00:03:00] discover workforce solutions, and unlock the power of algorithmic marketing through innovation, big data, and optimal tech stack performance.
[00:03:08] Paul Roetzer: , visit algomarketing.com/AIPOD and find out how Algomarketing can help you deliver deeper insights, faster executions, and streamlined operations through the power of AI. Okay, so this episode, Cathy and I did this. Mid year? I think May. Was it May? Oh, wow. Okay. That was a while ago.
[00:03:28] Paul Roetzer: So we did like a top 15 questions in AI, and I believe it's been our most popular, highly downloaded show of the year. I think so. The idea here is, if you're not familiar, Cathy and I run an intro to AI class every about three weeks on Zoom. 2021. It is free. I do about a 30 minute presentation and then we do 30 minutes of Q& A.
[00:03:50] Paul Roetzer: We've done 32 of these sessions. So the last one was last week. we had over 800 people registered for that one. We've had over 16, 000 people [00:04:00] register for the series since starting it in 2021. And so what happens is Cathy curates all the questions at the end, but we never get to more than like five to seven questions.
[00:04:10] Paul Roetzer: And there are dozens of questions every week. So what we've done is Cathy went through and kind of synthesized what are the most common themes of questions that we get. And so today what we're doing is Cathy has picked 15 questions that we haven't had time to answer on Intro to AI, or maybe I've answered in variations.
[00:04:30] Paul Roetzer: And we're going to go through these one by one. I'm going to shoot for about three minutes per question. So hopefully there are questions in here that you've been wondering as well, and this will help you figure this stuff out as you build your plan for 2024. Did I miss anything, Cathy?
[00:04:46] Cathy McPhillips: I don't think so.
[00:04:46] Cathy McPhillips: But the only thing I was going to say is when you were talking about the questions is that we have hundreds of questions to sift through because they're all so different. So I'm hoping that we captured many of those within these 15. So I'm [00:05:00] looking forward to this.
[00:05:01] Paul Roetzer: I got my water bottle ready. Don't lose my voice.
[00:05:05] Paul Roetzer: Yes. I'm ready.
[00:05:07] Cathy McPhillips: I feel like Mike's 45 in 45 presentation. We're going to be rattling through these.
[00:05:11] Paul Roetzer: And by the way, this is completely unscripted. I scanned the questions once just to make sure like there weren't any I wasn't going to be able to answer at all, but I have not prepared anything, even bullet points for this.
[00:05:20] Paul Roetzer: So this is like. Off the cuff, hopefully, again, it's really valuable for you. So let's go.
[00:05:27] Cathy McPhillips: All right. Number one, how does an enterprise go from one off approaches to a more holistic AI strategy where tools and apps are chained together and aligned to the ultimate goal for the company? Thinking about like individual use cases versus individual roles versus companies.
[00:05:43] Paul Roetzer: Yeah, so what we've seen throughout 2023 is a lot of siloed efforts. So obviously through Marketing Institute, we often hear from the marketers, but I've had conversations with CIOs, Chief Digital Officers, CEOs. And it's been very rare that we have seen a unified effort [00:06:00] across the company to pilot and scale AI.
[00:06:03] Paul Roetzer: I think we will see much more of that next year. But the way I think that happens is a unified approach, and it probably needs to come from the CEO. Like, I'm a huge believer that the CEO has to be very supportive and invested in the future of the company being built around AI capabilities. And so I think you need a unified approach to education and training, a unified approach to an AI council that infuses the different functions of the business together, and then eventually building an AI roadmap where you prioritize use cases and pilot projects, and resources.
[00:06:35] Paul Roetzer: by department or division of the company. So I know that that's probably hard to do in a lot of companies, but many organizations we talk to are just at the stage where someone still needs to just raise their hand and say, I think there's an opportunity here. I'd love to be a part of building an AI council.
[00:06:50] Paul Roetzer: we saw that with our friends at VMware. That's what happened back in like February, March, a couple of people there raised their hand and say, Hey, let's do this. Led to a marketing at a council, which then, led to a [00:07:00] larger. company wide AI council. And I think that's what needs to happen. So if you're the marketer, raise your hand to start a marketing AI council, but try and get in the conversations to build a bigger AI council that figures all this out and unifies the efforts.
[00:07:13] Cathy McPhillips: And if you are not in our Slack community, I would just say there's a group of folks in there right now within the past week or so that have gotten together all from different companies, trying to figure out how they each can approach their own leadership. and develop one within their organization. So there's lots of conversations happening.
[00:07:27] Cathy McPhillips: A lot of the VMware folks are in there. There's a university in there and people want to do this. So people are willing to share the path they have taken to where they've been successful. So I think it's a great, great, information in there in our Slack community. I'll put that in the show notes
[00:07:42] Paul Roetzer: I'll say the link to that thread would be great. I haven't looked at that thread yet, so I'll definitely check that out too. Yeah, that's great.
[00:07:46] Cathy McPhillips: Okay. Number two, how can I train AI on my company's content? We get this a lot.
[00:07:53] Paul Roetzer: Yeah. So, I mean, the first thing here to be aware of is like, you don't obviously want to put your company's content into a model that's going to keep [00:08:00] that data and use it for future training.
[00:08:01] Paul Roetzer: So like you don't want to give ChatGPT, for example, unless you have an enterprise license that has a separate user agreement. you don't necessarily want to give it your data, especially if it's not publicly available data. So in most instances, you're going to need to have a separate agreement with one of either the model companies, like an OpenAI or an Anthropic or a Cohere, where they're built to allow you this training and you keep that data.
[00:08:26] Paul Roetzer: or you're going to work with a company like a Writer or a Jasper, where again, you trust that your data is safe and secure and no confidential information is going to leak into these models. So. I would say, that's the fastest path or the safest path. You can certainly train like a GPT. You can go and open AI and train a GPT with publicly available data.
[00:08:48] Paul Roetzer: Like we were having this conversation actually this morning at the Institute about transcripts to our podcast. We were trying to find like I remembered saying something, I couldn't remember when I said it, and we're like, well, could we just build a GPT that just has [00:09:00] all of our transcripts trained into it?
[00:09:01] Paul Roetzer: Now again, that's publicly available data, so I don't personally see any issue with putting that data into a GPT, because it's already out there anyway. It's just making it, you know, faster for us to use it. So, I would say, make sure you have the right people in the room. This may involve CIOs. It may involve legal, like you have to do this the right way to protect your company's data, but, it's likely going to be through a separate user agreement with one of the providers.
[00:09:27] Cathy McPhillips: So when you talked about the podcast and us having all of those, you know, building our own GPT, what would that process be? Would it be simply exporting the transcripts and uploading them into one of these tools or how does that work?
[00:09:37] Paul Roetzer: I think so. I honestly haven't thought about it. I was sitting there having a cup of coffee when it like.
[00:09:42] Paul Roetzer: Popped in my head that I wanted to find this thing, but yeah, I mean you could Like for us, we probably already have all the transcripts exported into docs because we export the transcripts anyway. So, I mean, certainly one path is to manually go through and grab all the transcripts and then upload them separately.
[00:09:59] Paul Roetzer: There may be a way to [00:10:00] bulk upload them, like into a GPT. Again, it may be more efficient to work with like a Cohere or somebody like that and do it. I don't know. I haven't really thought this one through, but yeah, the simplest one is like, let's say we want to take five shows and upload it. I would, I would envision just uploading five files to GPT.
[00:10:16] Paul Roetzer: I do think there are limitations on how many files you can put into a GPT. I don't know off the top of my head, remember what that was.
[00:10:24] Cathy McPhillips: Interesting. Yeah. Cause I mean, same thing this morning when you were asking about where's that one episode, I had the same thought a few months ago when the same thing came up, you know, a different topic, but it's like, there has to be a better way than us doing search and find, you know, within our, our stuff.
[00:10:38] Cathy McPhillips: So you kind of address this, but I'm going to ask and see if you can go in a little more detail for number three, how much concern should I, should I, or my company have about the content we feed into these public models? How do we keep data Privacy top of mind. So one of the things you talked about was the, user agreements, what kind of things should we be looking for in those and who on the team should be involved in that and ensuring it's the right tool [00:11:00] for you?
[00:11:01] Paul Roetzer: Again, you know, my general feeling is if it's publicly available data, that's already out there. I don't think it's much of a concern. If it's confidential information, that's sitting on your server that you wouldn't open up public access to it, then I would probably default to don't put it into a model that's going to take it.
[00:11:17] Paul Roetzer: Now, the chances of some data eventually showing up in like GPT 5 in somebody's output, probably pretty minimal, but it's just not something you risk. And this is, you know, I think we'll probably, I assume one of these questions will get into like, you know, policies and principles and stuff, but this is why generative AI policies are so critical for your company.
[00:11:38] Paul Roetzer: You have a whole bunch of employees who may be doing this already, and you don't even know it because no one's told them not to. And people don't really know how this stuff works. And so they may take like the transcript from an internal meeting and, you know, one of their jobs is to like summarize it and send it to everybody.
[00:11:53] Paul Roetzer: And they're like, Oh, I'll just have ChatGPT do it. And they take the transcript, drop it into ChatGPT. They get a great summary. So they're going to keep doing it, but they're not thinking [00:12:00] about like who keeps that data and now GPT 5 is going to be trained on our internal meeting notes and things like that.
[00:12:06] Paul Roetzer: So generative AI policies are like essential. to put these guardrails in place so people know when they can do this and, you know, maintain the privacy and security of the data.
[00:12:16] Cathy McPhillips: Absolutely. And we can put Jasper's Gen AI policy template in the show notes as well. That's a good starting point for a lot of people.
[00:12:24] Cathy McPhillips: Okay, number four. What is or what will be generative AI's impact on search? How can marketers and businesses mitigate risk?
[00:12:33] Paul Roetzer: Yeah, this is one of the great unknowns that I think we're going to deal with in 2024. so my, my initial feeling is I've guided a lot of companies to just assume there will be a drop in organic traffic.
[00:12:47] Paul Roetzer: We're getting into like changes in consumer behavior here. And, You know, the, I've given a couple of examples on the podcast in past episodes where like I can feel my own personal behavior changing. So the last couple trips I did in November and [00:13:00] December, I was in cities I'd never been to before. I didn't go and Google what to see in those cities or learn about them.
[00:13:06] Paul Roetzer: I just went to ChatGPT and said, first time in Raleigh, tell me what I need to know about the city, any restaurants I should see, what sites should I see? And it just like, does it for me way better than like getting a bunch of sponsored links on Google. So. That's where I started feeling my own behavior changing.
[00:13:21] Paul Roetzer: And so I think we're going to see a change in behavior. The other thing we dealt with on a recent episode was our search results in Google and other search engines going to start to decay because they're going to be surfacing AI generated content in them. And so all this synthetic media that's being created is going to find its way into search results.
[00:13:42] Paul Roetzer: And so we're actually going to see a drop in the value or quality of search results because it's being powered by AI generated content. We just don't know. So I think what you have to do is one, diversify your strategy. We always say like the more human content wins, more off the script stuff, more live events.
[00:13:59] Paul Roetzer: More [00:14:00] videos, that's actually you and not a synthetic version of you. editorials, narratives where it's true points of view from a human based on experience and things like that. Like, really steer into the more human strategy. And then have a team or have a couple people who very closely monitor the impact on your organic traffic.
[00:14:19] Paul Roetzer: Are you seeing trends? Are you hearing about industry trends? So I would say like watch like a search engine land places like that is going to be covering this very closely and really keep an eye on it as part of your initiative for next year. Absolutely.
[00:14:34] Cathy McPhillips: Okay. Number five, to what extent does success with AI require good data?
[00:14:38] Cathy McPhillips: Should we be using the potential of AI as a trigger to elevate a culture of data hygiene across our
[00:14:43] Paul Roetzer: organization? Yeah, certainly for AI at scale within your enterprise, the data strategy is essential. So Tim Hayden, you know, a good friend of the Institute, has spoken at every Macon. He talks about CDPs all the time, customer data platforms, and how you need to unify all that data and use that [00:15:00] to drive your AI strategy.
[00:15:01] Paul Roetzer: Absolutely critical, especially when you start getting into things like, you you know, personalization at scale, building predictive models, things like that. To use an AI writing tool, not as essential, like you don't need to go and get, you know, your CDP in place to, to start using generative AI tools. But if you're thinking about it more broadly as an organization, you absolutely have to get the data.
[00:15:23] Paul Roetzer: in shape. And you probably, it's probably not going to be just the marketing team working on this. Like you're going to be, this can be a cross functional thing within the organization to make sure this is done right. So yeah, follow the people who are experts in this space. Tim Hayden is one of them. but there's a lot of people who, talk more about data.
[00:15:41] Paul Roetzer: Chris Penn talks a lot about data. You know, that would be another person that comes to mind. Cal L. Dubé from Pan Data, who spoke in our conference a bunch of times as well. So there's just some names that come to mind right away.
[00:15:56] Cathy McPhillips: Number six, what are your best recommendations for agencies looking to take their [00:16:00] teams and or their leadership team on this journey?
[00:16:03] Paul Roetzer: Yeah, so agencies, we just in November did a AI for Agencies Summit. So if you're an agency, listen to that. That whole, it was about five hour virtual summit. That's all available on demand.
[00:16:13] Paul Roetzer: I did a keynote on like state of AI and agencies. as I think a lot of people know, I used to run an agency. So I had my own agency for 16 years. So I kind of put my own agency leader hat back on to create that event and specifically my keynote. I think the way I would approach this is, is twofold.
[00:16:32] Paul Roetzer: First, I would look at your own business and say, how can we use AI to be a more efficient, more creative. more profitable business. And that means AI within your HR, your finance, your own marketing sales service. you're really looking at the application of AI to build a smarter version of an agency.
[00:16:51] Paul Roetzer: Then I would look at and say, okay, what can we be offering to clients? Where is the demand in the market right now? What are the services people need? So like. large [00:17:00] language model strategies or generative AI strategies. A lot of companies are struggling with that. Like, what do you do? change management, the implications of this technology as it starts to be diffused into organization.
[00:17:10] Paul Roetzer: What's going to happen? Education and training within those companies. Like we're seeing massive demand for that stuff. So I would really just be thinking about. Where is the market going? Where is AI going to take consumer behavior? And then what sort of services do you need to be building to create value?
[00:17:26] Paul Roetzer: In the process, I would absolutely be considering your own pricing model. Because if you charge billable hours, like, it's just not going to fly. And, again, people familiar with my background, I wrote Marketing Agency Blueprint in 2011. Chapter 1 was eliminate billable hours. Like, I've always been a proponent of value based pricing.
[00:17:45] Paul Roetzer: And I think now moving forward, it's, it's just essential. And then for the brand side who pays these agencies, you have to work with agencies that get it. Like if your agency isn't building internal AI education and training, isn't [00:18:00] preparing their teams, isn't building services around this stuff. Go find an agency that is, because they are not going to be able to create enough value for you a year from now.
[00:18:09] Paul Roetzer: Yeah.
[00:18:09] Cathy McPhillips: I think back to my agency days and I would just think there are so many efficiencies, especially with like reporting and stuff. You know, it's just, gosh, how much time can we have saved doing that? So we could really have been creative and done all the things that we loved doing. The monthly reporting was the worst, you know, but it was so critical to the business.
[00:18:25] Cathy McPhillips: So yeah, there were so many opportunities. Number seven, once AI usage in marketing scales and matures, how will it impact the marketing team structure and certain roles and responsibilities?
[00:18:35] Paul Roetzer: This is definitely in the unknown category. we had Dan Slagen from tomorrow. io did a talk on, on this at, Macon 2023, where he, you know, they had hired an AI specialist within their company.
[00:18:51] Paul Roetzer: And they were starting to look at like, what is the future of their org chart look like? Is there going to be AI ops people kind of across each function? It's going to be a single [00:19:00] person. you know, what are these roles and capabilities going to need to be? And so I think we're going to start to see more going into 2024 of new roles emerging.
[00:19:10] Paul Roetzer: I do think AIOps is probably one of the first ones we'll really see. We've seen this at one healthcare company that we've talked with where they took someone who was a project manager who's been there forever and they moved her into an AIOps role. And so I think you're going to see a lot of that people who have the domain expertise, have the comfort level within the organization, especially big enterprises.
[00:19:34] Paul Roetzer: where you understand the procurement process, you understand everything about the strategy, and there's going to be these new roles that emerge for people who really grasp AI, deep competency in it, are comfortable working with and talking to these vendors, are comfortable leading onboarding within their organization.
[00:19:51] Paul Roetzer: So if you go get, for example, like writer dot, writer, is it writer. com I think? So if you get Writer, and you infuse Writer into your [00:20:00] marketing team, there's all kinds of change that needs to happen. It's not just, oh great, we got a writing tool. Someone needs to teach that. Someone needs to maintain the prompt library.
[00:20:08] Paul Roetzer: Somebody has to do the brand style guides. Like, there's all these other things that go into enterprise adoption of these tools, and nobody's really Trained or prepared to do that. So I do think that we'll see new roles being created due to ai. It's hard to project what those will be, but I often go back to like, say, imagine like marketing in 2005, 2006, we didn't have mobile, we didn't have social really, we didn't really even have.
[00:20:35] Paul Roetzer: content marketing the way we know it today. And think about all the roles within your team today that didn't exist in 2005. Like they just weren't even possible because the tech hadn't really been diffused across society yet. I think that's what we're going to see in the next three to five years is just all kinds of new roles created.
[00:20:53] Paul Roetzer: And then what does that mean to our org chart? I really don't know yet. I think you'll see a lot more chief AI [00:21:00] officers in companies. You know, that's probably going to be a role we'll see a lot of, but beyond that within marketing, I think you're just going to have a lot of people who are already in communications or PR or advertising or analytics that are going to develop AI capabilities.
[00:21:13] Paul Roetzer: It's just going to be a skill set. and so I think the titles may actually stay the same. You just have to advance your skills.
[00:21:19] Cathy McPhillips: Sure. So how is this different from say, you know, getting a new CRM or getting a new piece of big technology? How is, what's, what's the difference or is there a difference?
[00:21:28] Paul Roetzer: Well, I mean, we always teach it as it's just smarter technology.
[00:21:34] Paul Roetzer: So I think in a lot of cases, it's just going to change the way you do your job. So, you know, you think Cathy about like our podcast, we talk about this as an example all the time, like if your job is to manage a podcast within a company, you're still going to be managing the podcast next year, but you may be using Descript and what are the other tools?
[00:21:52] Paul Roetzer: We use AnyClip, like you're going to use all these other tools to do your job. And the AI may start doing like 20 percent [00:22:00] of your job. Right. And you're either going to do more like promotion. Like you're going to fill that time with something else. You're going to launch another podcast, like whatever it is, but yeah, it's just smarter technology to help you be more efficient and creative at your job.
[00:22:14] Paul Roetzer: so there, I think people overcomplicate it sometimes thinking of it as like this whole new shift and everything, which it sort of is. But in the simplest way, it's just smarter technology.
[00:22:24] Cathy McPhillips: Right. I mean, it makes you wonder if like those, the term artificial intelligence is this roadblock and people just can't get past that when really they need to think about it differently.
[00:22:31] Paul Roetzer: Yeah, I think, you know, certainly that's why we look at education as so critical, like literacy and AI is essential because once you understand what it is, and this again, from the, Page one of our book, it's just smarter technology. We say it every time on the Intro to AI class, like it's just smarter technology.
[00:22:46] Paul Roetzer: And so once people realize like, Oh, okay. So me as the email marketer can go assess the AI tool because I'm just looking at saying, well, my job today is this with these 20 steps. Once I have this product, my job tomorrow is this, but the AI is going to do five of those steps for me. [00:23:00] Great. But that's basically all it is.
[00:23:03] Cathy McPhillips: Okay. It's almost like these questions were planned out, you know, to segue to the next one. number eight, in which marketing role or functions are you seeing the greatest increase or efficiency in product productivity and which roles in an agency or org need to be retrained or repositioned
[00:23:19] Paul Roetzer: quickly?
[00:23:21] Paul Roetzer: Yeah, I mean, the early ones we're certainly seeing would be writing, coding, anywhere where development is done. I think anything within the creative functions, so certainly like graphic design is going to see massive efficiency gains. 2024 will be where we start seeing huge efficiency gains in video production and audio production.
[00:23:40] Paul Roetzer: So when you think about, like, the five main categories of generative AI that we talk about, it's text, Image, Video, Audio, Code. You're going to see efficiency gains if your job is to do any of those things. 2023 was sort of the year of text. In some ways, also the year of image generation. video saw [00:24:00] major leaps in the last few months.
[00:24:01] Paul Roetzer: I think next year we'll see incredible advances in video capabilities. Audio is sort of having its moment. We're starting to see a lot of models coming out with audio built in. Gemini from Google will do that right away. and then coding is probably actually the first, like we've been seeing advancements in coding over the last couple of years with like GitHub Copilot and Repl.
[00:24:21] Paul Roetzer: it and things like that. So yeah, I mean, it's, there really isn't going to be an area of marketing that isn't touched, but if you do any of those generative AI categories, it will affect you and drive efficiencies. So
[00:24:32] Cathy McPhillips: what are some things we could be doing to reskill ourselves to fill some of those gaps so we stay
[00:24:37] Paul Roetzer: relevant?
[00:24:38] Paul Roetzer: You gotta, obviously, like whatever training you can get, but you have to experiment with them. So, you know, if you're an Adobe user, I was talking to a company the other day who, you know, is Adobe Firefly, they have it, and they do a ton of creative. And it's like, okay, like get training on it. Like, don't just have the tools, like have people go through and build experimentation, share what they're learning.
[00:24:58] Paul Roetzer: And so I think it's really more [00:25:00] formalizing, not only the training of the tools, but the sharing across teams of what works, because these things aren't like traditional software. You're going to find these really nuanced ways to get more value or a higher quality output from the system. And you need to be able to share that information.
[00:25:17] Paul Roetzer: I don't care if it's a Slack channel, like, I don't know how you do it. But you have to be very dynamic in being able to take learnings and spread them across the organization. Like, just in the last couple of weeks, we've seen this with large language models where people are finding ways to dramatically increase the quality of the output by just like tweaking the prompt a little bit.
[00:25:37] Paul Roetzer: and we'll see a lot more of that with all areas of generative AI next year.
[00:25:44] Cathy McPhillips: Okay. Number nine, some companies don't permit employees to use ChatGPT as a tool because of data privacy concerns. What are the pros and cons of using ChatGPT in the workplace?
[00:25:54] Paul Roetzer: This is definitely an issue. There are a lot of companies we talk to that don't have access to ChatGPT.
[00:25:59] Paul Roetzer: [00:26:00] So my guidance is usually, you know, especially to marketers, find a use case. Like first, understand why you're not allowed to use it. is it legal? Is it IT? You know, who, who has the concerns around the use of the technology? Who is the one that's actually shutting it off? Figure out what those are, and then find use cases where those concerns don't apply.
[00:26:24] Paul Roetzer: And then go get permission for that use case. Say, hey, can we use ChatGPT for this specific thing? It won't be connected to any data. There's no confidential information that can go into it. We will set specific guardrails. You can even monitor the usage if you want. This is what we're going to use it for.
[00:26:41] Paul Roetzer: Here is the business value we think can come from this. Get one use case approved, then do the next one. Like prove that that worked. proved it did not cause any undue risk or liability, and then do the next one. So if you're in a highly regulated organization, this may be your only path. I have talked with companies that are having [00:27:00] to do this one use case at a time.
[00:27:02] Paul Roetzer: So they can't get broad approval for ChatGPT Plus for the team, but they can maybe get specific individuals approved for specific use cases. So you have to know your own organization, but Understand there's probably good reason why they're not allowing it, like why legal is shutting it off. And I get that it can be very frustrating as a marketer to be told no and not always get the exact reasons why you're being told no or not agree with them.
[00:27:29] Paul Roetzer: But I think in this case, we have to give the benefit of the doubt to legal and IT that they're doing what's best for us and for the company and for the customers. so understand that, work with them, say, what would you need from me for us to begin testing this technology? because we think we can see.
[00:27:45] Paul Roetzer: 20, 30 percent gains in efficiency. And so like, if they say no, then go to the CEO and say, Hey, if I could give you 20 or 30 percent gains in efficiency next year and save this company 10, 000 hours, would you like that? Yes. Okay. I need your help getting legal out of the way so we can do this. [00:28:00] So I think like know your organization, know what the goals of the company are and talk about those.
[00:28:04] Paul Roetzer: Don't talk about, I want to use AI technology. You're not going to win the argument in that case.
[00:28:10] Cathy McPhillips: So it seemed like when ChatGPT first came out, a lot of companies said no, and then slowly some of them said, okay, okay, but here's the use cases or here's our guidelines or whatever. So it's like the ones that remain, they must have some, it's, it's been 12 months.
[00:28:23] Cathy McPhillips: They've had time to dig into it and have good reasons on why they're not allowing it.
[00:28:25] Paul Roetzer: Yeah, some are just highly regulated and it's just that they're conservative and how they approach that stuff. But we all, I mean, we talked about, I think it was episode, what, 75 or 76. We talked about the Salesforce study, about how many people are using generative AI tools and just not telling their employers.
[00:28:42] Paul Roetzer: And in many cases, like passing it off as their own work and not telling their employers. So people are doing this. Like it, there are a lot of. Like regulated companies I have talked to who don't like condone it, but they haven't told them they're not allowed. And so they basically just have people do whatever they want and they're just kind of turning the [00:29:00] other way and like pretending like it's not happening.
[00:29:03] Paul Roetzer: So I think just transparency on all parties. You know, clear guidance and guardrails. If you, if you have to put some policies in place that restrict use, that's fine, but at least be clear about what you're doing and why you're doing it and how they can maybe take steps to get it, get past those issues in the future.
[00:29:21] Cathy McPhillips: Yeah, a fluid policy is better than no policy, right? Yeah. Okay. Number 10. What advice do you have for marketers eager to take your advice who are hamstrung by overly restrictive AI policies from risk averse legal teams? Can or should I push back?
[00:29:36] Paul Roetzer: Yeah, this probably, we probably addressed this one in the previous one.
[00:29:41] Paul Roetzer: You know, I think again, it is sitting down and talking to them and understanding what's going on. Like why, why aren't they allowed to do this? and then try and find those use cases and the technologies where it's not an issue. Like for example, If you're just assuming we can't do anything because Ligo is not saying it, go and say, okay, well, can we use Descript because it's [00:30:00] just for our podcast?
[00:30:01] Paul Roetzer: All this information is already publicly available. There's nothing we're doing here that causes risks or liabilities. It's literally just taking it, doing a transcription, doing a summarization, chopping up some videos, putting it up on YouTube. Like, that is a very low risk thing. And because if you go back to the issue here is often understanding of AI.
[00:30:18] Paul Roetzer: You may have the attorneys or the IT people who don't really understand how marketing works or what you're trying to do. They're just thinking AI is a red flag and we got to shut it down. They're not going to think about all the use cases you have. So, involve them. Like, show them the things you could be doing that aren't going to cause any issues.
[00:30:38] Paul Roetzer: So, I just, be understanding, but be proactive, in a way that creates value for both of you.
[00:30:46] Cathy McPhillips: So, if some of that risk averse legal is due to copyrights, can you address that?
[00:30:51] Paul Roetzer: Yeah, I think, again, that goes back to you understanding copyright law and what is allowed. So, again, the The current [00:31:00] status is that the anything generated by AI cannot have a copyright to it.
[00:31:06] Paul Roetzer: You can use AI to assist in the generation of stuff, but you can't own a copyright if the AI creates it and the prompt doesn't equal authorship. So that's kind of the simplest way to think about it. That doesn't really matter for like emails, it doesn't apply to transcribing podcast transcripts. Like if we take this and we use AI to write this and then write a summary of this, this is still my work.
[00:31:28] Paul Roetzer: Like it's still human authored content that AI is just helping us. create in various formats. So that's a non issue. so I think you have to go through, how are we going to use it? Is it to write articles that we're going to publish that we want to have the copyright to? Or is it to help change the tone and make it, we have some sales people who aren't very empathetic or friendly, and we just want to make their emails more friendly.
[00:31:52] Paul Roetzer: Then fine, like who cares? Let the email, let the AI write the email if it's more empathetic than the human. So I think, again, it's understanding your own use [00:32:00] cases and whether or not copyright is really a concern in that use case. Okay.
[00:32:07] Cathy McPhillips: Number 11. Are there prompt engineering rules or best practices to help validate the accuracy of data included in the content?
[00:32:15] Cathy McPhillips: And is there a way to avoid controversial topics or inappropriate language? I've never seen an inappropriate language come out of
[00:32:21] Paul Roetzer: one of these before. Don't use grok.
[00:32:26] Paul Roetzer: If you want to avoid inappropriate languages, don't use, yeah, Grok.
[00:32:31] Cathy McPhillips: So prompt engineering rules or best practices?
[00:32:34] Paul Roetzer: Yes. So definitely. this is an evolving area. So early in 2023, there was all this buzz about like, is prompt engineering going to be like a role in a company? Are we going to have like prompt engineers as marketers?
[00:32:50] Paul Roetzer: And my feeling was always like, no, like, this isn't going to be a thing. It's going to be a skill set. And what's going to happen, and what is happening, is that the AI is [00:33:00] going to get better at improving our prompts. So we're seeing this now. Like if you go into ChatGPT, you can ask it to write you a prompt for something.
[00:33:08] Paul Roetzer: If you give it an image generation prompt in ChatGPT it rewrites that prompt. to be way better than what you're going to give it. So there is the fact that the AI is going to get better at improving your prompt. That's going to dramatically reduce the need for us to get really good at prompting.
[00:33:26] Paul Roetzer: However, we are seeing right now, that certain approaches to prompting do create better outputs. So the advice I often give people, without getting into like all the steps, I think we have an episode where we got into like all the actual steps that OpenAI recommends. Ethan Mollick had a great post about it.
[00:33:44] Paul Roetzer: The simplest way to think about this is, you are giving a project to a really smart intern, Describe what you want from them the way you would to a human. So, hey, here's the project, here's the parameters of the project, here's what we want the output to be, here's three [00:34:00] examples of previous outputs of this exact thing.
[00:34:02] Paul Roetzer: If you just do that, you will get better outputs from the AI. you'll also eventually see that things like telling the AI to take its time or telling it to think through it step by step or telling it to check its work. like for example, in DALI 3, when using ChatGPT if you want it to have words in the image, tell it to check its spelling before it creates the output.
[00:34:27] Paul Roetzer: And it will. Again, it's like weird stuff. And the only way to figure this stuff out. is to continually test these tools yourself. You cannot just go and listen to a podcast or read some stuff and think you understand it. The nuances are kind of unique sometimes and you got to figure out what those are.
[00:34:46] Paul Roetzer: So again, the high level, talk to it like a human, as weird as that sounds, you will get better outputs from the machine.
[00:34:57] Cathy McPhillips: So part of that question included, can you use the tools [00:35:00] to check the accuracy of the data? And no, right? I mean, I would rely on it for that. A human is needed for that part of it.
[00:35:07] Paul Roetzer: Anything that is like, important, that you're going to use in a research report, or in internal reports, or like, something like that, the human still has to check the work. Like, the AI is getting really good at data analysis and narratives of data. But it does still hallucinate and it does it in very subtle ways sometimes.
[00:35:29] Paul Roetzer: And it may be like 98 percent right. And you check, you know, eight out of 10 things and you're like, Oh, that's good. Like I don't need to check the rest. And the mistake may be in the 10th one. So, yeah, you can't rely on it for things that have a high risk, like if it's going to be wrong. and so data would be the same way right now.
[00:35:47] Paul Roetzer: Now they may resolve that in future versions, but there's no indication it's going to be fixed next year. That we're just going to be able to rely on these things 100%. But the way I think about this is You know, you can't rely on [00:36:00] humans 100%. Humans make mistakes with data all the time, but you still check the work.
[00:36:03] Paul Roetzer: So, almost think about it like if you're asking it to write, a report on your analytics, for example, and you're the director or the VP, and you're the one that's responsible for handing it to the CEO, and if it's something wrong, it's on you, not the junior person who did it. That's basically how you have to treat the system.
[00:36:20] Paul Roetzer: You have to check these things the way you would. If you're standing behind that work, as this is, this is accurate, then you have to go through the same process you would as if an intern did it for you. Absolutely.
[00:36:32] Cathy McPhillips: Okay. Number 12, what suggestions do you have for AI development for small companies with a specialized market focus?
[00:36:40] Paul Roetzer: AI development. So I don't know if that's like building their own thing or, I don't know about this one. So I'll try and answer this a couple ways and hopefully, I answer the question correctly. So first of all, most small companies aren't building their own AI. [00:37:00] They're just going to use all these tools.
[00:37:01] Paul Roetzer: So the beauty of a lot of these tools is they're incredibly affordable. We're a seven person team at the Institute. We use probably, I don't know, 12 to 15 different AI tools. Every week? Every week. And probably spend less than 500 a month on all of those tools, I would guess? I don't know. If I'm not including like our HubSpot license in one of those, I'm just talking about like point solutions for AI specific tasks.
[00:37:23] Paul Roetzer: so most small companies are, are just going to, you know, SaaS model it and they're just going to pay a monthly fee to do it.
[00:37:32] Cathy McPhillips: So maybe look at it from the
sense of, you know, if you're a small company and you are just have this specialized niche, can the ChatGPTs of the world still be useful?
[00:37:41] Paul Roetzer: Yeah, so what I, what I think we're going to start to see, so when OpenAI released GPTs, in the fall, what it did do was give any company the ability to develop, products.
[00:37:54] Paul Roetzer: Now, eventually they'll have an app marketplace where you can sell these things. So in theory, you could train a GPT on [00:38:00] some specific knowledge you have in a specific market and offer it like accounting, for example, or law, or marketing strategy. We could take like specific approaches to marketing strategy.
[00:38:10] Paul Roetzer: We could build a marketing strategy GPT. So I guess in that sense, small companies, I think in the near future, will probably have the ability to build whatever they want. That previously may have taken, you know, a million dollars in funding to go build. And then you had to hire a couple of developers. I think it's going to democratize everybody being able to build stuff.
[00:38:27] Paul Roetzer: Like Repl. it, the company I mentioned earlier, their, their mission is like a billion developers. Like they want everybody to be able to develop whatever they can imagine with language. Like you just say, this is what I want to build. And so like right now. I can do that with images and I can kind of do it with videos and I can sort of do it with audio.
[00:38:44] Paul Roetzer: But in the future, it'll be like text to app, like build me an app that does this. You can kind of do that now with some of the Python coding and ChatGPT. But I think that's going to become within the next 12 months, much easier that really anything you can imagine building, you could probably build a prototype pretty quickly with [00:39:00] just your words.
[00:39:01] Paul Roetzer: That's incredible.
[00:39:04] Cathy McPhillips: Okay, number 13. What marketing processes or tasks do you feel can or will benefit most from GenAI? And which ones do you feel are least likely to
[00:39:13] Paul Roetzer: benefit? So, I think writing is the answer everyone would assume I would give for what's going to benefit the most, but I would actually say it's strategy.
[00:39:24] Paul Roetzer: I think it's the most often overlooked way to use these tools is as a strategy assistant. I rarely hear it talked about as people doing it. So I think that the process that can benefit most is, is strategy. And in part, I think that because it's the hardest thing to teach. So I always said, like at my agency for 16 years, like I would hire people.
[00:39:48] Paul Roetzer: I felt like I could train writing. I could train a lot of things. I couldn't teach strategy to someone who didn't already have strategic abilities. It's like, You really need, like, it's a [00:40:00] very rare skill to be a deep strategic thinker, but it's so valuable in marketing. And I see these AI tools. as an ability to really democratize strategic planning, almost in a sense of how we're doing it with coding, where like anybody can build an app.
[00:40:17] Paul Roetzer: I think we could use these tools to give strategic ability to people who maybe struggled with that. so I think that's a huge area where we could see massive lift and for marketers and then least likely to benefit. I would say in the near term, it might go back to the data side because while it's going to be super helpful and while it can crunch numbers and do all kinds of interesting things with like ChatGPT code interpreter and Gemini from Google is going to have these abilities and things like that, it's not reliable yet.
[00:40:49] Paul Roetzer: And so I think in the near term, we're not going to see like dramatic shifts in how these tools. Drive huge efficiencies from a data perspective [00:41:00] until they become a little bit more reliable, which could happen next year. I mean, it's, it's going to happen. And I just don't know when it's going to happen.
[00:41:06] Cathy McPhillips: And just like with, you know, ChatGPT, when you're writing, you still need, you know, from a data side of things, you still need someone putting the data in the notes, what they're putting in and what they want from it. So there's a whole learning curve from that side of things as well. Okay, number 14. We talked a little on the last Intro to AI class about AI for LeadGen.
[00:41:26] Cathy McPhillips: You have some, you had some good thoughts about thinking bigger than generating leads and more about the content needed, analysis being done, etc. Can you talk more about that?
[00:41:35] Paul Roetzer: What did I say on the intro that was like so profound? I'm trying to remember back like what I talked about about.
[00:41:41] Cathy McPhillips: You talked about how someone said, what tools can I, can I use for LeadGen?
[00:41:45] Cathy McPhillips: And you said, well, LeadGen's everything we're doing. So there's so much more to it than just that, you know, just getting that lead. It's writing the content, doing all
[00:41:53] Paul Roetzer: those things. Yeah. So I, okay. that is a good point. So the, yeah, when people ask about like, how [00:42:00] do we increase leads? How do we use AI to increase leads?
[00:42:02] Paul Roetzer: I think that they think about Conversion rates, optimization of ad campaigns, things like that. And what we're saying is like, you got to take a first principles approach to this and start from the ground up and say, well, who are you, what are the leads you're trying to create? Like, who are they? And can you take an entirely new approach to that persona or that audience?
[00:42:22] Paul Roetzer: So start with the persona, like go into ChatGPT plus and say. We target, you know, I don't know, CEOs in the healthcare industry, and we want to try and find ways to reach them. Here's how we've previously done it. What are new ways we could do it? And it's going to probably come at you with some ideas.
[00:42:39] Paul Roetzer: Some you might throw out, some may be really good. Okay, expand on this idea. How could we do that? What, what is the value we create there? So like, really starting from that strategic level, and then using it to say, okay, let's We're going to create a blueprint for this. Like, let's go create that. Okay. Help me build an outline for that.
[00:42:55] Paul Roetzer: Should we do an online course about it? Should there be a webinar? Like, and have it help you [00:43:00] at each step of the way. And so in the end. Rather than just getting more leads, you may get way, way better at converting the people who are already there. So maybe your lead conversion percentage goes from 2 percent to like 5 percent because it was just a better plan.
[00:43:14] Paul Roetzer: It was better content. It was better everything. You understood that person in a more, you know, thorough way. And so I think that's stepping back from the traditional way of we think about we just need to get more leads and say. How can we just reimagine this whole process from step one is, is what probably these tools enable that a lot of people are overlooking.
[00:43:36] Cathy McPhillips: That reminds me of when I was doing a project a couple of years ago, we wanted to target people in DemandGen on LinkedIn. And like there were, there was a number of folks with DemandGen, but it was very small. And we're like, this looks wrong. We're like, well, that's not the job title. We're all, you know, we're all in DemandGen and LeadGen to some capacity.
[00:43:51] Cathy McPhillips: So there's so much more to it than that. This is, we're going, we're breezing through these
[00:43:56] Paul Roetzer: questions. I know, it's not bad. I told you like three minutes per, that was my goal.
[00:43:59] Cathy McPhillips: [00:44:00] Number 15. we get asked this a lot. I ask this at every single intro class. What's the best way, the best tool and the best use case to get started?
[00:44:10] Paul Roetzer: so high level, my answer is always ChatGPT plus. Still the best option on the market, best value. but the probably better answer is it depends on what you do for a living. If you are a writer, if you are an advertising professional in creative or media buying, if you do analytics, if you do social media, if you do communications, it is to make a list of all the things you do.
[00:44:36] Paul Roetzer: Like just take a spreadsheet and say, okay, this list out. These are the 20 things I do every month. And then say like, okay, create another column. How often do you do them? Daily, weekly, monthly? How many hours a month are you spending on them? you could put like a, like a joy column. Like how much do I enjoy this work?
[00:44:52] Paul Roetzer: Like, I hate this thing. I hate this report, but I do it 20 hours a month. Like just find the variables that will help you [00:45:00] determine which task or tasks that you do. Would you most like an AI to help? And it could be because you don't like the task. It could be because you spend 50 percent of your month doing it and it's not fulfilling to you.
[00:45:12] Paul Roetzer: Whatever the criteria is you determine, you just want to find the thing that's going to help you be better at what you do or enjoy what you do more. And so that's why I say it's, it's a very subjective. So when I talk about what is the most valuable tool, ChatGPT Plus is obvious for everyone. We love Descript.
[00:45:31] Paul Roetzer: Like Descript is transformative to our marketing efforts. So that's a very valuable one for us because our podcast is a core part of our growth strategy. So again, it's stepping back and saying, well, what are the things we're doing that matter? It's our Intro to AI class, gets a ton of people in and helps a lot of people.
[00:45:48] Paul Roetzer: Our podcast grows an audience in a way we never imagined possible. So like If I could find ways to be better at those two things, that's really valuable to us. so that's how I kind of think about it, is where are your [00:46:00] resources going, where are the campaigns, which things are helping you achieve your goals the most, or maybe are functioning below goal, like we're not hitting this goal.
[00:46:08] Paul Roetzer: So the best tool is going to be the one that helps you hit that goal, because that's how you're compensated. So it's very personal, I think, which tools are best. but at a high level, you know, there's some that we just think are great tools because that's the ones we use.
[00:46:23] Cathy McPhillips: And I say often a lot when we're talking about the podcast, because this is part of the presentation that I give, is I go through the whole podcast process and I say, you know, based on the things that I'm doing, let alone the things that Mike's doing with, with the writing the four, the three blog posts with, on each of the topics, is I know that I'm still, that should take me 17 hours of my life.
[00:46:43] Cathy McPhillips: to do that every single week, just me, not even Mike's part, and I'm doing it two and a half or, you know, Claire's helping now, but I can take those 15 hours and I'm in Slack an hour every day, talking to community members, I'm scheduling phone calls, I'm doing human things, I'm doing all these things, and that's what's important.
[00:46:59] Cathy McPhillips: That's what, that's what [00:47:00] brings me the joy, right? So, cool. And again, a lot of stuff with the podcast, I did not edit videos before using Descript. So it's teaching me a lot of things as well. So it's filling a knowledge gap and a skill gap that I didn't have. And now I, and now I'm talking about being a video editor.
[00:47:15] Cathy McPhillips: It's like that, who would have thought that was going to happen?
[00:47:18] Paul Roetzer: Yeah. And I think that's like our approach at the Institute is always, when we think about a strategy or we think about the use of a tool, We say, how does it make things more intelligent and how does it make it more human? So part of our plan, very intentionally, is to free Cathy up to do more interpersonal communication.
[00:47:35] Paul Roetzer: Like, spend time with people, build relationships, hear from them, what is going on? What could we create of value for them? What courses would be helpful? And that's, the AI, the AI is not going to do that. It's not going to build those relationships. And so if she can save 20 or 30 hours a month that she can rededicate to the community, great.
[00:47:51] Paul Roetzer: Like, that's the more human side. And so I think as you look at your strategy, moving forward, and you think about your use of AI tools, think about how do we become more intelligent as a [00:48:00] marketing department or as a company, and how do we become more human? And the more intelligent part is what frees you up to do the more human stuff.
[00:48:07] Cathy McPhillips: And quite honestly, the more human stuff that I'm doing is, brings me the most joy, but it's also driving revenue.
[00:48:15] Paul Roetzer: No doubt. Yeah. And we can attribute that and we can look at it and say, okay, that's time well spent. But yeah, I think the first one is like you said, it just brings you more enjoyment. Like I I'm a huge believer that AI should make us enjoy our work more.
[00:48:28] Paul Roetzer: The stuff that we don't enjoy, the repetitive data driven stuff that many of us don't really care for, we just do it because we have to. Maybe that's what the AI does and we get to do the other stuff that's a little bit more fulfilling. I love strategy. I mean, I, I'm never happier in my job than when I can spend eight hours with no meetings, no calls.
[00:48:45] Paul Roetzer: And I can just like stare at whiteboards and I can pencil things out on paper and I can just think. So I love plane rides sometimes. Like I just think uninterrupted. And so that to me is like joy. And if I can spend my time Doing that kind of stuff, then, you know, I [00:49:00] really love what I do. I don't love like, where's this report?
[00:49:03] Paul Roetzer: Where's that report? Oh man, we're not even tracking this thing. we knew we should do, there's all these things we know we should do. We don't have the time to do it. I would love for AI to do all that stuff, just surface things for me, almost like that Apple Journal when I said at the beginning, I would love an Apple Journal for our business, just like, it's seeing things, it's servings them, what's going on here, and you're just like, like, maybe somebody should build that.
[00:49:26] Cathy McPhillips: Well, speaking of joy, talk about a little bit about, you know, people talk about Gen AI and writing blog posts and writing all these things for us. And most of our team, we don't use it for that.
[00:49:35] Paul Roetzer: No, we don't. We don't. Yeah, I've said this often when I'll do like the intro class. I don't. I think like AI writing for you is like the least interesting thing it does.
[00:49:46] Paul Roetzer: because to me, writing is a critical part of thinking and it's like how I analyze ideas. It's how I form my own thoughts and points of view on thing. And if I just like, let AI do it, then I can't stand behind it. Like I can't explain [00:50:00] it to you. I actually find this, even with summarization, a lot of people like use it to just summarize stuff.
[00:50:05] Paul Roetzer: And then they just read the cliff notes version. If I'm going to talk about something on our podcast each week, I read every single thing that we talk about. If I was just using AI to do summaries of each of those articles, I could never off the cuff, just talk deeply about topics. So I don't think we can shortcut like human knowledge.
[00:50:26] Paul Roetzer: I don't think we can shortcut critical thinking and strategy. I think it's just there to assist us. It's a, it's there to kind of. enhance what we're capable of doing. Like maybe I can read another three research reports a week because of the AI, but it doesn't replace the fact that I'm still doing it. It can't go listen to podcasts for me.
[00:50:44] Paul Roetzer: It can do a summary of the transcript, but it's not the same as investing the hour to like listen deeply and hear the inflection in someone's voice and be like, Oh, Sam Altman really means that. Like, it's not words on paper. It's like he was hurt by the fire that you can feel it in his voice. And like, AI doesn't get [00:51:00] that stuff.
[00:51:00] Paul Roetzer: So there's things that are going to remain uniquely human, and I think that's part of the path forward is figuring out what those are for you and figuring out what they are for the industry.
[00:51:09] Cathy McPhillips: So I was on a, I was being interviewed for an ebook a few weeks ago, and the woman who was interviewing me was writing, taking notes.
[00:51:16] Cathy McPhillips: And I said, do you want me to record this just so we can have a conversation and you can just transcribe it? And that way you can have my quotes right. We can just have a conversation. You're not trying to talk and think and listen and write at the same time. And she was like, Oh, that would actually be amazing.
[00:51:29] Cathy McPhillips: And just like things like that, like what, and she said it, she's like, that was the best interview I've done because I actually was able to focus on you and not trying to do a couple of things at one time.
[00:51:37] Paul Roetzer: So it's interesting. I still type my notes while it's happening, even if Zoom is transcribing it for me and sending me a summary because I can't break from the fact that when I type it, I remember it.
[00:51:49] Paul Roetzer: And. Like, I don't know. It's one of those things I can't get away from yet. I literally, it's like every conversation I have, I still take notes on the conversations and I [00:52:00] do feel like. Putting it down in words is actually what puts it into my memory versus just rereading it. I don't know. That's an interesting one.
[00:52:07] Paul Roetzer: Like I don't really thought about that, but I do use the zoom companion, but I don't, I still take the notes. And even if I know there's three other people on our team on the call taking notes, I still take my own notes. You've probably noticed that. Like I take notes for everything. Yeah. Interesting.
[00:52:24] Paul Roetzer: Well, those are our 15 questions. Yeah. Hopefully this was really helpful for everyone. we'll, we'll probably try and do one of these each quarter, at least half in 2024. We're going to continue on with the Intro to AI class. The next one is Cathy January 11th, January 11th. That's when I started dating my wife.
[00:52:42] Paul Roetzer: January 11th, 1995. We started dating when we were in high school. Random, but random. No, that is an important milestone in my life. So, okay, cool. Well, we appreciate everyone being with us, not only for this episode, but throughout the year, the podcast has grown. [00:53:00] exponentially this year, it's pretty remarkable and not only grown, but like the feedback we hear is incredible, like so many people reach out and share, you know, the impact the podcast has had on them.
[00:53:12] Paul Roetzer: So we're very appreciative of that and definitely going to continue next year with our weekly, format. And we'll probably start mixing in some more special additions that are like deep dives into certain topics and Q and A's like this. so yeah, so thank you for being with us. Thanks for being part of the community.
[00:53:28] Paul Roetzer: We hope you and your families have, happy holidays and a wonderful new year, and we will be back January 9th, Cathy, I think is the next podcast.
[00:53:37] Cathy McPhillips: January 9th. So if they're watch Paul's LinkedIn over the next few weeks, January 9th, there'll be a doozy.
[00:53:43] Paul Roetzer: I don't even know how we're going to cover it all.
[00:53:45] Paul Roetzer: And I was thinking we need to do it a year in review. We did like the mid year in review. We may need to do a special edition year in review in, early to mid January and kind of look back at 2023 and. The milestones, which might be a need to be a three hour episode. I don't even [00:54:00] know.
[00:54:01] Cathy McPhillips: That was a great episode. And I compare that to the one you, you did about Sam Altman, because you went through so many things and tied them all together. So me as a marketer, that helped me kind of understand why all this stuff mattered. So I'm looking forward to that one.
[00:54:14] Paul Roetzer: Well, if I wasn't taking the next two weeks off, I would do it now, but we'll, we'll look at early January and we'll, we'll recap that.
[00:54:19] Paul Roetzer: It'd be fascinating probably to go back and look at everything. With your fresh, relaxed brain. Yeah, seriously. Give my brain a little time to relax. All right. Well, thank you everyone. Thank you, Cathy. We will talk to you all in the new year. Thanks everyone.
[00:54:31] 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 www.marketingaiinstitute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.
[00:54:53] Paul Roetzer: Until next time, stay curious and explore AI.[00:55:00]