Navigating the AI Transition: Strategy, Adoption, and the Future of Work

[Paul Ritzer]: Based on all the conversations I have had with leaders of major companies, I have yet to find one that's prepared for it, and that worries me a lot. Welcome to AI Answers, a special Q&A series from the Artificial Intelligence Show. I'm Paul Ritzer, founder and CEO of Smarter X and Marketing AI Institute. Every time we host our live virtual events and online classes, we get dozens of great questions from business leaders and practitioners who are navigating this fast-moving world of AI. But we never have enough time to get to all of them. So we created the AI Answers series to address more of these questions and share real-time insights into the topics and challenges professionals like you are facing. Whether you're just starting your AI journey or already putting it to work in your organization, these are the practical insights, use cases, and strategies you need to grow smarter. Let's explore AI together.

Welcome to episode 206 of the Artificial Intelligence Show. I'm your host Paul Ritzer along with my co-host today, Kathy McFillips, our chief marketing officer at Smarter X. Welcome back, Kathy.

Kathy McFillips: Thank you.

[Paul Ritzer]: If you are a regular listener, you know that we do these special AI Answers editions in addition to our weekly episodes, and Kathy is my co-host for these. So if you're expecting to hear Mike's voice, tune in for episode 207, our next weekly episode. AI Answers is a series we introduced probably about a year and a half ago now because this is our 17th episode. We do about two a month. The basic premise here is we teach two free classes every month. If you haven't attended them or if you've got someone in your organization that's trying to just figure this stuff out, it is a great entry point for them. We teach Intro to AI every month. We've been doing that one for number 57.

[Paul Ritzer]: We've been doing that one every month for 57 months. You can do the math on how many years that is. Probably close to 60,000 people have registered for that class alone. It is a great way, like I said, just as an intro level for everybody. We do a 30-minute presentation, then we do about 30 minutes of questions. Then we do the same thing with "Five Essential Steps to Scaling AI," and that one's more for the leadership level. Both of those are completely free. They're done through Zoom webinars. We will put the links in the show notes if you want to attend one of those. We've got two of them coming up in April.

Each time we do that, we get dozens of questions. Kathy and I usually get through maybe 10 to 12 of those questions in the live sessions. So what we do with this AI Answers podcast series is we take the unanswered questions and we curate them. Sometimes we'll handpick some of the best ones from the webinar as well because it's good to repeat those answers for people who weren't there. We go through those. Claire and Kathy on our team go through them, they curate the questions, they put them together, and then Kathy sends me a link five minutes before we're getting on. I have not looked at these questions. We just answer them the same way I would live, unscripted and however I'm feeling at that moment. If I don't have a great answer to it, I'll be honest with you and tell you that. But we do our best to just try and provide as much context and perspective for the non-technical audience. If you're new to the show, most of what we do is try and cater to the actual practitioners and business leaders on the non-technical side. We're talking to marketers, sales people, customer success people, company leaders, ops, finance, things like that.

The questions today that we're going to go through are from our Scaling AI class, the 15th version of that class that we taught on March 18th. We're recording this on March 25th. Sometimes that's relevant depending on what the question is. I may throw in something like current event news that I haven't even talked about on the podcast yet. So that's it. Am I missing anything there, Kathy, on format?

Kathy McFillips: I don't think so.

[Paul Ritzer]: All right. This episode is brought to us by AI Academy by Smarter X that helps individuals and businesses accelerate their AI literacy and transformation through personalized learning journeys and an AI-powered learning platform. New educational content is added weekly so you always stay up-to-date with the latest AI trends and technologies. Our AI for Industries collection is one of the great features of the platform. We have six course series right now with professional certificates of completion that are designed to jumpstart AI understanding and adoption across industries. There are six: professional services, healthcare, software and tech, insurance, financial services, and retail and CPG, which actually just launched last Friday. These series are an ideal launchpad for organizations that want to level up their teams and accelerate AI adoption and impact. Individual and business account plans are available now. You can also buy single course series for one-time fees.

If you're not ready to do the AI Mastery membership and do the annual program and take advantage of all those, you can just do an individual thing. I will say on the business account side, it's actually cheaper to buy the annual thing than it is for the single course series. But if that's what you're looking for, those are both available. Go to academy.smarterx.ai. You can learn all about not only the industries series, the department series, the foundations collection, AI Academy lives, GenAI app reviews, everything that's a part of AI Academy. Check that out again at academy.smarterx.ai. Okay, Kathy, let's do it. We're going to try and do this in about 40 minutes, 45 minutes to get through. Looks like we have 15 questions. I'm going to try and be efficient with my answers today.

Kathy McFillips: Real quick on the academy stuff, I just think of my agency days. I would have loved to have taken the professional services obviously, but then I had a CPG client, a financial services client. That would have been so helpful.

[Paul Ritzer]: For context, I owned an agency for 16 years. Kathy also comes from the agency world. She did her own business for a while as well. Both of us are very deep in that. When you're in the agency world, you need this depth of knowledge across industries, across your client portfolio. It is a great thing for agencies as well as people who already work on the corporate side within the brand side within those industries.

Amazon’s AI Rollout and Market Maturity

Kathy McFillips: Okay, question number one. We've been talking about Amazon slowing parts of its AI rollout due to quality issues. My gut says that might actually be a sign of maturity, but how do you see it?

[Paul Ritzer]: This question I think is tied to Amazon having issues recently with AI agents kind of going rogue and doing some things they shouldn't have done. I'm thinking that's the context of this question. I think there's just growing pains for everyone right now. Not just on the brand side, not just the people who are trying to figure out how to do just the basic stuff within a corporation and drive pilot use cases and get an AI council going and get support from the C-suite. We're all fighting that fight, but even the tech companies themselves—the big frontier lab companies, the cloud companies like Amazon—they're all trying to move really fast. The tech is advancing so quickly and the capabilities, like the ability to build these agents and these agent swarms and give them access to files and the ability to make decisions and take actions, it's really messy.

We had an issue we talked about on episode 205 where Meta had a similar problem where the agents just sort of went rogue and did something crazy. I do think that it could be a sign of maturity, but I think it's probably more likely a sign of everyone moving really fast and trying to keep up with the competition, especially in the technology industries. It's hard and there's lots of unknowns about this tech, but you don't want to be caught sitting back and not experimenting with it. I think it's more that people have to realize how to experiment responsibly and safely within their enterprises and that's an ongoing learning curve. It's why all of us need to lean on our technical partners within the organization or outside consultants who can make sure as we're experimenting with the latest and the greatest in AI, we're doing it in a way that doesn't put our data or our companies at risk.

Kathy McFillips: I think back to when ChatGPT came out and the other models were like, "Wait, we have something too," and they rolled it out when they weren't quite ready. They were iterating in real time and I think just the risk is so much greater now.

[Paul Ritzer]: Especially with the agent stuff. A lot of the generative AI—just being able to create things or use reasoning models—the human is still very much in the loop and in control. As we start to experiment with these things like OpenDevin or even Claude where you're giving it access to files or Claude Computer Use to a degree, you're giving it access to a bunch of stuff and we don't really understand how these agents do things or why they do things. It's just a whole other surface area of risk and it's why most enterprises are going to be very slow to move aggressively into this space.

Large Enterprises vs. AI Natives

Kathy McFillips: Do you think large enterprises are structurally disadvantaged in the AI era or do they have assets that will ultimately let them win?

[Paul Ritzer]: This is a mix. I wrote in 2023 an article called "The Future of Business is AI or Obsolete." In that article, which we can put a link in the show notes, my theory was there's going to be three types of companies in the future: AI Native, AI Emergent, and Obsolete. The AI Natives are built smarter from the ground up. They don't have legacy systems or legacy talent that they have to convince to use AI. They aren't stuck in legacy pricing models. They have all these advantages where they can just use the smartest tech and they can build the organization on the fly around what it enables.

The AI Native idea is you look at an opportunity in an industry and say, "We can build a smarter version of that company." Take a law firm, a marketing agency as we were just talking about, or a software company and just say, "Let's just build from the ground up with fewer people, be more efficient and use AI in everything we can do." Then you have the AI Emergent companies, which is basically everybody else—all the other existing companies who now have to figure out how to adapt and how to change their pricing model. If you're in a services industry and you're charging by the hour, it doesn't work. You can't do that. You're just going to completely undercut yourself and destroy your financial model.

If you have legacy tech that's hard to move people off of or it's not AI native and you're trying to force-fit AI capabilities into an existing software stack, it's really hard to do. If you have customers—we experienced this when I started launching AI services back in 2017 at my agency—we had all these legacy customers who wanted nothing to do with it. They didn't understand it. They didn't get why we would be building AI capabilities into a service company like we were doing. You're known for something, and to try and change that perspective and become known for something else is really hard in any industry.

The AI Emergent, though, they have talent, they have a customer base, and they are more likely to have financial strength. If they can move fast enough through a combination of vision from leadership and then a strategic approach to change management, they can push off the AI Native competitors that are going to emerge from everywhere. But it's hard and we're definitely seeing a lot of organizations struggle. We're starting to see turnover at the top. I think you're going to see a lot of that. I think you're going to see a lot of leadership in the C-suite in particular that just aren't getting it or aren't moving fast enough with a high enough sense of urgency, and I think it's going to cause a lot of shifts. We also see it in the stock market when you start looking at the valuations and the market caps of these companies that just haven't figured it out yet. Even Apple is a great example. Now, Apple's somehow managed to keep their stock price relatively strong, but it's just an organization that has really struggled and they have everything you could ever want from an AI Emergent company: more money than anybody, incredible talent, and an amazing brand that we all are envious of. And yet for three years now, they have yet to figure out how to infuse AI properly.

Ownership of the AI Problem

Kathy McFillips: If we say the bottleneck is something like adoption or data readiness, who actually owns that problem inside an enterprise and why hasn't it been solved yet?

[Paul Ritzer]: Part of the reason it hasn't been solved is because nobody knows who owns it. I think what happened—and I say I think, but I can say confidently I know this in many enterprises—when GenAI showed up in late 2022 and then we got GPT-4 in spring 2023, it started becoming very apparent to enterprises that this is a shift in not only consumer behavior, but it's going to be a shift in the way we do everything from the ideation of products to the marketing of those products to our sales and our success. All these things are going to shift and a lot of C-suite turned to the IT department, turned to the CIO, and said, "Go figure this out. This is a technology problem," which it wasn't at a macro level. It is part technology problem, but it was treated as a pure technology problem.

What then happens is they didn't take the initiative to educate and empower the leaders of each of the different business units or teams within an organization and then let them democratize the ability for them to then build their plans, figure out what tech stack they needed, and figure out how it was going to evolve their org chart. You have people like chief marketing officers, heads of sales, chief customer success officers, CROs—if they don't understand AI deeply and aren't using it themselves and becoming very competent in it, then they can't own the diffusion of it across their departments and teams and business units.

I think that's largely what happened. We had this lack of adoption in part because we didn't prioritize literacy and competency of the tools. The data readiness is a separate but related issue because some of the most important, highest-value uses of AI in enterprises are going to require clean data that is infused into the AI processes and workflows. But that's often a red herring in terms of why adoption slows because what happens is the IT department will say, "Well, we're just not ready. We got to get the data. Make sure it's safe. You can't touch this because it puts this data at risk." When in reality, in most organizations, take a marketing team as an example, 90% of the use cases they would tackle in the first 12 months have nothing to do with the data. You don't even need any data access. That's a misconception I see time and time again when I talk to enterprises. They think they have to solve data first and they don't. It can happen in parallel while you're stacking all these use cases that don't touch the data.

The AI Divide and the Future of Employment

Kathy McFillips: I keep thinking about this idea of an AI divide inside companies between power users and everyone else. Are you seeing that too? And what actually happens to the people who don't keep up?

[Paul Ritzer]: We see this every day. This is a major problem. I'll try and give a tangible example. Let's say our AI Academy is a good example. We will have companies come in and they'll buy, let's say, a hundred licenses for their team, one of the divisions of the company. You buy a hundred licenses that gives them access to all this education: piloting AI, scaling AI, fundamentals by industry, by department, all this stuff. It's sitting there and they can learn it in a short time and become highly competent with AI. And yet, if you take that segment of a hundred people and you start breaking it down, there's going to be 20 to 30%, or whatever the number is, of people who hate AI, want nothing to do with it, find it threatening, find it abstract, think it's going to destroy the environment—they have some reason why they want nothing to do with it.

Then you're going to have a middle-of-the-road group who are like, "Yeah, I'm dabbling in it. I'm using the co-pilot a little bit the company gave us, but mainly for summarizing meeting notes and doing some emails." They would answer the question "Do you use AI regularly?" as "Yes, I use it weekly." You're going to have this misconception that they actually know what they're doing when in reality they're just doing these surface-level things.

Then you're going to have a portion of the company who are racing ahead. The day you give them access to Academy, they're in there, they're doing their first three certificates in the first week. You give them a co-pilot license and they are daily active users grinding all day. It's doing all kinds of amazing things. They are racing ahead, becoming infinitely more productive than their peers while still getting paid the same as their peers, by the way. They're actually leveling up the organization because they're uncovering all these use cases and ways to infuse AI that drive efficiency, productivity, and innovation.

The difference becomes you have these people who are intrinsically motivated to solve AI even with all its challenges and the negatives it's going to have on the economy and jobs. They're just like, "Okay, we get that, but let's figure out how to do this in a responsible way." Those power users really start to separate themselves in a way we almost haven't seen in a very long time. The only thing I can think of is back in, say, 2000 when the internet really started to take off within corporations. You had people who figured that out, knew how to use Google, got really good at doing email, and the people who didn't, who refused to do those things. That's the closest analogy I can probably get to, where you're just going to have the people who do it and the people who don't.

What we are seeing, what I've heard behind closed doors many times over the last two years and what we're now seeing happen publicly, is the people who don't won't have jobs. It is one of the hardest realities and I don't want to say it in that way to seem crude or inhumane about it. The reality is if you run a company and you know that a tool or a capability enables that company to grow more efficiently, to accelerate its growth, and you have people who refuse to use it, they won't be employed at your company anymore.

What we're telling leaders is: give them a runway. Tell them that that's the case. Don't just spring it on them in three months and say, "Okay, everybody who said on the survey that you didn't like AI, you're no longer employed." No, tell them from the CEO on down, "We are going to move in this direction. We are going to become an AI-forward company. We're going to empower you with GenAI applications. We're going to give you ChatGPT, Google Gemini, Anthropic Claude, Copilot, whatever it is. We're going to give you the tools. We're going to personally train you on those tools so you know the use cases that are more valuable to you. We're going to provide AI Academy courses to you so you can go through and take this training and do the live events and learn every week. We want you to listen to the Artificial Intelligence Show podcast. We're going to tell you the blueprint to become more valuable in this company and you're going to be assessed based on it. It's going to be part of your performance reviews annually."

Now, if you've done that and you've clearly integrated it into the business and they still don't do it, then there's nothing you can do as a leader but help transition them to somewhere else where they would prefer to be because they obviously don't want to be there. I think that's the only way to do it. I don't think most companies will take that human-centered approach to it. I think a lot of companies are just going to cut people. But my hope is more and more companies take the approach of at least being transparent about what you want from them. Set expectations clearly, give them the resources to meet those expectations, and then if they don't do it, there's nothing you can do. Replace AI with any technology advancement of the last 50 years, the same thing would be true. "You're a salesperson, we're going to give you Salesforce." If for 12 months the person refuses to use Salesforce and they still manage their sales leads in an Excel spreadsheet that no one else has access to, you're fired. It's not just an AI thing. It's empowerment of tools and education. If you choose not to take advantage of that, then you don't have a job there.

Kathy McFillips: And you said we must be doing this because our competitors are doing it.

[Paul Ritzer]: Enterprises have a really hard time with this because you're going to have a large percentage of your employees who either think it's too abstract and technical, they don't like it, or they find it threatening. AI Native companies are like, "We're only hiring AI-forward people. We're not even bringing you in unless you already listen to a podcast, you work with ChatGPT daily—you're not getting a job here unless this is you." In that case, the AI Native companies have a massive advantage from a hiring and development of talent perspective.

Kathy McFillips: We talked about email and Google back in the day. It made me think we don't need inter-office mail like we talked about this week.

[Paul Ritzer]: We were at a gala together on Saturday night for my kids' school and this came up about inter-office mail, which I totally forgot about. But yes, that was a job. It's not a job now.

Disagreeable Takes and Automation Risks

Kathy McFillips: What's an AI take you have right now that most people would disagree with?

[Paul Ritzer]: My take for the last two years was that we were going to lose millions of jobs. Most people, including leading economists, argued with me about this and thought I was insane. I think people are coming around to this idea, but I do still get a lot of pushback on this—nowhere near as much as I did six months ago. I think in the end AI is a net positive for the economy. I'm not sure exactly how from a jobs perspective because I think there's just going to be fewer jobs. I feel like we're going to go through a very challenging period from an employment perspective, both unemployment and underemployment. I actually am more concerned about underemployment. Meaning your kid graduates college in May and they take a job at a retail store even though they have a double major in economics and marketing because you just have to get out into the world and start making a living. I think a lot of jobs are going to be hard to come by. I get less disagreement now than I used to, but I think the next few years there's just a ton of unknowns about how this plays out. Based on all the conversations I have had with leaders of major companies, I have yet to find one that's prepared for it, and that worries me a lot.

Kathy McFillips: Is there a world where companies look back and feel like they automated too much too fast? And has that already happened?

[Paul Ritzer]: I think it's going to happen all the time and it is probably just going to be part of business moving forward. I think there's always going to be this push to the limit of what we think this AI can do and then a realization that it couldn't do that. One prominent example we've talked about on the podcast is Klarna where they're like, "Hey, we're never hiring people again. We're just going to do everything through AI agents. All our customer success is going to be agents." And then now they're hiring a bunch of people. You have OpenAI who you would think would be the perfect example of needing as few people as possible. We just talked on episode 205 about the fact that they're planning to double their staff from like 4,500 to 8,000 in the coming months.

I think a lot of companies are just going to try real hard to use this and then there's going to be a pullback. Another example I could think of is companies that race ahead to use AI avatars because they're cool and they save you time and people can just talk to the AI avatar in a customer success call, or you build your online learning with AI avatars instead of the human. I think that's going to snap back fast. I think a lot of humans want to know they're actually talking to a human or hearing from a human. While it's fun and efficient to push the limits of this tech and try all these ways to do automation, at the end, I think a lot of it's going to fail. We're going to fall back to the importance of the human element of business. I think that's just going to be a constant learning curve for organizations as they experiment.

The ones who are out on the frontiers of this trying all the things, we're going to learn a lot of lessons from them and they're going to be painful lessons for them and hopefully others. A fast follower is probably where most companies want to be here. I don't think very many are going to want to be on the true edge. I saw a tweet last night from Andrej Karpathy that there was a Trojan horse put into this codebase that was downloaded like 95 million times and it ended up exposing all this stuff. You had all these developers using this thing and it created this massive security risk. Again, if you were out there and you were doing the thing and trying all the open code stuff and all these new agent things, it sounds great in an abstract sense that you're doing these things and then all of a sudden it's like, "Oh, probably shouldn't have done that." That is why IT and legal, as much as they can be roadblocks, you've got to work in alignment with them, especially as it comes to using these frontier technologies. There are lots of risks ahead.

Even very simple things—we turned on a chatbot a while ago and we're like, "Turn it off." We had to go through a lot of due diligence before we could turn it back on. It seemed on paper to be right and then it wasn't. We had to self-correct on that. Other things, we're un-surfacing all these ideas on ways we can streamline our processes with our free classes. Jeremy, who's heading up marketing for Academy, he and I have been working on this and had this great idea and we tried it and we're like, "Stop." I love the ideas, but we need to make sure that they're ready to actually roll out.

Augmentation vs. Automation

Kathy McFillips: We've talked before about augmentation versus automation. Do you still think we land somewhere in the middle or does automation eventually take over?

[Paul Ritzer]: It's not going to be evenly distributed. I think it's going to depend on your role, what the tasks are that make up your role, what workflows you do on a regular basis, and what industry you're in. Is it in a highly regulated industry? I think everybody's going to experience the augmentation versus automation spectrum differently depending on what you do. For me as a CEO, I would say like 95% of the way I use AI is augmentation. It truly is just an enhanced strategic partner. It's enabling me to think more intelligently, to think more broadly about any implications of decisions before I make those decisions. There are definitely pieces of my work I'm automating, but for the most part, it's really just enabling me to do more work better and more thoroughly.

I think of it that way, but then if you move down the channel in terms of the roles within an organization, I think entry-level is going to probably flip. It's going to take 90% of what an entry-level person would do and automate it. I think augmentation at the senior level is the more likely scenario, while automation is more likely at the entry to mid-level where you do the tactical work. I think maybe that's a distinction: if your job is to do tactical work as part of a strategy or a campaign, a lot of that tactical work, regardless of what industry you're in, is going to be automated with minimal human-in-the-loop in the coming one to two years. Senior-level people are going to be using it more as a strategic thought partner, assisting in decisions and problem-solving, building strategies, and stuff like that.

Kathy McFillips: If we fast forward three years, what does the average knowledge worker's job actually look like because of AI?

[Paul Ritzer]: If I had the perfect answer to this, I'd be in a different financial state. This is the multi-trillion dollar question. Nobody really has the answer to this. It's why OpenAI and Google and Anthropic are hiring economists. They're trying to model this stuff. It's why you can look to places like the Brookings Institution; they do some really good stuff in this area. Some of the content we're actually going to plan for MAICON is going to focus on this future of work and future of economy kind of stuff. My keynote at MAICON—I don't think we've announced it yet, so I'm not going to share exactly the plan because I'm not 100% sure how I want to announce that.

I have a working hypothesis of what I think the future looks like. To be quite honest with you, it was literally at a bar on a Friday night two weeks ago where I was picking up food for my family and I had this thought of what I thought it would be. It was based on a couple of conversations we'd had that morning, a meeting Kathy and I had actually been in that morning with a leader of a major university, and we were talking about the future of work for students coming out of college. Something else happened that day that made my mind go to this. I developed this hypothesis of what I thought the different roles in an organization would be. I wrote it down in five minutes and then I actually sent it to the team while I was sitting at the restaurant and I was like, "Hey, I think this might be the MAICON keynote, but I have to play with it a little bit more and think it through myself."

Without divulging the whole concept, I think at a very high level you're going to have leaders with extensive experience and expertise who oversee agents and a swarm of agents and a team of people. Those leaders can do most of the work that their lower-level employees, associates, and things like that used to do—the tactical stuff. It goes back to the previous automation versus augmentation. If I'm the CEO and I want to launch a new product, I can build the product myself in Claude code. I don't need to hire designers and developers; I can actually go in and just do the thing in 20 minutes. Then if I want to launch that, rather than turning it over to the marketing team and saying, "Okay, go build the landing page and write the emails and do all the things," I'll just tell Claude to do it. "All right, great. We're locked in. This is what it's going to be. I want to launch it in 30 days. Build me a game plan to launch this thing. Here's all the past game plans we've done." Great. Three minutes later, I have the game plan. "Great. That looks awesome. Let's go start building all the components to it." Build all the components, finish it up, package it, turn it over to the marketing team, and say, "Here you go. Got drafts ready to go. You guys do your thing now and edit, vet it, do whatever."

Imagine that scenario for all knowledge work. There's nothing stopping a senior-level person from doing the work of the lower-level people, especially as the AI models start to learn that business and learn the preferences of that senior-level person. You don't have to teach it to the entry-level people. Then the question becomes: what happens to the entry-level people? That's the part I think I may have cracked the code on, but I'm not ready yet to explain it deeply. What I'm trying to solve for is how we create entry-level employment at scale when the senior people can do the work of those entry-level people. That's the fundamental problem statement I'm trying to solve for and I think I have a direction. To be continued, but I think the first part is the way I just explained: senior-level people doing most of the tactical work themselves and then turning it over to people to get it to the finish line versus building the strategy and then hoping someone else figures out how to do it.

Kathy McFillips: Getting that idea out of your head is a very valuable part of the process. If AI can help you get to that point versus passing it off to somebody else, that's huge. Katie Robbert from Trust Insights and I try to talk once a month and we're about five months behind. We talked on Wednesday and it turned into "I have notes from Claude" that she was talking about and I was like, "I can't wait to get in there."

[Paul Ritzer]: I have one right now I'm working on that I've been trying to do for at least two years and it's a visual thing and I'm not the best at it. I've tried different drawings. I've worked in Freeform on my iPad sketching it out on flights. I've got paper drawings. I've tried every different way to do this. It's a very complex idea that needs to be simplified into a simple interactive visualization. It came up again roughly in one of the meetings we were in yesterday, Kathy. Last night I'm literally just like, "Okay, I just need to write this as a prompt. I need to explain the challenges I've tried, the reasons why it's not been working the way I've tried to do it." I tried to work with a designer to do it and they didn't get it either. I don't know what it needs to be, but I'll know it when I see it.

I started working on this prompt last night. I continued this morning and it's all I can think about right now. In a perfect way, I'm going to finish it by the end of the day. But it's one of those where once you get the prompt written, you literally just give it to Claude and ChatGPT and then you just sit back and pray for three minutes. Maybe it'll do it, maybe it'll nail it, maybe it's a one-shot thing. I'm going to give it this prompt and there it is. This happened to us last week with a couple of visualizations I was building for the team retreat. By the time this podcast episode comes out, I may be the happiest person in the business world tomorrow if it works. It'll be transformational for me and maybe for the company and maybe even for some of our AI Academy customers. I have no idea; I might put the prompt in and it could take a while.

This is an example of what the future of work looks like: I don't need all those people. I just need them to get it to the finish line once I have it. I have endless of these things. I have a sandbox of five of these things a day I would love to bring to life that I just didn't have the resources or capacity to do before.

Strategy and Literacy

Kathy McFillips: What are companies still getting fundamentally wrong about AI strategy right now?

[Paul Ritzer]: They don't have one. I don't ever want it to feel like I'm saying AI education is the core just because we offer AI education—it's what we do as a company—but even if I wasn't doing that, if I was just consulting and speaking, I would lead with AI literacy every single time I gave a talk. It's the most fundamental thing to understand. How can you build a strategy if you don't understand the technology? If you say the Chief Marketing Officer is in charge of the AI strategy for the marketing team, great. If that CMO is using AI as a chatbot and an answer engine and has no idea of its reasoning capabilities, has never done a no-code app development project, never done deep research, doesn't understand multimodalities and the ability to do video and image and how the social team can be using all that—if they don't know those things, how in the world are they ever going to build an optimal AI strategy of the people they need, the technology they need, how to reimagine workflows, and what the future of the org chart looks like? The way everyone gets AI strategy wrong is that they don't start with AI literacy. I can just stop talking there; it is literally the answer to almost every flawed AI strategy: they didn't start with a deep understanding of the technology itself.

Kathy McFillips: One of the biggest challenges I hear is just keeping up since everything seems to change weekly. How should leaders decide what actually matters versus what's just noise?

[Paul Ritzer]: I think about this a lot. This is actually related to the visualization project I was just alluding to. I think that when I was designing the product roadmap for AI Academy, when I reinvented what our AI Academy was back in fall of 2023, this was actually the problem I was trying to set out to solve. What is the fundamental knowledge everyone needs to know? What is the base level of understanding they need to have about artificial intelligence so that they can then figure out how to apply it to their department, their personal role, and their industry? Then how do they keep up with the stuff that's relevant to them?

If you look at how we've structured the learning journeys and the collections within our Academy, this was meant to answer this question. I'm going to start with taking the Foundations collection. I'm going to go through fundamentals, I'm going to take Piloting, I'm going to take Scaling. Then I'm going to take AI for Marketing because I'm in marketing. If I'm in the insurance industry, then I'm going to take AI for Insurance. And the GenAI apps that drop every week—when one pops up that's relevant to my job, I'm going to take 20 minutes and watch that GenAI app review. I'm going to attend some Mastery Live classes because I want to do the AMAs and be able to ask some questions. I want to get a trends briefing every quarter. That was how we thought about it: what is a learning journey you need to go through?

I think everybody has to figure out their role. It doesn't mean having to buy Academy from us. Come to the free Intro class; start there if you need to. Listen to the podcast each week. That'll keep you updated on what the 10 things you actually need to know each week are. Find a couple books that are super relevant. Find some people to follow on LinkedIn or X that you really trust. Find a couple other podcasts. You can do all this for free. That's the beauty of all this. You can even go take some courses on LinkedIn Learning or Coursera. You don't have to just do our stuff. Our stuff is meant to be complementary to whatever else you need to do to learn.

All of us learn in very different ways. I used NotebookLM last night with my son. He was studying for a Spanish test and he needed to—I would give him the Spanish word, which is funny because I took Spanish but I can't pronounce a lot of Spanish words. I'm trying to say the word in Spanish then he's supposed to tell me what it means in English. So I instead went into NotebookLM. I took a picture of the thing he was studying. I said, "Make me flashcards where we want you to show me the Spanish and then turn it to English." And it did it. We prepared for his quiz through flashcards. I think that's the kind of thing: you have to understand what the tech is and then you have to find the ways you best learn. In some ways you're using the tech itself, like NotebookLM, to create quizzes, flashcards, mind maps, whatever you do.

You need to think about a personalized learning journey for yourself. You need to think about what education you have access to and then what are just those free amazing resources that are going to keep you up to date and on the leading edge of where you want to be. It's not for everyone to try and consume everything every week. It's a lot. There are times I've said on the podcast that I personally get overloaded by it. I have days where I want it to stop. I want to not think about everything today. I don't want to think about the political implications, I don't want to deal with the negative impacts on humans. But I do because that's my job. But I get it if you can't do that or you don't want to do that. You want to dip in, know what you need to know, and then get out and go on with your life. Everybody's got to figure out what their goal with this stuff is, and then you can adapt your learning journey based on that.

Rapid Fire: Councils, Governance, and Leadership

Kathy McFillips: Okay, we're starting a new segment on AI Answers called Rapid Fire because we have five questions left and not a lot of time. We hear a lot about AI councils, but I've also seen them slow things down. What separates the ones that drive progress from the ones that don't?

[Paul Ritzer]: I teach an entire class on AI councils. There's an amazing channel within our Slack community of people who have built AI councils within their organizations. Every council is different. I think it all probably starts with what is the mission of the council and what's within the charter—their responsibilities and their goals and how it's governed. You have to contemplate that from the beginning of the formation of the council or if you're trying to evolve a council. What do you need to have to actually move the organization forward? If you find the overall council is slowing things down, maybe you can split off a subcommittee that's focused on a specific thing that isn't hindered by the politics of the overall council.

It's like anything else, especially in an enterprise. Everything gets bloated and too many people get involved and there's too many meetings and too many emails and nobody actually owns anything or has the responsibility to move anything forward. You just got to try and avoid that. However you do it, whether it's splitting off a Center of Excellence that is allowed to be more innovative and take some more risks because you can test things in a sandbox, you've got to find the thing in your organization that allows you to keep moving it forward. If you find the council is slowing things down, then find a spin-off of that that allows you to keep moving.

Kathy McFillips: Where do you think governance is necessary right now and where is it actually getting in the way?

[Paul Ritzer]: This isn't probably going to be a complete thought, but the two things that came to my mind the second you were reading this were: anywhere that touches data and anything that has to do with agents automating outcomes, actions, and decision-making. If you're at a point where you're actually allowing agents to do things—whether it's a customer interface where it's the chatbot and it's giving recommendations and actions, or personalization of emails based on behavior and you're just letting the AI write the emails and send them—you need governance when it touches high-value instances or stakeholders that are important to the organization. The more prominent the use case is, the more it accesses data, the more it has decision-making or autonomy, the higher the risk it becomes and the more you want to welcome that governance. You want guidance on how to do it properly because you do not want to screw that up. We're in a whole new world where we don't even know what the precedents are around liabilities and insurance. All the ecosystem and infrastructure is being built around this. You've got to be safe when the use cases call for responsibility.

Kathy McFillips: One of the things that scares me the most is the relationships that I've built and then setting up automations that would come across as an automation to people that I have relationships with. I'm probably being overly cautious, which I don't think is a bad thing.

[Paul Ritzer]: I don't think it is either. I wrote this in one of my books—I have no idea if it came from somewhere else—but a brand takes a lifetime to build and a moment to lose. That could be a personal brand or it can be a business brand. You could invest in relationship building and spend time on calls with someone like Katie Robbert, and then Katie gets some crappy automated email. She's going to give you the grace of understanding what you're doing, but if you think about your customer base as a whole and you try and install some new AI-powered personalization and that goes haywire, you could start to chip away at that brand equity and trust. That terrifies CEOs and CMOs. The company is based on that brand trust and if you ruin it, what do you have left? I think people give a little bit more grace as we move forward when they realize companies are experimenting, but that's a fine line and I don't know that you're going to know when you crossed it until it's too late.

Kathy McFillips: When you're showing AI to leadership, is it more important to show the system itself or real outputs and results?

[Paul Ritzer]: This goes back to knowing your audience. Who is the CEO and what is their familiarity with AI? Do they care? Have they already given resources and support to AI initiatives? Are they still trying to be convinced that it matters? You've got to know your CEO and what is actually relevant to them. At a general level, if we approach it as a CEO who is skeptical of what AI can do and the need for urgency, show the results. How you did it is not as relevant to them. If you say, "Listen, we've been working on this initiative using some new capabilities and we took this thing that used to be 50 hours a week for the sales team and we condensed it down to seven minutes. The outcomes are actually better value. Here's what it looked like before, here's what it looks like after. We think we can save the company 700 hours a month, which equates to this amount of value, and we think we can redistribute that to launch two new products next quarter." Sold. I don't even care which tool you're using to do it. Now, if that CEO is like, "That sounds amazing. Show me how you did this," great. Now show the demo. But if you lead with, "Yeah, we're doing this cool AI thing. Let me show you some prompts," and they're skeptical, you've got to know the leader.

Kathy McFillips: What's one AI use case that feels like a no-brainer at this point, but most companies still have not implemented?

[Paul Ritzer]: Strategic thought partner. If you're not using the thinking versions of ChatGPT, or if you're not playing around with reasoning models—depending on which platform you have a license to—you need paid licenses. Pay the 20 bucks a month. If you're using Gemini Pro, Claude 3.5 Sonnet or Opus, and then ChatGPT with the thinking models, you've got to use the reasoning models. They're a cheat code. If you're not using the reasoning model, you're just leaving so much intelligence on the table that you're not applying to what you're doing. Using it as a thought partner to help you with decision-making, problem-solving, and strategy building is an absolute game-changer. It is the dominant use of how I work with AI every single day.

I work across three models every day. If it's a high-value situation, I will put the same starter prompt into ChatGPT, Gemini, and Claude. I will monitor the outputs of them. We're working on a project right now that we met about yesterday. I put a high-value prompt into six different models. I actually tried variations. I did Claude Opus and Sonnet. I did Gemini. I did ChatGPT with and without my custom instructions. I will try everything on these high-value ones and then the model that seems like it's best suited for that use case will become the dominant thread. Then I'll use the other models as a critic to test the outputs of the primary model. I'll say, "Hey, what do you think about this final strategy?" and I'll let the other models critique the primary model. You might think that's taking a lot of time, but it's still saving you hours and weeks of your life. It might take me four or five hours to do it that way instead of 50 hours. Because we can just prompt something, we think we should just be able to get it done super fast. No, sometimes you have to finish the process and be patient and see it through.

Kathy McFillips: Last one. I still see people waiting to be told how to use AI instead of just experimenting. Why do you think that is and what actually works to change that behavior?

[Paul Ritzer]: It's human nature. That's just how we are. That's never going to change. Go back to 2020. How often were you saying, "Just Google it"? The answer is three seconds away by just putting it in the search engine. And yet 20-some years after the invention of the search engine, you had people who still wouldn't Google it. I think it's just human nature that there's just people who don't want to learn the new thing. They're not going to naturally experiment with it because it's abstract or because it's just not what they're comfortable with.

The only way I've found to change behavior is to show them a use case that changes their life in a positive way. Solve a pain point that they have that they didn't know how to solve otherwise or give them the ability to do something creatively they couldn't do before. Hold their hand through those first few use cases until they realize this isn't that bad. This is how you do adoption in an enterprise, too. When you assign Copilot licenses or ChatGPT, you should assign those licenses with the first three to five use cases baked in for the people that you're giving them to. Show the sales team how to use it to do SDR work, to segment databases, or write better proposals. Give them a GPT trained to do those things on the company policies and the brand. The easier you make it for people who still need to learn it, the faster you're going to get adoption and buy-in.

The example I always give is: no one wants to spend their Sunday night away from their family for two hours because they have to write the report that's going to get turned into the C-suite Monday morning that they know the C-suite isn't going to read anyway. All of us hate that. What if that's the thing you can take off someone's plate? "Hey Kathy, I'm going to give you your Sunday night back. We're going to build a GPT that's actually going to do this and we're going to set up an automation where it just emails it to you Monday morning. You edit it and then you just send it to me with the summary of what I need to know. How about that?" Do that five times for somebody and there's no way they don't start to experiment themselves.

Kathy McFillips: That is the one thing I remember from MAICON 2019 when I came as an attendee was sitting in Keith Moehring's session about automated reporting with AI. I left and I was like, "That's amazing."

[Paul Ritzer]: Funny enough, that is the service that we introduced in 2017 that our clients wanted nothing to do with. We were using a tool called Automated Insights. We were doing rules-based automation of analytics reports. Google Analytics reports storytelling that was automated was the first thing we built back in 2017 or 2018. Then Keith did that talk in 2019 about how we were doing it. We had zero clients paying us to use it at the time. It was shocking, but that's where we are. All right, another 15 questions through AI Answers and we're done right in time.

Kathy McFillips: Thanks, Paul.

[Paul Ritzer]: Thank you, Kathy. Thanks for joining us. We will be back with our regular weekly as scheduled. Just warning everyone in advance, I am on spring break with my family April 1st to the 10th. So we will not have a weekly on April 7th. We will be back with the weekly on April 14th. Maybe some amazing stories to share of our travels. I'm really excited. Thanks everyone. Have a great weekend.

Thanks for listening to AI Answers. To keep learning, visit smarterx.ai where you'll find on-demand courses, upcoming classes, and practical resources to guide your AI journey. And if you've got a question for a future episode, we'd love to hear it. That's it for now. Continue exploring and keep asking great questions about AI.

Key Takeaways