AI Answers: Your Questions on Careers, Strategy, and the Future of Work

A conversation with Paul Roetzer, founder and CEO of Marketing AI Institute, and Kathy McPhillips, Chief Marketing Officer at Smarter X, tackling key questions about navigating the fast-moving world of AI.

How can someone with a marketing background transition into the AI space without strong coding skills?

Increasingly, a coding background is just not necessary. In many ways, for people who are expert coders, AI is giving them superpowers. But for people who aren't coders, a world of opportunities is opening up in marketing. If anything, AI is enriching traditional marketing roles by enabling coding with natural language.

People who traditionally have no coding ability can now go into a tool like Google Gemini and perform coding tasks using simple language. There are tremendous opportunities to blend what marketers have always done and enhance it with the new coding capabilities of these AI models.

What are the best AI skills to learn while job searching?

The most critical skill is prompting—learning how to talk to the machine and treat it like a collaborator or a thought partner. Focus on your prompting abilities to enhance not only your creative output but also your critical thinking. Before that, however, you need a deep, basic understanding of what AI is capable of and the different features and apps within various platforms.

To become a stronger candidate, you need to show demonstrated competency. Use the tools. If you're unemployed, you can gain this experience by earning certificates to show you're continuing your education. More importantly, build things. Develop Gems in Google Gemini, GPTs in ChatGPT, or skills in Claude. These can be for your personal life—a health tracker, a trip planner—just do things with the technology.

When we interview candidates, we ask, "How are you using AI personally and professionally?" If you're job searching, you can focus on the personal side. Talk about building a guided learning tool to help your kids with homework. Build things that provide value in your own life to demonstrate your competencies. It’s also helpful to use AI in a subject you know really well. That way, when you see the output, you can learn how to better prompt it to get a richer, more accurate answer.

For consultants who bill by the hour, how should they think about time spent experimenting with AI?

The concept of billing by the hour is becoming obsolete because the value exchange is broken. When you hire an advisor—legal, HR, or any consultant—you're not paying for their hours. You're paying for the output and the value they create. We know they can work more efficiently with AI, and that’s a good thing.

If a consultant solves a high-value problem in 25 minutes using a great prompt, they shouldn't be paid for just 25 minutes. They should be paid because they solved the problem. The challenge is for people who have always charged by the hour. As the person hiring a contractor, I expect a certain level of AI literacy, competency in prompting, and workflow building. I assume you're using AI to give you superpowers.

Wherever possible, get rid of billable hours. Charge for the reasonable value of the output you're creating or the problem you're solving. It will always be a better scenario for everybody involved.

How do we ensure AI productivity isn't quietly weakening our thinking?

This is a significant concern. The idea that increasing productivity should give us back time is powerful, but there's a challenge of becoming too reliant on AI and losing the training grounds, especially for younger employees. How do we create entry-level job opportunities and develop them into the experts of the future?

There's no solid organizational structure or change management process I've seen that properly addresses this. The same challenge exists in schools. The shortcut is there for all of us; it's easy to give a few prompts, get an output, and think the work is done. But this approach can prevent deep thinking about the topic and gaining the confidence to argue a position.

There is always an exchange. As leaders, we have to be very intentional to not let critical thinking skills slip. This is especially true in environments where there's pressure to reduce staff, which just compiles the work on those who remain. People will just use AI even more and push work upward. This is a new frontier, and while hopefully more companies are thinking about it, elegant solutions have not yet emerged.

What's the best reframe for creatives who see AI as a threat?

It ultimately comes down to a choice. Writers, artists, and musicians will all have to come to grips with the fact that AI will be able to create at a human-expert level in all domains. The output will often be indistinguishable from human work on the surface.

However, just because AI can do something doesn't mean you have to let it. There will increasingly be a human preference for content and creative work when they know the story behind the human creator. If I show you two pieces of art and tell you one was made by an AI with three prompts and the other by a human with a rich story, your emotional connection will almost always transfer to the human one.

The imperfections of human creativity are probably what will make it so special in the future, along with the stories behind how creators learned their craft. AI is just not going to have those stories. I’m very optimistic about an explosion of human creativity, and also human-plus-AI creativity, as long as it's presented in an authentic way.

How do you wrangle a "Wild West" AI free-for-all at your company?

You wrangle your team by putting guardrails in place and providing approved tools. Too often, the first misstep is that no one has given the team a structured platform. Companies need to get licenses for tools like Gemini, give everyone access, and define standard use cases. You can even build the first five custom Gems for them, personalized by role or department.

You have to approach it from a change management perspective, and that has to come from the top. It's very difficult for a department head to do this without support from the C-suite and alignment with IT, procurement, and legal. The two most fundamental steps every organization needs to take are providing access to the technology and then delivering proper training.

How can you personalize AI training at the enterprise level?

The first step is to survey your people. You need to figure out where they are with their comprehension and competency with AI tools, but also how they feel about AI. Some people may be resistant because they feel it threatens their job or they simply hate it. You have to first break down that human barrier of resistance to change.

From there, you have to help them realize the value through a use case that matters to them, which often involves a task they hate doing. Find a part of their job they don't enjoy and ask, "Can we create a custom AI to help you with that?" This can break down barriers and open their minds.

Once you move past the initial resistance, you can structure the training.

The goal is to create a collection of resources that allows individuals to personalize their learning journey or lets administrators assign training based on where people are and where they want to go in their careers.

How do you get legal stakeholders to enable AI adoption instead of blocking it?

You have to involve legal, IT, and procurement from day one. Even if you have the autonomy as a leader to get licenses for your team, you want to be aligned with these other departments. You need to understand where they are resistant to infusing AI and, more importantly, why they are resistant. The more you work together and understand each other's perspectives, the better you'll be able to drive adoption without running into roadblocks.

Do an audit upfront. Sit down with legal and ask:

Then, you can figure out how to steer toward low-risk applications and use cases. Openness and a collaborative approach are key. A position that seems illogical might make perfect sense once you understand the reasoning behind it. There's a reason IT and legal are hesitant; there are many unknowns and risks, and you have to accept that and work together.

How will AI adoption pick up in traditional industries like manufacturing?

There are always segments within any vertical. In manufacturing, some pockets are racing ahead with digital transformation and AI adoption, while others are much slower. With any slow adoption, you have to create a sense of urgency at the highest level. If the CEO isn't bought in and the board isn't pushing for change, AI adoption will only happen in silos.

You have to find the trigger points that get leadership to move. That could be an executive briefing with a trusted outside expert who can explain what's happening. In these sessions, leaders can ask questions in a setting where they won't feel stupid for not knowing the answers. You have to know what moves your management team, but increasingly, that drive has to be C-suite driven. The top leaders must be convinced of the need to prioritize AI transformation and the urgency with which they need to do it.

Can companies behind on digitalization leapfrog ahead with AI?

Yes, it can be an opportunity. We're spending more time thinking about what the future organizational chart looks like. Significant changes are coming to the hiring process, talent evaluation, org charts, and career paths.

My theory is that the roles we've all grown up with won't look anything like they do now in three years. There's an opportunity to completely reimagine what we do, how we're trained, and the roles we play at every level. This presents a leapfrog opportunity because you can rethink work, roles, and career paths in a way that translates directly into your HR processes—how you recruit, interview, assess, and guide people's careers, especially when AI is doing much of the entry-level work.

There's going to be a way to do this transformationally. The challenge for large enterprises is that major change is never easy. But for an AI-native company, there has never been a greater time. You can just decide to approach things differently, define new titles, and infuse new skills into training from day one. You can't just flip a switch and do that with big companies.

Will AI companies eventually price based on the labor they replace?

I believe labor replacement cost is one of the most logical pricing models for AI, though I don't necessarily advocate for it. Legacy software companies may struggle to position their products this way due to the PR nightmare, but AI-native companies will absolutely do this.

They will approach a company and say, "You have five customer service agents spending $800,000 a year. We can triple their output, cut response time in half, and do it with one agent for $250,000 a year." If that's true, no publicly traded CEO can say no; they have a fiduciary responsibility to consider it.

The current models of metering, utility credits, and abstract pricing are impossible to budget for and will likely see their day pass quickly. Outcome-based and value-based pricing makes the most sense. If you go to a CFO and say, "For $50,000 a month, you get unlimited use for this level of output, which is equivalent to the work of 10 marketers," that's a conversation they understand. It seems too obvious that this is the eventual direction, even if it's messy.

What is a swarm of agents and why does it matter?

A swarm of agents is an informal way to describe a group of AI agents working together. You can think of it as a symphony of agents. For example, in marketing, you might have:

Just like a human marketing team, these agents are highly trained for specific functions but collaborate within a shared environment. A leader could set a goal, provide the necessary data and context, and hit "go." The swarm of agents would work together, plan, execute, and ping the human leader for approvals at key checkpoints. The human becomes an orchestrator of these agent groups. I believe that by the end of this year, early adopters will be running their marketing, sales, and customer success teams in this way.

Do reasoning models actually reason or just predict the next word?

This is more of a philosophical question. We don't fully understand how humans reason. Some would argue the human process isn't that different from what machines are doing. Even leading AI researchers aren't 100% clear on the answer.

Some believe it's simply next-token prediction happening thousands of times per second, with self-verification. Others believe the models are approaching metacognition—awareness of their own thinking. We may never get a definitive answer, and it brings up deeper questions about consciousness itself.

Ultimately, what matters to me is that it simulates reasoning incredibly well. Whether it actually has empathy or not is less important than the fact that it can simulate empathy better than some humans. The fact that we are even debating this shows how advanced the simulation is. The outcome is what matters.

Should AI companies be regulated to preserve diversity of thought?

This is a complex issue with legal and societal elements, including the biases of who builds the models and what they're trained on. Right now, governments run the risk of mandating a certain type of thinking in AI models. For instance, the friction between Anthropic and the Pentagon involves the government wanting AI models to align with its philosophies.

Elon Musk's mission for Grok is to seek "truth," but it's his truth, based on his specific beliefs and sources. That's his prerogative as the builder. The question is, should we allow that? Should there be models that appeal more to Republicans and others to Democrats? It's dangerous for humans to be the sole arbiters of truth, but that's how it has always been with media outlets. They present the same facts in different lights to shape perception. I don't know the answer, but it's a critical conversation.

If AI can solve advanced math, why can't it solve technological unemployment?

AI can solve advanced math problems because it has been specifically trained by math experts on data with verifiable outputs. You can train a model on math and know definitively if the answer is right or wrong, which allows for structured training.

The technological unemployment conundrum, like other major societal issues, has no clear, single "right" answer. Therefore, it's hard to train a model to solve it because we don't know what the correct output looks like. When we get into superhuman intelligence for societal, defense, or philosophical problems, how can a human even evaluate the output?

The bet that labs like Google DeepMind are making is: solve intelligence first, then solve everything else. The mission is that once we achieve AGI, it can help us solve these messy, complex problems. I choose to believe this is possible because it’s the most optimistic outlook.

How do we make sure AI gives us time back instead of just more work?

There's a lot of chatter that companies will just do what they've always done: take productivity gains and create more work. We're already seeing a slow slide down that cliff, where staff is reduced and those who remain are pressured to be superhuman.

As leaders, we have a choice. At some point, there's a profit level that's enough and a growth level that's enough. This is especially true for privately held companies. You have to make a conscious decision to give time back. This doesn't necessarily mean a four-day workweek, which can often be more of a PR move. It means implementing real, tangible changes:

AI is giving us a gift of time, but you have to be intentional about taking advantage of it. It requires deliberately stepping back, taking care of yourself, and prioritizing things outside of work. As an organization, we have an opportunity to fully embrace and operationalize this mindset.