I’m here today with Milena Traikovich, a leading expert in demand generation whose work focuses on the intersection of analytics, performance optimization, and technology. She helps businesses navigate the rapidly changing landscape of digital marketing, making her the perfect person to discuss the often-feared, yet potentially transformative, impact of artificial intelligence. We’ll be exploring a counterintuitive idethat AI, much like the steam engine of the 19th century, might actually create more marketing jobs than it eliminates. Our conversation will touch on how this economic paradox is reshaping the day-to-day reality for marketing professionals, the new skills required to thrive, and the emerging roles that will define the industry’s future.
The article frames its argument around the 19th-century Jevons paradox. Could you walk us through how this coal consumption theory translates to today’s digital marketing, and give a specific example of how a tool like Amazon’s Ads Agent mirrors the efficiency impact of Watt’s steam engine?
It’s a fascinating and powerful parallel. Back in 1865, William Stanley Jevons noticed something strange: James Watt’s steam engine made using coal far more efficient, yet instead of coal consumption decreasing, it skyrocketed. The reason was that this newfound efficiency made the technology accessible for countless new applications that were previously unthinkable, driving up overall demand. Today, AI is our steam engine, and marketing expertise is our coal. AI tools are making the execution of marketing tasks dramatically cheaper and faster. Look at Amazon’s Ads Agent, which launched in November 2025. It allows a marketer to manage complex campaigns across different platforms using simple natural language. This is the modern equivalent of Watt’s engine; it takes a complex, resource-intensive process and makes it so efficient that the barrier to entry collapses, enabling a whole new scale of activity.
Box CEO Aaron Levie predicted AI would enable “the marketing campaign that wouldn’t have been launched otherwise.” Can you share a metric-driven anecdote of a small business using new AI tools to launch a campaign that was previously unfeasible, detailing the resources saved and the outcomes achieved?
Absolutely. This is where the theory becomes tangible. Think of a 10-person services firm. Before, launching a sophisticated, multi-channel campaign was a pipe dream. They’d face brutal tradeoffs: should we hire a developer to build a landing page, or a designer for creative assets? They couldn’t do both. Now, with generative AI tools like those from Meta, which are already used by over 4 million advertisers, that same firm can do it all. An account manager can generate a dozen variations of ad creative in an afternoon, a task that would have taken weeks and thousands of dollars with an agency. Meta even reported that these tools cut the time spent on certain DSP tasks by a staggering 75%. That 10-person firm can now launch a campaign that proves their value proposition quickly, without the crippling upfront investment. They’re not just saving money; they’re unlocking growth opportunities that were completely out of reach.
The text mentions agencies aim to increase account manager portfolios from 35 to 64 clients. What does the day-to-day reality look like for that manager? Describe the new high-value skills they’ll need as tasks like budget pacing and campaign setup are automated by over 80%.
The day-to-day reality for that account manager is a fundamental transformation from technician to strategist. Imagine the “before” picture: hours spent staring at spreadsheets, manually adjusting bids, and painstakingly setting up campaigns. The data shows that automation is set to reduce time on budget pacing by 90% and campaign setup by 80%. That mental energy is now freed up. The “after” picture is a manager who spends their morning analyzing AI-surfaced insights, their afternoon in deep conversation with clients about long-term goals and market positioning, and their evening directing AI agents to test a dozen new creative angles. The essential skills are no longer about clicking the right buttons. They are about strategic oversight, critical thinking, client empathy, and the ability to ask the AI the right questions. You become the conductor of an orchestra of intelligent agents, not just a single instrumentalist.
Harvard research identified several pitfalls, such as people blaming “black box” AI for failures. What practical, step-by-step process should a marketing leader implement to build internal and client trust in AI, and how should they manage the fallout when an automated campaign underperforms?
Building trust is paramount, and it requires a deliberate, human-centric approach. First, a leader must champion transparency. This means educating both their internal teams and their clients on what the AI does and, just as importantly, what it doesn’t do. Avoid overstating its capabilities, which the Harvard research shows leads to harsher judgment. Second, implement a “human-in-the-loop” model. Let AI handle 80% of the work, but ensure a human expert reviews and approves the outputs, providing that crucial layer of context and oversight. This builds confidence that the machine isn’t running unchecked. When a campaign underperforms, the process is critical. Instead of just saying “the algorithm failed,” the leader must guide a collaborative post-mortem. Was it a flawed prompt? Was the initial data biased? Was the strategic direction wrong? By deconstructing the failure as a team, you shift the blame from the “black box” to a correctable process, reinforcing that AI is a tool managed by people, not an infallible oracle.
Unlike the inelastic demand for food that led to fewer agricultural jobs, marketing demand is described as elastic. Based on this, what specific new marketing roles or specializations do you see emerging as AI handles more tactical work, and what existing roles are most at risk?
This elasticity is the key. When food became cheaper to produce, people didn’t start eating ten meals a day—demand was inelastic. But when marketing becomes cheaper, businesses absolutely will do more of it. This opens the door for new specializations. I see the rise of roles like “AI Marketing Strategist,” who will be experts at orchestrating various AI agents to achieve business goals, or “Creative Prompt Engineers” who specialize in translating brand identity into instructions for generative AI. We’ll also need “AI Ethics & Governance Specialists” to manage the risks. On the other hand, the roles most at risk are those defined by repetitive, deterministic tasks. Think of junior roles focused purely on manual campaign setup, A/B testing implementation, or basic data reporting. These functions are precisely what AI agents are being designed to do with superhuman efficiency. The value will shift from doing the task to defining and overseeing the task.
What is your forecast for the marketing job landscape in 2030? Specifically, how will the division of labor between human strategists and AI agents evolve, and what core competency will be most critical for a marketing professional’s long-term success?
By 2030, I forecast a marketing landscape that operates like a high-level consultancy, even within brand teams. The division of labor will be crystal clear: AI agents will handle the vast majority of execution—the bidding, the creative generation, the data processing, the reporting. The human’s role will be entirely strategic. We will be the architects, defining the goals, providing the essential context, and making the final judgment calls that an AI, no matter how advanced, cannot. Today’s entire job might become a single task for a future AI. The single most critical competency for a marketing professional’s long-term success will be strategic curiosity. It’s the ability to ask powerful questions, to challenge the AI’s outputs, to synthesize information from disparate sources, and to translate a complex business problem into a clear directive for your suite of AI tools. Your value will no longer be in the answers you can provide, but in the questions you can formulate.
