For the past decade I have served on the program advisory council for a local post-secondary business school. Just this past year, we met so the marketing profs could present their proposed course changes to us for our feedback.

I, for the first time in one of these meetings, didn’t know what to say. It was nothing about the proposed changes themselves. It was because I honestly didn’t know what role humans would play in marketing in the future.

Some of the course changes were for first year students who would still have 4 or 5 years of the program ahead of them. That means they’d be looking for their first job in marketing in 2030 or 2031. I assume that would be an entry level job. So what, I wondered, would you teach marketing students today to make them a wanted commodity in the marketing job market 5 years hence?

Given this question that has been lodged in my brain ever since, I read with interest fellow Media Insider Maarten Albarda’s post last week on “Humans in the Loop” in the marketing platform stack. In it, he said, “For decades, advertising relied on a strict human-in-the-loop chain with clear approval protocols. In today’s world, that has become optional and comes at a cost.”

The new advertising ecosystem is being built by AI technicians to eliminate friction, ensuring higher profitability generated by self-scaling programs without restraints. In that world, humans are nothing more than profit-sapping friction and therefore should be eliminated. I’m not sure what openings that leaves for new marketers trying to enter the business 5 years from now. What should we be teaching them to give them a fighting chance to avoid becoming both redundant and irrelevant?

Supposedly, the reason to keep a human “in the loop” is so they can step in when something unexpected arises that AI doesn’t have the operating parameters to cope with. These are called “edge cases.” The thinking is that with a human somewhere in the mix, we can all rest easy knowing that when these “edge cases” arise, expert judgement can hit the stop button and keep us all from being sent to AI Armageddon.

There are several things wrong with that comforting assumption.

First, who or what flags these “edge cases?” If it’s AI, we’re assuming that it can recognize when something is above its pay grade and call in a human. But even if it does do that, what about all the other edge cases that AI didn’t recognize as such, assuming that these tasks fit within its normal operating procedures?

Secondly, even if exceptions are properly identified as an edge case and a human is called in to oversee, the places where human intervention is required is a constantly moving target. AI is light years ahead of where it was just a year or two ago, which means today’s edge case is tomorrow’s routine.

One example of this constantly expanding AI horizon comes from self-driving cars. Waymo self-driving taxis are coming to Toronto. One of the reasons they haven’t come to Canada sooner was because our notorious winter driving conditions represented an “edge case” for self-driving AI. But Waymo feels that they have now collected enough data from northern US cities to allow it to reliably model what is required to stay on an icy street during a Canadian winter storm. In this case, what was an “edge case” has now been incorporated into a “business-as-usual” data set.

Finally, to bring us back to our marketing students, the only reason to have a human in the loop is to allow the human to use their expertise and intuition to keep AI from going off-track. But this intuition only comes from thousands and thousands of hours of experience. And, as I have said before, as we off load more and more cognitive drudge work to AI, we don’t build that expertise because we haven’t immersed ourselves in the work required to gain it. Will a brand-new marketing student ever become an expert when 99% of the day-to-day work required to put a campaign together is now handled by AI?

These are just some of the reasons why I didn’t have a ready answer when the question came in that PAC meeting, “What should we be including in our marketing programs?” Someone else in the group quickly said, “Teach them how to work with AI.” And while I didn’t say anything out loud, the thought that thundered through my head was this, “Yeah. Good luck with that!”