How Curious Are You, Really?

AI will automate the answers. The only question left is whether you still want to ask them.

Rabbit Strat Matt

Jun 09, 2026

I took a step back a bit for this one.

I wrote before about how to fight the machine - the practical stuff, the moves, the method.

This time I want to ask something more uncomfortable, and more honest: whether you actually want to.

We were all four once.

Me at 4 below with a rabbit:

At four years old, you asked questions relentlessly. Research suggests pre-schoolers fire off somewhere between 73 and 300 questions a day.

Then something happens. School rewards knowing the answer, not asking the question. Work finishes the job. By the time we’re sitting in marketing meetings, the muscle has quietly atrophied.

And the data on what’s left is bleak. In Francesca Gino’s Harvard Business School research, across 3,000+ employees, only about 24% reported feeling curious in their jobs on a regular basis, and roughly 70% said they face barriers to asking more questions at work.

A separate survey of 16,000 employees found only 52% believed their leaders encouraged curiosity, and 81% felt curiosity had no bearing on their pay.

In Gino’s same study, 92% of people credited curious colleagues with bringing new ideas. We know it’s where the value is. We just systematically train it out of ourselves.

I’m telling you this because I think it’s the single most important thing in our industry (marketing).

The accidental industry

There are roughly 6.5 million marketers in the world, around 15 million counting adjacent roles - PR, UX, design, sales. In the US that’s about 2.5 million people. In Europe it’s a serious workforce too: ad agencies employ around 189,000 in Germany, 112,000 in the UK, 79,000 in Spain, 70,000 in Poland and 69,000 in France (Statista, 2024).

The open secret = most of those people fell into it. Only about 23% of agency staff hold a marketing-related degree and the CIM has found 35% received no training at all in the previous two years.

That’s not a criticism - it’s the opposite. Some of the best marketers I know arrived sideways, and that matters. You don’t want everyone coming up through the same microcosm, trained on the same playbook, reaching for the same moves. That’s how you get convergence - the human version of the average answer. You want diversity of thought, the kind that actually solves problems and differentiates, precisely because the people in the room didn’t all think their way in from the same door. It’s genuinely fine for an industry to give millions of people careers - and the messy, accidental way they got here might hold some great benefits.

But a new challenge is now in front of that workforce, because AI is about to ask all of them the same uncomfortable question. And it isn’t “can you do the work?” because maybe these untrained people have gotten good enough at doing their job without a Mark Ritson MiniMBA… but the question is = “how curious are you, really?”

The trap: we got obsessed with the answers

Somewhere in the last decade, marketing fell in love with prediction. We built a profession around measuring incrementality, proving attribution, confirming lift -convinced reality could be modelled and forecast.

But markets aren’t weather. Yuval Noah Harari distinguishes between a “level-one” chaotic system like weather, which doesn’t react to predictions about it, and a “level-two” (second-order) system, which reacts to the prediction itself and changes. Build a model that perfectly predicts the oil price, and traders act on it, moving the price before the forecast lands. The prediction eats itself.

I’ve banged on about this before…

Marketing is level-two. The moment everyone optimises toward the same predicted outcome, the outcome moves. We poured our best minds into perfecting forecasts in a system that is, by nature, un-forecastable - and called the small slice we could confirm “results.”

We chose the answers over the questions.

We chose to stop being four 😢

The part AI eats first

And this is exactly where AI is strongest. McKinsey estimates generative AI has lifted the share of work hours technically automatable from ~50% to 60–70%, with automated hours projected to reach 45% in the EU and 48% in the US by 2035 (I know what some of us think about Mckinsey predictions but this one proves my point so…).

The answerable, the optimisable, the pattern-following - that’s the easiest thing in the world to automate. A model is, in essence, a machine for confirming patterns. If your value is built on efficiency, you’re not building a moat - you’re volunteering for the exact work AI does better and cheaper. Play that game long enough and you don’t win it. You end up inside an automated ecosystem owned by the people who own the automation.

Efficiency is fine. It’s just not a moat, and it has an expiry date.

Curiosity is a human advantage

A model can give you the answer. It cannot want to ask a better question. The psychologist Todd Kashdan maps curiosity across several dimensions - joyous exploration, deprivation sensitivity (the itch of a knowledge gap), and crucially stress tolerance: the willingness to sit in doubt, confusion and not-knowing long enough to find something genuinely new.

That last one is the whole game. AI removes the discomfort of not-knowing instantly - and that discomfort is exactly what real discovery requires. The personality research agrees: Openness to Experience - intellectual curiosity, appetite for novelty - is the single best personality predictor of creative output. The same trait that makes someone restless with routine is the one that produces the genuinely new. The marketers who’ll thrive aren’t the best at following the process. They’re the ones least comfortable inside it.

You can’t hide behind the process anymore

We treated process as a hiding place - “I followed the steps, I hit the KPI.” AI takes that away, because the process becomes the AI. What’s left for humans is the part that requires leaning in harder: agency, client and team all having to ask bigger, deeper, more curious questions.

So here’s the honest question - and it really is honest, not a trap. When you sit in front of an AI workflow, what are you in this for?

If it’s efficiency and cost-cutting, that’s fine - but be clear-eyed: it won’t differentiate you, and that game has an end.