Marketing teams are getting more confident and less right.
There's a name for why. Wharton researchers Steven Shaw and Gideon Nave call it 'cognitive surrender': the point where we let AI lead our thinking, then present the result as our own. Most of us already sense the slippery slope. If we don't use our noggin, we might lose our noggin.
But is that really so bad? Getting the brainy kid to help with your homework has always been a decent shortcut. The brainy kid usually knows the answers.
Usually. In their paper, Thinking - Fast, Slow, and Artificial (Wharton, January 2026), Shaw and Nave ran three experiments with 1,372 people. Participants had a chatbot to hand that sometimes gave confident wrong answers. They accepted those wrong answers 80% of the time.
That's worrying enough. But the bit that should make marketing leaders sit up is what happened to confidence. People using the AI rated their confidence in 'their' answers 11.7% higher than a group working without it. When the AI was right, accuracy jumped 25 percentage points. When it was wrong, accuracy fell 15 points below people working on their own. And the confidence stayed high either way.
Accuracy down, certainty up.
That's a bad combination in any team, and an awkward one in marketing, where so much of the job is judgement. Which audience, which idea, which bet to back. If the AI gets it wrong and everyone feels sure, nobody goes looking for the mistake.
The job for leaders is to put the thinking back where they can see it. Three places to start.
Ask to see the working out.
Your maths teacher was onto something. An answer on its own tells you very little. Ask what was explored and ruled out, how the options were weighed, what the strongest counter-argument is and why it didn't win. If nobody can tell you, the thinking probably didn't happen.
Get open about AI use.
'Shadow AI', where people use tools your IT team hasn't signed off, is one problem. Teams hiding that they use AI at all is another, and arguably a worse one, because you can't ask for the working out if you don't know who did it. Aim for maximum visibility and go first. If you're open about where AI helped you and where it led you astray, your team has permission to be too.
Watch the seniors, not the juniors.
We're good at checking the output of the whippersnappers. We've always marked their work. Experienced people don't get marked, they present. And a confident senior with a polished answer is very hard to challenge, even when the polish came from a chatbot. If they're not conscious of how much brainpower they're handing over to the bots, nobody else is checking either.
None of this is about using AI less. It's about owning the thinking.
If you'd like a clearer picture of how AI is really being used across your team, and where the thinking might be going missing, that's exactly the sort of thing we get into in a Waypoint. It's a free, confidential session for marketing leaders, and within 48 hours you get a written read on what's slowing your function down. You can book yours here.