AI is becoming a source of social proof

For a few weeks now, we've been tracking how buyers find you inside AI answers. A multi-million-event study found that when an AI platform recommends a brand, the buyer is 182% more likely to search for it within a week. A passing mention does a fraction of that. It's becoming a new form of social proof.

Some marketers are already manipulating the behavior. Researchers showed that a 13-word snippet planted on Reddit or Wikipedia can affect what AI search tools cite as authoritative.

In some ways, the choice is between faking it and deserving it. Structure your expertise as clear claims, genuine comparisons, named experts, and evidence that holds up under a buyer's question. Then go ask an AI what it recommends in your category, the way an actual buyer would, and see what comes back.

Telling people you used AI can cost you

Last week, the worry was telling strong AI output from confident-sounding noise. A recent controlled experiment replicated the same work with two audiences and changed only one thing: whether the person said they'd used AI. Those who disclosed it were rated 10 times lazier and 24 points less recommendable for visible work. Even framing AI as a team helper didn't fully close the gap.

I suspect that's even more the case with how high-trust audiences judge authorship right now. Visible human judgment on the page matters, like named experts, operator detail, and reasoning that a reader can follow back to a person. A skeptical institutional reader is scanning for trust, whether they'd say so or not. You probably don't want a lot of press talking about how you cut marketing costs using AI.

Build genuine human authorship into your published thought leadership and make it visible. Put the named experts and the operator detail where a reader sees them first. And think carefully about how to describe AI's role in the work. That description plays a part in how your work gets received.

When AI executes, can you explain your thinking?

Something I've believed for a long time is getting easier to prove. As AI absorbs execution work in thought leadership, you're still expected to reproduce its reasoning. The same is true for designs, messaging, and other artifacts from marketing activity. One leader said this week that 9 in 10 marketers can't explain why they made the call they did when you ask them.

You can also see it in what people are asking for. Demand has moved toward strategic frameworks that make sense in the real-world context of individual readers. There's an inherent gap in tactical but generic how-to. Pieces that build trust show a point of view and the reasoning under it. You might get caught short if an LLM did that work.

Put your scarce senior time into thinking, even if you do let AI take on production work. Get the reasoning onto the page where internal and external stakeholders can see it. Being able to defend why looks very different from just producing more of what.