The Marketers Who Think With AI Will Outperform the Ones Who Think Like AI

By Lisa Peyton and her team of AI super heroes

You’re on deadline. You paste a prompt into your favorite AI tool. The response comes back fast, confident, and polished. You scan it, nod, tweak a word or two, and hit publish. It felt right. But here’s the question that should keep every content marketer up at night: was it?

A new research paper from Wharton just put a name on something you’ve probably experienced but never articulated. Researchers Steven Shaw and Gideon Nave call it “cognitive surrender”: the tendency to adopt AI-generated outputs with minimal scrutiny, bypassing your own critical thinking in the process. Across three experiments with nearly 1,400 participants, they found that when people have access to AI, they don’t just use it. They defer to it, even when it’s wrong.

This isn’t a reason to panic. It’s a reason to get sharper. The content marketers who understand what’s happening in their own brains when they work with AI will consistently outperform the ones running on autopilot. This post breaks down the research, translates it into actionable strategy, and gives you hands-on exercises to become a more intentional, more effective AI-powered marketer.

You probably know Daniel Kahneman’s framework: System 1 is your fast, gut-instinct brain. System 2 is your slow, deliberate, analytical brain. Shaw and Nave’s big move is adding a third player: System 3, artificial cognition. That’s AI. And they argue it’s not just a tool you pick up and put down. It’s become an active participant in how you think, reason, and make decisions.

The Science of Surrender

Here’s what their experiments revealed, and why it matters for anyone creating content with AI:

We use AI most of the time, and we follow its lead. Participants consulted the AI chatbot on over 50% of trials and followed its recommendations 80–93% of the time. When the AI was deliberately giving wrong answers, people still followed it roughly 4 out of 5 times.

Good AI made people better. Bad AI made people worse than having no AI at all. When the AI gave correct answers, participant accuracy jumped 25 percentage points above the no-AI baseline. When it gave wrong answers, accuracy dropped 15 points below baseline. Read that again: people performed worse with bad AI than with no AI. Your output quality is a mirror of your AI’s accuracy.

Confidence went up across the board, regardless of whether AI was right or wrong. Access to AI boosted confidence by nearly 12 percentage points, even when half the AI’s answers were intentionally incorrect. This is the “false certainty” problem, and for content marketers, it’s a credibility time bomb. You feel more sure about work that may be less accurate.

Time pressure made surrender worse. Incentives and feedback helped, but didn’t eliminate it. When people were rushed, they leaned on AI harder and couldn’t distinguish good advice from bad. When they were given financial incentives and immediate feedback on their accuracy, they got better at overriding bad AI. But even then, the cognitive surrender pattern persisted.

Who’s most vulnerable? People with high trust in AI and a low inclination toward effortful thinking surrendered most. Who’s most resilient? People who enjoy critical thinking and actively verify. The takeaway: your disposition toward your own thinking matters as much as the AI tool you’re using.

The Playbook: Leading Teams and Leveling Up

The research gives us a clear picture of the problem. Now let’s talk about what to do about it. Whether you’re leading a team or flying solo, these strategies are built directly on what Shaw and Nave found.

The Leader’s Job: Building a Culture That Verifies

Build a verification culture, not a suspicion culture. The goal isn’t to make your team distrust AI. It’s to make verification a professional standard, like fact-checking has always been in journalism. Frame it as “AI quality assurance,” not “checking up on the robot.” The Wharton data shows that feedback and accountability structures meaningfully reduce cognitive surrender. Your team culture is the feedback loop.

Resist the speed trap. The research is unambiguous: time pressure is the number one accelerant of cognitive surrender. When your team is rushed, they’re most likely to accept AI outputs without scrutiny. Leaders who build in review buffers, even brief ones, protect their brand’s credibility. That extra 20 minutes before publish isn’t a bottleneck. It’s insurance.

Track AI-assisted content performance separately. You can’t manage what you don’t measure. Start tracking which content was heavily AI-assisted versus human-led, then compare performance over time. This creates the feedback loop the research shows is critical for recalibrating trust and catching blind spots before they become patterns.

The Practitioner’s Edge: Staying in the Loop

Know your surrender triggers. Are you most vulnerable when you’re on deadline? When the topic is outside your expertise? When the AI sounds confident and articulate? The research shows cognitive surrender isn’t random. It’s predictable. Self-awareness is your first line of defense.

Adopt the “offloading, not surrendering” mindset. Shaw and Nave found a meaningful distinction between two types of AI users. People who used AI as a collaborator, checking its work, overriding when it conflicted with their knowledge, performed best. People who used it as an autopilot performed worst. The difference isn’t how much you use AI. It’s how you use it.

Diversify your AI bench. If cognitive surrender means your output tracks a single AI’s accuracy, then relying on one tool for everything maximizes your exposure to its blind spots. Using multiple specialized tools, with your judgment as the orchestrator, is a natural hedge. I’ve been saying this for a while: coordinating multiple AI tools is the defining marketing skill of the moment. This research backs it up with hard data.

Your Anti-Surrender Toolkit

Strategy is great. But you need concrete practices to make it stick. Here are four exercises designed to build the cognitive muscles that protect you from surrender while keeping AI’s benefits fully intact.

Exercise 1: The “Brain-Only Baseline” Test

Once a week, draft one piece of content, or even one section, entirely without AI before bringing any tools in. Then compare the two versions side by side. Where did AI genuinely improve the work? Where did it just make it faster without making it better? Where did it lead you somewhere you wouldn’t have gone on your own, for better or worse? This practice calibrates your sense of your own capabilities and helps you spot where AI is adding real value versus where you’re outsourcing thinking you should own.

Exercise 2: The Red Flag Review

Before publishing any AI-assisted content, run a three-question check. First: did I verify the specific claims, statistics, and names? Second: can I explain why this argument holds without re-reading the AI output? Third: does this sound like me, or does it sound like a chatbot? If you can’t clear all three, the piece needs another pass. Print these out. Tape them to your monitor. Make them a habit, not a checklist.

Exercise 3: The Confidence Audit

After finishing an AI-assisted draft, rate your confidence on a scale of 1 to 10. Then fact-check the three claims you’re most confident about. Track your hit rate over a month. The Wharton research shows that AI inflates your sense of certainty. This exercise builds a personal early warning system. If your confidence consistently outpaces your accuracy, that’s a flashing neon sign that surrender is creeping in.

Workflow: The “AI Cross-Check” System

This is the workflow I use in my own practice, and it’s built on the core insight of this research: no single AI shares your blind spots, so use that to your advantage.

Step one: draft or research with your primary AI platform. Step two: take the output to a different AI platform and ask it to verify claims, challenge assumptions, and flag inconsistencies. Step three: apply your own expertise as the final filter. For example, I might draft with Claude, fact-check with Perplexity, and then review with my own knowledge and editorial judgment.

Think of it like getting a second opinion from a different doctor. Each AI model has different training data, different strengths, different failure modes. Running your work through a second model catches errors that a single-tool workflow would miss entirely. This isn’t extra work. It’s smarter work. And it’s the simplest, most scalable defense against cognitive surrender that exists today.

The Bottom Line: You’re the Operator, Not the Passenger

Nothing in this research says “stop using AI.” Quite the opposite. When AI is accurate, the people who used it outperformed everyone else, even under time pressure. The problem isn’t AI. The problem is uncritical AI use.

The competitive advantage going forward won’t belong to the marketers who use AI the most, or the ones who avoid it entirely. It will belong to the ones who think with AI instead of handing their thinking over to it. The ones who verify, cross-check, and stay engaged. The ones who treat every AI output as a strong first draft, never a final answer.

Shaw and Nave call it the difference between “cognitive offloading” and “cognitive surrender.” I call it the difference between being the operator and being the passenger. You get to choose which one you are, every single day.

Choose to stay in the driver’s seat.

Want more frameworks, tools, and strategies to sharpen your AI-powered marketing? Visit my AI Marketing Resources Hub for the latest.

Lisa Peyton is an AI practitioner, professor, and pioneer helping marketers navigate the evolving AI landscape. Find more resources at lisapeyton.com/ai-marketing-resources or connect with her at linktr.ee/lisapeyton.