For two years, the AI conversation in every boardroom has been about adoption: who’s using it, how much, how fast. That conversation is over. Adoption happened. What should keep leaders up at night now is harder to see on a dashboard: What are people doing after adoption? We recently conducted our inaugural Scribd Understanding Index, surveying 1,500 Americans about how they build knowledge today. The clearest signal in the data wasn’t about tools at all. In fact, 37% of those surveyed said the hardest part of any research session is knowing which sources to trust—twice the rate of the next biggest challenge. Confidence in information is the bottleneck.
Verification is already a workflow that shows how people are handling the trust problem: They’re fact-checking. We found that 66% verify AI results nearly every time they use them, and 47% say validation is so critical they always make time for it, even when they’re busy.
I read that as maturity. When a technology is novel, we marvel at it. When it becomes routine, we hold it to a working standard. The stakes on trust rise precisely because AI is now embedded in real decisions. And this habit already exists inside your organization, whether you’ve designed for it or not.
Business professionals are the heaviest AI users we surveyed, with 55% using AI tools daily. They’re also the most disciplined about verification: 60% say they always make time to validate. Your best people are spending real hours grounding AI’s answers. The choice for leaders is whether verification happens ad hoc, invisibly, and inconsistently, or as a designed part of how work gets done.
Where do people go when they verify? To humans. In our research, human-authored expert documents earn a net trust score nearly four times higher than of AI-generated summaries alone. And people can tell you exactly why: 76% say author expertise is highly important to whether they trust a source, and 72% say the same of citations.
This data perfectly validates a core belief of mine: AI significantly raises the bar for human judgment. The lesson for leaders is that expertise isn’t a nice-to-have that AI replaces. AI generates answers; accountable human knowledge is what lets your teams act on them.
I constantly ask my own leaders to optimize for clarity. Three moves follow from the data to help you do just that:
*1. *** Build verification into the workflow itself: Establish checkpoints, source requirements, and review norms rather than leaving it to individual diligence.
2. Reward accuracy and sourcing, not just speed: If your incentives only celebrate velocity, you’re training people to skip the step your best performers already consider non-negotiable.
3. Don’t outsource judgment to a black box: Keep investing in the subject-matter experts, primary documents, and institutional knowledge your people lean on to ground AI’s output.
Adoption is no longer the differentiator. It’s not about deploying technology. The next competitive advantage is how much trust you’ve built around it. The companies that pull ahead will have leaders treating trust as a system and building it into workflows.
Ultimately, the future belongs to people who understand. And that understanding is built by other people.
Tony Grimminck is chief executive officer of Scribd, Inc.