The recent jury verdict against Meta and YouTube was framed as a Big Tobacco moment for tech—a long-overdue reckoning with platforms designed to drive behavioral dependence in young people.
But while the public focuses on the addiction crisis, something quieter and more consequential is happening now.
Children and adults are developing a different kind of dependence: cognitive dependence. They’re outsourcing not just answers, but reasoning itself. In doing so, they risk weakening the very skills required to learn, adapt and make better decisions.
In classrooms, this looks like a student who uses AI to generate a history essay in seconds but cannot explain the underlying arguments or apply them to a new historical context. These students are not just taking a shortcut; they’re bypassing the struggle that builds neural pathways and true understanding.
Conversely, a classroom grounded in intentional, constructive AI use can change this dynamic. Instead of asking AI to write an essay on the causes of World War I, a student trained in decision-making might instruct AI to generate three conflicting arguments about the war’s causes.
The student’s task shifts to evaluating the evidence, identifying cognitive biases within the AI’s sourced arguments, and forming their own conclusion. Here, AI functions as a sparring partner, not a surrogate thinker.
Risk is not evenly distributed
You cannot reskill a workforce if the underlying capacity for critical thought has already been eroded. And yet that is exactly the contradiction the education sector is contending with.
School districts and curriculum designers must connect workforce development with AI policies. One determines whether people will have jobs. The other determines whether they will have the capacity to do them.
And that capacity is increasingly well understood. Research from the Burning Glass Institute’s Decision Skills in the Workplace report shows that employers are not just hiring for technical expertise.
They are seeking durable, transferable decision-making skills like evaluating evidence, reasoning under uncertainty, generating options, and recognizing cognitive biases. These skills are strong predictors of career mobility.
While the future of work depends on how well people make decisions, those are exactly the skills most at risk when thinking is outsourced to AI.
The risk is not evenly distributed. Just as access to computers widened the gap in an internet-driven economy, access to strong cognitive habits will widen it in an AI-driven one.
Students who see adults questioning AI’s outputs, revising drafts, and treating AI as a starting point will learn to think with it. Everyone else will simply defer to it.
AI and critical thinking
So how can school systems ensure that all students develop the necessary skills to use and navigate AI effectively?
There is already an established pedagogical approach: integrating systemic decision-making into the curriculum. Teaching decision skills starts from the simple premise that decision-making is a cognitive habit that can be analyzed, taught, and improved.
These aren’t “soft skills.” They’re the cognitive infrastructure of a resilient workforce.
When students develop these critical dispositions, they transcend passive consumption. They learn to audit data, spot algorithmic bias, and treat AI as a tool to expand their choices rather than dictate them.
Most importantly, students tap into their human advantage and think alongside AI; students can bring context and priorities, and use AI as a tool to support and refine their thinking.
If educational leaders are serious about preparing students for an AI-driven economy, cultivating these formal decision skills cannot remain peripheral. Deliberate cognitive training must be integrated into how students learn and how they engage with AI.
State policymakers must integrate explicit decision-making frameworks into core K-12 academic standards—not as a standalone elective, but woven directly into existing mathematics, civics, and media literacy curricula.
State departments of education must establish standards for AI tools used in classrooms. School districts should only purchase AI software that aligns with evidence-based learning and prioritizes active student inquiry over passive retrieval.
A narrower window
Technology companies also have a role to play. AI systems should prompt students to form hypotheses, make reasoning visible, and engage actively before sharing answers. In this context, design “friction”—the intentional introduction of pauses and cognitive challenges—is not an inefficiency to be engineered away; rather, it’s a vital pedagogical tool.
For instance, instead of an AI chatbot instantly spitting out a completed math proof or essay outline, the software should be intentionally designed to pause, prompting the student to first state their hypothesis or defend their initial assumption before unlocking the next layer of assistance.
Policymakers and the public recognized the psychological harms of social media too late. That window is narrower with AI.
The habits that parents and educators model today will shape how the next generation thinks. Technology companies can choose to build AI systems that actively support those decision-making skills.
Or we can watch them erode.
