People often ask me what the next stage of artificial intelligence (AI) adoption will look like. My answer is that we are no longer talking about AI literacy—we are beginning to see AI dependency.

AI literacy is a good thing. It means knowing what AI can and cannot do: how to write effective prompts, fact-check outputs, identify bias, and leverage tools to boost productivity. Every organization should aim for that level, as AI has become a standard workplace utility, much like spreadsheets or search engines.

AI dependency, however, is something else entirely. It begins when people stop using AI as an assistant and start letting it think for them. That line is becoming easier to cross than many realize.

The recent decision by the University of the Philippines College of Law to restrict the use of AI and digital devices in first-year Juris Doctor classes is a clear signal of growing concern among educators. Faculty members report a sharp decline in the quality of student work, particularly in legal reasoning, analysis, and research. The policy does not ban AI; rather, it delays extensive adoption until students master the foundational skills required of legal professionals. Faculty members who permit AI in limited situations are expected to supervise its use closely.

This distinction is crucial. The issue is not AI itself; the issue is what happens when people never learn to think before asking AI to think for them.

This challenge is not unique to the Philippines. The University of Chicago Law School recently introduced one of the strictest AI policies in the United States, prohibiting first-year students from using laptops and phones in required classes to help them build critical thinking, strategic analysis, and independent reasoning before incorporating AI into their workflow. As their dean argued, future lawyers must still be able to solve problems when AI is unavailable.

Ironically, these institutions are not anti-AI—they are preparing students to use AI more effectively later on.

The challenge extends far beyond academia. Inside many companies, employees routinely rely on AI to draft emails, summarize reports, build presentations, analyze spreadsheets, and record meeting minutes. These are all practical applications; I use AI myself for research and ideation. Yet I have noticed a worrying trend: many people now accept the first response AI offers. They no longer verify facts, challenge assumptions, or question whether a conclusion makes sense. That is where dependence sets in.

Recent surveys show that more than 80 percent of college students have used AI for academic work in the past six months, primarily for brainstorming, summarizing, and research. Notably, nearly four in 10 express concern that AI could erode independent thinking, seeking clear guidelines for responsible use rather than outright bans. Students themselves recognize the risk.

Research supports these concerns. One recent study found that standard AI tools can boost overall performance and shrink productivity gaps across varying educational backgrounds. However, long-term skill retention depends on whether users continue to exert cognitive effort. Those who delegate thinking to AI gain far less than those who actively engage with the tool.

This finding offers a vital lesson for business leaders: the objective should not be to replace thinking with AI, but to amplify it.

Consider navigation apps. GPS has simplified travel, yet many people can no longer navigate familiar places without it because their spatial memory has atrophied from disuse. AI threatens to produce the same effect in writing, analysis, problem-solving, and creativity.

When employees rely on AI to draft every report, summarize every meeting, and generate every idea, they gradually lose confidence in their own judgment. Instead of becoming better decision-makers, they become prompt operators.

Organizations must be careful not to reward speed alone. By prioritizing raw output volume, management risks inadvertently encouraging surface-level thinking. Faster output does not equate to better judgment.

How do we prevent AI dependency?

First, organizations should insist that employees formulate their own ideas before consulting AI. Workers should draft initial solutions first, using AI only afterward to review, critique, or refine them.

Second, AI-generated work must always be challenged. Managers should ask employees why they agree with a given recommendation. If an employee cannot explain the rationale without referencing the AI response, true learning has not occurred.

Third, leaders must redesign learning and evaluation. Instead of assessing whether employees can produce answers, assess whether they can defend them. Oral presentations, live problem-solving exercises, and collaborative workshops remain essential because they reveal genuine understanding.

Finally, we must remember that AI literacy is not about mastering software—it is about strengthening human judgment while using software.

Every major technology transforms work: calculators changed mathematics, search engines changed research, and GPS changed navigation. AI is now changing thought itself. The real question is whether we will allow it to replace thinking altogether.

If we are not careful, the next generation may excel at asking questions of machines while losing the ability to solve complex problems independently. That is a steep price to pay for convenience.

The future belongs to those who know when to trust AI, when to question it, and when to step away and think for themselves. That is the fundamental difference between AI literacy and AI dependency.