AI Is Becoming Your Second Brain — at the Cost of Your First

When the 'External Brain' Quietly Replaces the 'Original Brain'

When was the last time you independently thought through a complex problem?

Not the three seconds of reflection before opening ChatGPT, but truly sitting down for twenty minutes or longer, using your own brain to reason from scratch, weigh options, and reach a conclusion. If you hesitate to answer that question, you're not alone. In 2025, AI has deeply embedded itself into the daily workflows of hundreds of millions of people — writing emails, drafting proposals, even deciding what to have for dinner all require consulting AI. We are witnessing an unprecedented wave of "cognitive outsourcing," and its cost may be far heavier than we imagine.

From 'Tool' to 'Crutch': The Slippery Slope of Cognitive Dependence

Tech optimists love to describe AI as a "second brain" — a powerful tool that helps us expand memory, accelerate retrieval, and boost efficiency. The metaphor itself is fine, just as a calculator is a tool for math and a search engine is a tool for information retrieval. But here's the problem: when a tool is powerful and convenient enough, humans' willingness and frequency to use their original abilities drops precipitously.

The widespread adoption of calculators caused most people to lose their mental arithmetic abilities. GPS navigation has led to the continuous deterioration of our spatial orientation. These observations are well supported by cognitive science research. Columbia University psychologist Betsy Sparrow proposed the "Google Effect" as early as 2011 — the finding that when people know information can be easily searched, the brain actively reduces the encoding intensity of memory for that information.

But the cognitive outsourcing brought by AI is fundamentally different from the search engine era. Search engines replaced "information retrieval," while large language models replace the "thinking process" itself. When you ask AI to write a tactful apology letter, you're outsourcing not just the ability to choose words, but the entire emotional reasoning process of "how to understand the other person's feelings and how to find balance between sincerity and boundaries." When you ask AI to do a pros-and-cons analysis for a business decision, you're giving up not just the labor of making a list, but the exercise of deep value prioritization — "what factors truly matter to me."

The Real Danger: The Atrophy of Judgment

This is where the issue becomes most acute. The real risk isn't merely that we're getting lazier or that our critical thinking skills are declining — though that's bad enough. The deeper risk is that we are wholesale outsourcing our judgment and may consequently lose the ability to make qualitative judgments, moral judgments, and interpersonal judgments altogether.

These three types of judgment are precisely the ones that should never be outsourced.

Qualitative judgment involves distinguishing "good from bad," "beautiful from ugly," and "important from unimportant." This kind of judgment has no standard answer; it is built on personal experience, aesthetic accumulation, and intuition. When an editor repeatedly relies on AI to determine whether an article is "good enough," their internal sense of textual quality gradually dulls. When a designer always lets AI choose color palettes, their own aesthetic judgment muscle atrophies.

Moral judgment is even more critical. Ethical choices often have no clear right or wrong; they require us to painstakingly weigh trade-offs between conflicting values. A doctor's triage decision when resources are limited, a manager's choices about whom to lay off, an ordinary person's boundary-setting when a friend asks for help — the core of these judgments is not logical reasoning but the deep moral intuition of "what do I believe is right." AI can list every ethical framework for your reference, but if you never go through the painful weighing process yourself, your moral judgment will be like a muscle that has never been exercised — ultimately atrophying beyond use.

Interpersonal judgment concerns our core capacity as social beings. Reading the emotions behind someone's words, sensing subtle power dynamics in a relationship, judging when to speak candidly and when to remain silent — these abilities are built on decades of social experience and emotional investment. When people increasingly use AI to ghostwrite social messages, analyze their partner's chat logs, or even decide how to respond to a friend's confiding, we are systematically abandoning opportunities to train our interpersonal perception.

'Use It or Lose It': The Cognitive Muscle Principle

There is a fundamental principle in cognitive neuroscience: neural pathways follow a "use it or lose it" rule. Cognitive functions that are frequently used develop stronger, more efficient neural networks; functions left unused see their neural connections gradually weaken and even get pruned away. This is not a metaphor — it is a biological fact supported by extensive brain imaging research.

London taxi drivers, who rely on memory for navigation over long periods, show significant enlargement of the hippocampus — the brain region responsible for spatial memory. When GPS navigation became widespread, this advantage disappeared in the new generation of drivers. The same logic applies to cognitive abilities in the AI era: if we continuously outsource judgment, reasoning, and decision-making to AI, the prefrontal cortex responsible for these higher-order cognitive functions will face the risk of "detraining."

What's more concerning is that this process may be irreversible — or at least extremely difficult to reverse. A young person who has relied on AI for thinking tasks since childhood may never have fully developed the neural infrastructure for independent judgment. This is different from an adult who temporarily gets lazy and then picks up old skills again — at least there are already established neural pathways that can be reactivated.

Collective Deterioration of 'Cognitive Infrastructure'

If this cognitive outsourcing were merely a matter of personal choice, it would at most be a regret in self-development. But when it becomes a societal, generational trend, its impact reaches the level of civilization.

A functioning society depends on a large number of individuals possessing independent judgment. Democracy assumes citizens can independently evaluate policies, distinguish truth from falsehood, and make value-based choices. The judicial system assumes judges and jurors can make moral judgments in complex situations. The business ecosystem assumes entrepreneurs and managers can make strategic decisions amid uncertainty. When these abilities systematically deteriorate across the population, the foundations of social institutions will be eroded.

We are already seeing some disturbing early signals. Multiple surveys in 2024 showed that university professors widely reported a perceptible decline in students' argumentation and critical thinking abilities following the widespread adoption of AI tools. Corporate managers reported that younger employees increasingly tend to seek AI "confirmation" at every decision point rather than making decisions and bearing the consequences themselves.

Not Against AI, but Against Using AI Thoughtlessly

To be clear, pointing out these risks is not the same as opposing AI. AI's value as a tool is beyond question: it can help us manage information overload, accelerate repetitive work, and provide reference opinions in areas beyond our expertise. The problem is not AI itself, but how we interact with it.

The difference between a conscious AI user and an unconscious AI dependent is enormous:

• The former thinks first, then uses AI to verify and supplement; the latter lets AI think directly and accepts the result.
• The former critically evaluates AI output as "another opinion"; the latter directly adopts AI output as "the answer."
• The former deliberately preserves space for independent thinking practice; the latter chooses to outsource at every opportunity to take a shortcut.

This distinction sounds simple but is extremely difficult to maintain in practice. Because cognitive effort is inherently against human nature — the brain naturally tends to conserve energy, and AI offers an unprecedented "cognitive energy-saving" option. Resisting this temptation requires deliberate, sustained, counter-instinctive effort.

We Need a Culture of 'Cognitive Fitness'

Just as modern sedentary office workers need to deliberately schedule exercise to maintain physical function, knowledge workers in the AI era may need to deliberately schedule "cognitive fitness" to maintain their thinking abilities.

This means: regularly engaging in deep thinking exercises without AI assistance; insisting on forming your own judgment before consulting AI on important decisions; consciously preserving spaces in daily life that require independent judgment; and re-emphasizing training in argumentation, debate, and independent analysis within education systems.

Some forward-thinking educational institutions have already begun taking action. Rather than simply banning students from using AI, they have designed "think first, then use" teaching frameworks that require students to submit their own preliminary analysis before using AI, then compare AI output with their own thinking and reflect on the reasons for any differences. This approach neither rejects AI nor compromises the opportunity to practice independent thinking.

The Final Question

Throughout human history, every major technological revolution has been accompanied by the loss of certain abilities and the acquisition of others. The invention of writing weakened the tradition of oral memory. The printing press changed the monopoly of knowledge elites. The internet reshaped how attention is distributed. AI's impact on cognitive abilities may be the latest — and most profound — link in this chain.

But unlike previous shifts, what is being affected this time is not a specific skill but "judgment" — the core capacity of human cognition. If we lose the ability to judge independently, we will no longer be users of AI but executors of AI — receiving instructions, completing tasks.