From Noise to Clarity | Track 4

AI AND THINKING — What Happens When You Rely Too Much on AI #10

AI and Thinking | Post 10 of 15

Before you read: think of a thinking task you used to do yourself that you now routinely hand to AI. Do you still have the capability to do it without the tool? Are you certain?

The Risk Nobody Is Talking About

Most people, when asked about the risks of AI, point to misinformation. To hallucinations. To bias in the data. These are real concerns and they deserve attention. But there is a risk that is less visible, less dramatic, and possibly more consequential for the professionals who encounter it.

The risk is dependency. Not the dramatic kind that announces itself, but the quiet kind that develops through repeated small decisions to let the tool handle something you could have handled yourself. And over time, through accumulation, those small decisions change what you are capable of.

AI becomes dangerous not when it gives wrong answers but when it makes you less able to recognise weak ones.

Visual Prompt 1: “The Quiet Risk” A two-column hand-drawn sketchnote on a white background. Left column header: “visible AI risks” with a handwritten underline. Three items: “misinformation,” “hallucinations,” “bias in data.” A handwritten label beneath: “people talk about these.” Right column header: “less visible AI risk” with a handwritten underline. One item in blue, larger than the others: “dependency.” A handwritten label beneath: “fewer people talk about this.” A handwritten note between the columns at the bottom: “the quieter risk may be the more consequential one.” Blue accent on the right column header, the word dependency, and the handwritten note. Black ink throughout. White background. Rough sketch aesthetic.

How the Trade-off Works

The shift into dependency does not feel like a risk when it is happening. It feels like efficiency. You save time. You move faster. You reduce effort. You avoid cognitive strain. Each of these is a genuine benefit, and in the short term the trade feels entirely reasonable.

But thinking is not like a physical skill that stays intact while unused. It is more like a muscle that responds to the level of demand placed on it. When AI consistently handles the questioning, the analysis, the synthesis, the judgment, those capabilities get less use. And capabilities that get less use become less reliable over time, particularly under the pressure of high-stakes situations where they are needed most.

This pattern is not new. Navigation applications weakened spatial memory in people who used them exclusively. Search engines reduced the kind of active recall that builds robust long-term memory. Autocorrect weakened spelling awareness in people who stopped proofreading their own work. In each case, the tool was genuinely useful. And in each case, the underlying capability quietly eroded.

AI now operates at a different level. It affects not spatial memory or spelling, but higher-order thinking: strategic reasoning, synthesis across complex information, judgment under uncertainty, pattern recognition across domains, the ability to frame a decision well before attempting to answer it. These are the capabilities that matter most in leadership. And they are the ones most at risk from the kind of outsourcing that AI makes easy.

Visual Prompt 2: “The Erosion Pattern” A timeline hand-drawn sketchnote on a white background. A horizontal axis labelled “consistent AI use without deliberate maintenance.” Three historical reference points shown as small icons above the axis with handwritten labels: “navigation apps: spatial memory weakened,” “search engines: active recall reduced,” “autocorrect: spelling awareness declined.” Below the axis, a larger downward slope labelled “AI era” with four handwritten labels along the slope: “strategic reasoning,” “synthesis,” “judgment under uncertainty,” “decision framing.” A handwritten note at the bottom of the slope: “same pattern. higher-order capabilities.” Blue accent on the downward slope and its labels only. Black ink throughout. White background. Rough sketch aesthetic.

The Hidden Leadership Risk

There is a specific version of this risk that is worth naming for leaders, because it is harder to see from the outside.

A leader who over-relies on AI can still appear highly productive. The outputs are polished, the responses fast, the recommendations confident. Nothing in the presentation signals that something important is eroding beneath it. Which is precisely what makes it a hidden risk.

What erodes is not visible in the output. It is visible in the quality of thinking that the leader brings to situations where AI is not enough: where the context is too specific to be captured in a prompt, where the stakes are too high to delegate to a tool that cannot be held accountable, where the ambiguity is too genuine to be resolved by a confident-sounding recommendation, where the judgment required is the kind that only comes from sustained independent thinking.

Over time, leaders who outsource too much of their cognitive effort find that independent thinking becomes harder precisely when the pressure is highest. The capacity that was being quietly replaced rather than supplemented is not available when it is needed most.

What Strong AI Users Do Differently

The professionals who use AI well over the long term share a common discipline. They treat AI as augmentation, not substitution. The distinction is not semantic. It describes a fundamentally different relationship with the tool.

Augmentation means AI operates on thinking that the person has already done. They think before they prompt. They form a view before they ask for one. They build their own interpretation of the output before accepting it. They compare what AI generates against a standard they brought to the interaction rather than one the AI supplied. They slow down when decisions matter, not because AI is slow, but because judgment is not something that can be safely accelerated in high-stakes situations.

Substitution means AI does the thinking that the person has not done. The prompt replaces the reflection. The output replaces the interpretation. The recommendation replaces the judgment. Each substitution is individually small. Cumulatively, they change what the person is capable of.

The most valuable capability in the AI era may not be the ability to use AI effectively. It may be the ability to think clearly when the machine is unavailable, incomplete, persuasive, or wrong. That capability is built through deliberate practice. And it erodes through neglect.

Visual Prompt 3: “Augmentation vs Substitution” A two-column hand-drawn sketchnote on a white background. Left column header: “augmentation” with a handwritten underline in blue. A figure with a full structured thought cloud above their head. An arrow pointing from the thought cloud down to a prompt box, then right to an AI icon, then right to an output box. A handwritten label beneath: “thinking first. AI extends it.” A solid upward arrow beside the figure labelled: “capability maintained.” Right column header: “substitution” with a handwritten underline in black dotted style. A figure with an empty thought cloud above their head. An arrow pointing directly from the figure to an AI icon, bypassing any prior thinking, then right to an output box. A handwritten label beneath: “AI thinks first. person follows.” A dotted downward arrow beside the figure labelled: “capability erodes.” A handwritten note between the columns at the bottom: “same tool. opposite trajectory.” Blue accent on the left column header, the full thought cloud, and the upward arrow only. Black ink throughout. White background. Rough sketch aesthetic.

A Reflection Worth Sitting With

The question is not whether to use AI. The question is whether your use of AI is building your thinking or quietly replacing it. That distinction is not always comfortable to examine. But it is worth examining regularly, because the erosion, when it happens, does not announce itself.

Identify one thinking capability you consider important to your role. Now ask honestly: are you currently exercising that capability, or are you routinely handing it to AI? What would it take to maintain it deliberately?

The next post looks at what it means to develop AI literacy as a leadership capability, and why understanding how AI thinks is as important as knowing how to use it.

I am Ai Yat Goh, a visual thinking strategist and co-author of The S.T.A.R. System. I write about thinking clearly in a world full of noise. Follow me here on Medium for the rest of this series, or connect with me on LinkedIn.

VisualThinking #ArtificialIntelligence #Leadership #CriticalThinking #StrategicThinking

BK HAN GOW Hui Yian Candy Chan RH Malini Angela Toh Germaine Kwek