2 AI habits that can weaken your critical thinking, according to a BCG exec

2 AI Habits That Weaken Critical Thinking, And What They Mean for the Future of AI at Work

By · Published October 5, 2026 · Updated October 5, 2026

AI tools are now woven into the daily rhythm of white-collar work. Chat assistants draft emails, summarize research, write code, build slides, and answer questions in seconds. The productivity gains are real, and for many teams they are already baked into how work gets done.

But a warning is now coming from inside the highest levels of the consulting industry. A Boston Consulting Group executive has flagged two specific AI habits that quietly erode the very skill that makes professionals valuable: critical thinking. The message is not "stop using AI." It is sharper and more uncomfortable than that. The habits themselves are the problem, and they are spreading fast because they feel efficient, frictionless, and harmless.

Here is what those habits look like, why they matter so much right now, and what the future of AI adoption depends on getting right.

The Warning Comes From Where AI Advice Matters Most

Consulting firms sit at the crossroads of strategy and execution. They are hired to think. So when a senior figure at BCG points at AI habits that dull thinking, it is worth paying attention, not because it is anti-technology, but because it reflects what happens inside organizations that already deploy AI at scale across thousands of knowledge workers.

The concern is structural, not moral. AI is very good at producing plausible output. That is precisely what makes it dangerous to the human habit of reasoning. When an answer arrives instantly and reads confidently, the brain does what brains do: it accepts the path of least resistance. Over time, the mental muscle that used to be exercised by wrestling with a hard problem simply stops getting used.

The two habits at the center of this warning share one root cause: handing over cognitive work that used to belong to you.

Habit 1: Letting AI Think First, Instead of Thinking With It

The first habit is treating AI as a substitute for your own thinking rather than a partner in it. The tell is simple: the tool generates the first idea, the first outline, the first analysis, and the human never forms an independent view before reading it.

This matters because of how human judgment actually forms. When you build a rough answer in your own head first, even a wrong one, you create a scaffold. You know where the holes are. You know which assumptions you made. You can then compare the AI's answer against your own and spot where it diverges.

When you skip that step, you lose the comparison entirely. There is nothing to check the output against. The AI's answer becomes the answer, and your role shrinks from thinker to reader.

The subtle part is that this habit does not feel like a loss. It feels like speed. The task is done, the document is written, the meeting moves on. Nobody notices the missing step because the output looks the same. That is what makes it so hard to catch, the damage is invisible on the surface and only shows up months later, when the team can no longer explain why it believes what it believes.

Why the First Draft in Your Head Still Matters

Professionals who keep this habit intact describe a small but powerful routine: spend five minutes forming your own rough take before opening the AI. It does not need to be good. It needs to exist. That rough sketch becomes the anchor that keeps you in the driver's seat instead of the passenger seat.

Habit 2: Accepting the Confident Answer Without Interrogation

The second habit is trusting AI output at face value. This one is more visible than the first, but it is just as sticky, because the output is often polished in ways that signal reliability: clean grammar, structured arguments, confident phrasing, and specific-sounding details.

Confidence, however, is not accuracy. AI systems are built to generate likely text, which means they produce answers that sound right. Polished language is a style, not a proof. A well-written paragraph can still contain a fabricated detail, an outdated assumption, or reasoning that collapses under one good question.

The habit of interrogation, asking "why," "says who," "what's missing," "what would make this wrong", used to be a natural part of professional work. It was slow, but it was automatic. When AI compresses that work into a single acceptance step, the questioning muscle atrophies.

The compounding effect is the real risk. One unchecked answer is harmless. A thousand unchecked answers become the shared knowledge base of a team, then of a company. At that point, nobody can trace where a belief came from, and correcting it becomes far more expensive than checking it would have been.

Why These Habits Spread So Easily

Both habits are reinforced by the same incentive: speed. AI's core promise is faster output, and faster output is measured, rewarded, and promoted. Slower, more deliberate use, pausing to think first, pausing to verify, looks like friction in a world that treats friction as waste.

This is the central tension of the AI era. The tool removes effort, and effort is exactly what builds skill. Remove too much of it, and you get short-term gains paid for with long-term capability.

There is also a generational and cultural layer. Workers who learned their craft before AI assistants existed have a library of independent judgment to draw on. Workers entering the field now may never build that library if their first instinct is always to ask a model. That gap, between people who can evaluate AI output and people who can only consume it, may become one of the most important divides in the workforce.

What This Means for the Future of AI

Two big shifts follow from this warning.

First, the value of AI will shift from generation to evaluation. Producing a draft is becoming cheap and nearly universal. Knowing which draft is right, which detail is fabricated, and which assumption is fragile is becoming the scarce skill. Companies that win with AI will not be the ones that generate the most, they will be the ones that judge the best.

Second, AI tools will be judged on whether they make people smarter or just faster. The current wave is optimized for output. The next wave will be judged on whether it strengthens or replaces human reasoning. That is a design problem, a training problem, and a measurement problem all at once.

Expect this to change how AI is built. Tools that show their reasoning, flag uncertainty, surface their sources, and push back on the user are already being positioned as the more "responsible" option, not just because they reduce errors, but because they keep the human in the loop instead of letting the loop close without them.

What It Means for Businesses

For leaders, the practical implication is that AI adoption cannot be measured by usage alone. "Everyone is using it" is not a strategy. The companies that get this right will track something harder to see: whether people are still thinking.

A few shifts are likely to matter:

None of this requires slowing AI adoption down. It requires deciding what the human is still there for.

A Practical Playbook for Individuals and Teams

The good news is that these habits are easy to interrupt. Small changes restore most of the lost benefit.

The Bigger Picture

The most important question about AI is not "how smart is the model?" It is "what happens to the people using it?"

AI is a genuine leap forward. It removes drudgery, widens access to expertise, and lets small teams do work that once required large ones. None of that goes away because of these two habits. But the habits determine whether AI becomes a tool that amplifies human judgment or a tool that quietly replaces it.

The future of AI at work is not a contest between people and machines. It is a contest between two versions of the same worker: the one who thinks first and checks the answer, and the one who accepts the first confident paragraph. The first version gets faster and sharper. The second gets faster and, over time, less capable, and often does not notice until the moment it matters most.

That is the real lesson behind the warning. AI does not weaken critical thinking on its own. Habits do. And habits, unlike models, are something we choose.

TLDR: A Boston Consulting Group executive warns that two common AI habits, letting the tool do your thinking instead of thinking with it, and accepting confident output without questioning it, can quietly erode critical thinking at work. As AI makes drafting and analysis nearly free, the scarce skills become evaluation, verification, and judgment. Businesses should design workflows where AI sits after human reasoning rather than before it, treat verification as a required step, and measure quality alongside speed. The future of AI depends less on how smart the models get and more on whether the people using them keep thinking first.