Conclusion First

The future divide in the workplace won’t be about “using AI” but whether you “control AI” or “are controlled by AI.”
The biggest risk of AI is not job loss, but rather the slow “outsourcing” of our cognitive abilities, leading to cognitive decline.
Don’t see AI as an “outsourced worker” to get tasks done; view it as a “sparring partner” to stimulate your thinking. Every question should be a deep dialogue you lead.
The core competency in the AI era: when faced with AI outputs, the most valuable response is not “one-click acceptance,” but your active questioning based on professional judgment.

The Loss of Deep Thinking Is Not Far Off

The world is going crazy for AI. In just two years, nearly 1 billion people have flocked to OpenAI’s products. This exemplifies the typical Silicon Valley growth flywheel: create outstanding products, offer affordable prices, make them irresistible, and then find a way to reap billions in profit.

We are embracing AI because it provides unprecedented “cognitive shortcuts.” However, for most people, the destination of this “fast train” might not be pleasant. Initially, we allow AI to take on some tasks, but soon we discover it can do almost everything. Ultimately, we might lose our ability for deep thinking, our essential jobs, and future opportunities.

But this isn’t inevitable. You can choose to see the path ahead clearly and carve out a completely different route.

The Beginning of the Endgame

In March 2023, I used ChatGPT for the first time. Now, ChatGPT or Claude has become an indispensable part of my daily routine. AI has drastically increased my work efficiency, but at the same time, I find myself growing cognitively lazier.

In the past, I would meticulously check every word of the drafts produced by AI. But nowadays, it can deliver a decent initial draft 90% of the time, and I’ve gradually lost the motivation to proofread carefully.

A year ago, I thought the future workplace would split into two types of people: those who use AI and those who don’t. Now, I’ve realized I was wrong. Within five years, AI will be a necessary tool for everyone. At that point, the real divide will be between those who harness AI and those who outsource their thinking to AI.

“Thinking Outsourcing”: The Start of Capability Decline

Humans have long sought ways to “outsource” cognitive tasks. Before books existed, bards could recite the entire text of Homer’s “Iliad.” Today, technology extends our brains, allowing us to “outsource” tasks like mathematical calculations, navigation, and note-taking.

But AI is different. It can handle almost any cognitive task and gives the impression of high efficiency. Thus, “AI outsourcing” often begins unwittingly. You ask AI to draft an email, and it completes the task swiftly and well, saving you 10 minutes. Next, you ask it to outline a presentation, and it hits the mark perfectly.

Gradually, you start using it for more complex tasks, like strategic planning. You find yourself relying on AI to complete your work, and your professional skills slowly begin to atrophy.

A paper published by Microsoft and Carnegie Mellon University indicates that generative AI can undermine our critical thinking skills. When knowledge workers are overly confident in AI outputs, they become less inclined to engage their own brains.

Like many who trust AI, I often rely on my own judgment to fact-check its outputs. However, there are two issues: 1) we often overestimate our ability to spot AI errors; 2) the temptation to skip fact-checking grows stronger.

AI Controllers vs. AI Passengers

In the next decade, knowledge workers will be divided into two categories: AI Controllers and AI Passengers.

AI Passengers will eagerly delegate their cognitive work to AI. They will paste prompts into ChatGPT, copy the results, and submit them as their own.

In the short term, they may be rewarded for their speed. However, as AI operates without human oversight, these passengers will be regarded as redundant since they contribute no additional value to AI outputs.

AI Controllers, on the other hand, will remain in command of AI. They will treat AI outputs as initial drafts and scrutinize its work rigorously. They will also occasionally turn off AI to give themselves time for independent thought.

Over the long term, the economic gap between these two groups will widen dramatically. AI Controllers will amass disproportionate wealth, while AI Passengers will become easily replaceable.

How to Become an AI Controller

Here are steps to ensure you remain the master of AI:

Start in areas you know. Use AI in fields where you have expertise and maintain a critical stance toward its outputs.
Engage in dialogue with AI, rather than just asking for answers. Instead of simply asking, “How should we allocate our marketing budget?” set parameters, provide information, offer options, and debate with it.
Stay highly vigilant. Participate actively and don’t take outputs for granted. Challenge yourself to reflect: “Is this really a good suggestion?”
Practice active questioning. Continuously test AI with your viewpoints. “Does this plan underestimate the risks involved in the project?”
Resist the urge to outsource all initial drafts. Facing a blank page might be daunting, but it’s a crucial step to activating your brain.
Make the final decision and take responsibility for it. AI should assist with every medium to high-risk decision, but it cannot make the call for you. As a human, you must be accountable for your choices.

Your Brain Should Not Be Wasted

With AI, you now have a round-the-clock thinking partner equipped with “expertise” on any topic.

Yet, you stand at a crossroads. You will see many colleagues choosing to give up “active thinking” and outsource their decision-making to AI. Many won’t realize their cognitive abilities are declining until it’s too late to salvage much.

Don’t be that person. Use AI to challenge and enhance your thinking, not to replace it.

The question is not, “Will you use AI?” but rather, “What kind of AI user do you want to be: a Controller or a Passenger?”