At the end of last year, I conducted an experiment: I used four AI models to simultaneously analyze the same business decision, had them debate each other, and finally used Perplexity for fact-checking.

GPT-4o provided a robust, mainstream answer. Gemini identified three risks that GPT hadn’t considered. Grok approached from a completely different angle, proposing an alternative solution I hadn’t even thought of. After three rounds of debate, what I gained wasn’t one answer, but a decision map.

In that moment, I realized: my role isn’t to answer questions—it’s to design how questions get answered.

This is the position I found in the AI storm.

Tool Anxiety is a Trap

When generative AI exploded, many people’s first reaction was anxiety. Writing gets replaced, code gets generated, reports get automated. As if all the skills we spent time practicing suddenly lost their scarcity overnight.

But the real question isn’t “will I be replaced,” but rather: what is your original value built upon? If value comes from repetitive output, then yes, it will be compressed. But if value comes from structural design and judgment capability, the situation is completely different.

Many people, faced with AI, choose to learn more tools, subscribe to more platforms, chase more updates. This is speed anxiety. But tools will always update, models will always upgrade. If competition is built on proficiency, you’ll never catch up to algorithms. What truly needs to change isn’t your skill list, but your role positioning.

From Executor to Orchestrator

Traditional professional division of labor emphasized “completing tasks”—writing articles, making presentations, analyzing data. Now these can all be partially automated. So where is human value?

My own experience is that value shifts to three levels.

Problem Definition. AI is very good at answering questions, but it cannot judge which questions are worth asking. In my piece “The AI Era’s Always-On Economy,” I discussed how when decision frequency far exceeds human comprehension speed, humans slip from decision-makers to observers. Those who can redefine problems are the ones who control direction.

Structural Design. Models can generate content, but who designs the overall architecture? Who decides priorities? I now manage eight social media platforms simultaneously, not through manual posting, but by designing an automated pipeline from Apple Notes to Google Sheet to OneUp API. Content output is five times what it used to be, but I spend less time “writing”—because I spend time designing processes instead.

Value Judgment. Efficiency doesn’t equal value. AI can help you create optimized solutions, but it won’t bear the consequences for you. When I run debate engines, three models often give mutually contradictory advice. I’m still the one making the final decision. Judgment is humanity’s last line of defense.

Role Army Isn’t a Metaphor

When I say “role army,” it’s not rhetoric—it’s the working method I use every day.

GPT-4o is a steady analyst, skilled at structured output. Gemini is a nitpicking critic who finds holes you didn’t think of. Grok is an unconventional creative source, often proposing unorthodox but valuable perspectives. Perplexity is the fact-checker, responsible for filtering out hallucinations.

The key isn’t which model is better, but how you orchestrate them. As I wrote in “Super Individual Case Study,” one person plus a model army can generate the output of a small team. But the prerequisite is upgrading from “user” to “orchestrator.”

Draw Your Strategy Map

Facing the storm, I chose to do one thing: draw my own strategy map.

Three questions: Which capabilities will be compressed? Which capabilities will be amplified? Which capabilities can only I integrate?

For me, writing speed will be compressed, but cross-domain integration won’t. Data analysis will be compressed, but the intuition from ten years of industry experience won’t. Individual skills will be compressed, but the ability to connect theological training, entrepreneurial practice, and technical understanding won’t.

This isn’t skill inventory—it’s positioning reconstruction. No longer pursuing faster execution, but thinking about how to design better order.

Every technological revolution brings elimination, but what’s usually eliminated isn’t people—it’s old division-of-labor logic. If you still define yourself by old order, you’ll feel squeezed. If you start designing new order, you’ll see the space is much larger than imagined.