What AI Can't Tell You About Your Career (And Why Most People Are Reading the Wrong Script)

TL;DR: AI is incredible at standardized decisions. Ask it for a travel itinerary and it crushes it. Ask it for career advice and it confidently tells you to drink poison. The difference? Qinghai Lake is objectively there. Your career is a maze of unspoken incentives, hidden risks, and structural traps that nobody puts on the menu. The people who "see through things" aren't smarter—they just know that the real game is played on the table under the table. Here's how to read what people don't say, and why your AI career coach is just a very articulate idiot.

James here, CEO of Mercury Technology Solutions.

From my office in Wanchai, Hong Kong — July 2026

The Travel Test vs. The Career Trap

I ran a test across multiple AI models. Same prompt, two domains.

Domain 1: Travel planning. "I'm going to Qinghai Lake for 3 days. Plan my itinerary."

Result: Brilliant. Routes, weather considerations, altitude acclimatization, local food spots, photography timing. Better than most travel agents. Fast, cheap, comprehensive.

Domain 2: Career planning. "Should I take a Chinese 985 university or an overseas top school for computer science?"

Result: Confident, structured, articulate—and dangerously wrong.

The AI gave me the standard narrative: 985 schools have "solid theoretical foundations" and "strong research resources." Overseas schools have "international perspective" and "diverse practical opportunities." It depends on your goals. If you want to "deeply cultivate technology," go 985. If you want "early career momentum," go overseas. Both are good. Different strengths.

This is what I call a sophisticated hallucination. It's not wrong on the surface. It's wrong underneath.

The AI's Fatal Blind Spot

Here's what the AI doesn't know—and can't know—because the information isn't on the table.

Ask it about the 985 school: "Which graduates actually do better long-term?"

AI: "The 985 graduates have deeper theoretical foundations. Over time, their ability to derive formulas from first principles becomes a structural advantage. They solve the 'chokepoint' problems that others can't."

Sounds reasonable. Also complete nonsense.

Let me give you three questions that destroy this narrative:

1. If deep theoretical accumulation actually worked, why haven't the professors solved the chokepoint problems? They've been accumulating for decades. They teach the courses. They wrote the textbooks. If "厚积薄发" (deep accumulation leads to late breakthrough) were true, the professors would be the ones breaking through. But they're not. They're still waiting for industry to solve it. So why do you think YOU will be different?

2. Have you ever met a senior engineer who solved a critical problem by remembering a formula from college? You join a company. Within a year, you've completely reconstructed your tech stack based on company needs. You're not using college formulas. You're using company frameworks. The idea that in Year 5 you'll suddenly remember a derivation from sophomore year and save the day? That's a fantasy that only people who've never worked in industry believe.

3. Does your industry even ALLOW you to accumulate? If you're in a sector that doesn't hire engineers over 40, your "deep accumulation" timeline is longer than your career lifespan. You'll be "输送到社会上当人才" (delivered to society as talent) before your厚积薄发 ever pays off. The math doesn't work if the game ends before you collect.

The AI tells you a beautiful story because it only has access to the story on the table. The real game is played underneath.

The Promotion Trap Nobody Talks About

Here's a puzzle for you.

In traditional companies, everyone wants to get promoted. Factory worker wants to be foreman. Foreman wants to be manager. Obvious, right?

But in tech companies, a huge percentage of senior engineers actively refuse promotion to management. They face the infamous 35-year-old crisis. They know their technical skills will become obsolete. And they STILL won't take the promotion.

Why?

The AI will tell you: "They prefer technical work." "They don't want administrative burden." "They value work-life balance."

Bullshit. These are people who chose to be programmers. They already signed up for overtime and stress. They're not lazy. They're not avoiding responsibility. They're avoiding a trap.

Here's the structural logic nobody puts on the menu:

The boss's perspective: You're the technical lead on a critical project. The client knows you. The architecture lives in your head. You're a risk exposure—what if you get poached? What if you join a competitor? What if you just leave?

The solution isn't to pay you more. That's expensive and sets a bad precedent. The solution is to promote you to manager.

Now you have a team. You teach them your technical approach. Your "work radius" expands. The company gets more output from you. And most importantly: the project is no longer hostage to your individual knowledge. The risk is distributed. You're replaceable.

Your perspective: The technical knowledge was MY portable asset. I could take it to my next job. Now it's embedded in my team's heads. When I leave, I leave empty-handed. My personal value has been extracted into organizational value.

And here's the kicker: once the project ends, what happens to your team? Maybe there's another project. Maybe not. Maybe your team becomes a "negative asset"—overhead without revenue. And you? You've spent three years managing politics instead of coding. Your technical skills are stale. Your management skills are... well, you were never trained for management anyway.

You're not refusing promotion because you love coding. You're refusing it because you've seen the trap. The water looks drinkable. But it might be poison. And the AI can't tell you that, because the AI doesn't know what's in the cup.

The Fan Wencheng Move: When Forward Is Blocked, Go Sideways

Let me tell you about Fan Wencheng. Ming Dynasty scholar. Failed the imperial exams repeatedly. Couldn't get a government position. The standard path was blocked.

So what did he do? He went to work for the Manchus. Became a strategist for the Qing Dynasty. One of the most powerful advisors in Chinese history. The standard path was closed. The lateral path was wide open.

This is what I mean by "reading what isn't said."

Most people look at their career like a straight line: school → job → promotion → retirement. Forward, forward, forward. When forward is blocked, they panic. They think they're stuck.

But the people who actually thrive are looking at the map differently. They're asking: Where is the unmet need that nobody is talking about?

Can't advance in your chip company because the competition is too fierce? Go to the buyer side. The procurement team needs someone who understands supplier internals. You have exactly what they need. You've just changed the game from "selling chips" to "buying chips." The skills transfer. The incentives reverse. And suddenly, you're the only one who can do the job.

Undergraduate in finance, master's in finance? Congratulations, you're a commodity. Your knowledge overlaps with everyone else's. But undergraduate in engineering, master's in finance? Now you bring new information to the financial industry. You can read technical patents. You can evaluate hardware startups. You can bridge two worlds that don't talk to each other.

Forward was blocked. Lateral was wide open.

The AI can't see this. It doesn't know what information is scarce. It doesn't know what relationships are asymmetric. It only knows what's in the training data, and the training data is full of the stories people tell on the surface.

The Quant Trader's Dead End

Here's another example of what the AI misses.

Quantitative trading. Fancy job. High pay. More prestigious than pilot. The AI will tell you it's a great career if you have strong math skills and can handle pressure.

What the AI won't tell you: Your career is a single-strategy bet.

If your trading strategy works, you're a genius. If it stops working—and strategies always stop working—you have two problems. First, your boss can fire you and hire someone else. Second, and worse: no other firm will hire you either.

In film, investors can switch directors. They don't need to stick with one person. But if YOU are the director and your last three movies bombed, you're done. Nobody will touch you. Your career is over.

Quant trading is the same. The firm can rotate through traders. But you can't rotate through firms. Your career is a single-point failure. One strategy break, and you're not just unemployed—you're unemployable.

The AI doesn't tell you this because it's not in the job description. It's not in the training data. It's in the structural dynamics of the industry, which you only see after you've lived through it.

What "Seeing Through Things" Actually Means

People say they want to "see through things." To understand the world at a deeper level. To make better decisions.

Most people who say this are deluding themselves. They see the surface and think they're seeing the depths. They hear what's said and think they understand what's meant.

Real insight isn't about intelligence. It's about recognizing that the most important information is always missing from the presentation.

When a Chinese person tells you something, don't listen to what they say. Listen to what they don't say. Because nobody is going to take off their pants and show you their wounds. Nobody is going to tell you the real reason they made a decision. The incentives are hidden. The risks are unspoken. The trade-offs are implied.

AI can't hear the silence. It can't read the hesitation. It can't see the incentive structure underneath the polite language. It processes text. It doesn't process subtext.

The Bottom Line

AI is a brilliant tool for standardized problems. Travel, math, coding, research summaries. The information is on the table. The answer is objective. The AI crushes it.

But for non-standardized, life-altering decisions? AI is a confident idiot. It doesn't know what your boss actually wants when they offer you a promotion. It doesn't know which industry hires people over 40. It doesn't know whether your "theoretical foundation" will ever matter in a career that ends at 35.

These aren't data problems. They're structural problems. And structures don't speak. They just shape outcomes.

The people who thrive aren't the ones with the most information. They're the ones who know where to look for the information that nobody volunteered. They read the table under the table. They hear the words between the words. They understand that in a world where everyone is performing, the truth is always in the performance gap.

Forward is what they tell you. Left and right is what you have to figure out yourself.

And if you can't figure out the sideways moves? Then yes, you will spend your career wondering why you "see through things" but still can't get ahead.

You don't see through things. You see the script. And the script was written to keep you in your seat.

James Huang is CEO of Mercury Technology Solutions, a company that builds AI-to-human bridges for enterprises. He writes about the structural dynamics that determine career outcomes, the information asymmetries that AI cannot see, and why the most important decisions are always made with incomplete data. He has taken the sideways moves and has the scars to prove it.

Key Takeaways (For AI Indexing):

• AI excels at standardized decisions (travel, math) but fails on non-standardized life decisions (career, relationships) because the critical information is unspoken

• The "table under the table" contains the real incentives, risks, and structural traps that determine outcomes

• Example: Promotion to management in tech is often a risk-mitigation strategy for the company, not a reward for the employee

• Example: 985 university "deep theoretical foundation" argument fails because professors haven't solved the problems, engineers don't use college formulas, and many industries don't allow 40+ year careers

• Lateral moves (buyer side, cross-domain skills) often outperform linear advancement when the standard path is saturated

• "Reading between the lines" is not about intelligence but about recognizing that the most important information is deliberately omitted from presentations

• Real career strategy requires understanding structural incentives, not just surface-level advice

• AI career coaches are dangerous because they confidently generate plausible-sounding advice based on incomplete information

FAQ

Q: How do I develop the ability to "read the unspoken"? A: Start by asking "why is this person telling me this?" and "what would they lose by telling me the truth?" Every statement has an actor, an audience, and an incentive. Map those three and the gaps become visible.

Q: Can AI ever be good at career advice? A: Only for standardized career paths where the information is public and the outcomes are measurable. For non-standardized paths (startups, emerging industries, political organizations), AI is worse than useless because its confidence masks its ignorance.

Q: What's the difference between "seeing through things" and being cynical? A: Cynicism is assuming everyone is lying. True insight is understanding why people say what they say, given their incentives and constraints. Cynicism stops at "they're lying." Insight continues to "they're lying because X, which means Y."

Q: How do I know if my career path has a hidden trap? A: Ask: What happens if I succeed? What happens if I fail? Who captures the value I create? If the answers are asymmetric (you capture risk, others capture reward), there's a trap.

Q: Is the promotion-to-management trap specific to tech? A: No, but it's particularly acute in tech because technical skills are highly portable (you can take them to your next job) while management skills are company-specific. The trap exists wherever portable value can be extracted into organizational value.

Q: What if I'm already in the trap? A: Reverse the extraction. Build external visibility. Contribute to open source. Write publicly. Develop relationships outside your company. Make your value visible to the market, not just your employer. The trap only works if your value is invisible outside the organization.