Everyone's 'AI-Proof Career' Advice Is Wrong. Here's The Real Plan B.
Q: Everyone is talking about AI. How are you thinking about your own career to stay relevant over the next 10 years? What is your "Plan B"?
Why this matters: This isn't a question about your coding skills. It's a test of your strategic thinking. Can you see the second-order effects of a technological revolution? Do you react to trends, or do you reason from first principles?
Interview frequency: High and increasing. This is the new "where do you see yourself in 5 years?"
❌ The Death Trap
The candidate gives a generic, low-conviction answer. They either repeat headlines ("I'm learning prompt engineering") or retreat to platitudes ("I'll focus on soft skills"). The worst answer is clinging to a craft, like plumbing, based on a shallow understanding of automation.
"Most people say: 'AI won't be able to do everything, so I'm focusing on uniquely human skills like communication. Or I might specialize in a trade like plumbing, because robots can't do that. As an engineer, I'll just keep up with the latest AI tools.'"
This answer shows you're a consumer of opinions, not a producer of insights. It fails to grasp the physics of the new economy.
🔄 The Reframe
What they're really asking: "Do you know how to identify the critical, non-obvious bottlenecks created by exponential technological growth, and can you position yourself there to capture value?"
This reveals you understand that wealth and relevance aren't created by doing what the new technology does. They're created by providing what the new technology *needs*. You're not trying to out-race AI; you're looking for the chokepoint where you can tax it.
🧠 The Mental Model
My "Plan B" isn't a specific job title. It's a framework I call the **"Bottleneck Hunter's Thesis."** The greatest opportunities aren't in building the AI; they're in solving the five systemic bottlenecks that AI's own success creates.
📖 The War Story
Situation: "During the early days of the cloud computing boom, I saw my peers rushing to become web developers, building applications on top of AWS."
Challenge: "It was a gold rush. Everyone was building apps, and the space became incredibly crowded. The value was being captured by the platform—Amazon—not by the thousands of developers building on top of it. I realized racing to do the obvious thing is a losing game."
Stakes: "Choosing the wrong specialization means spending years building a skill set that becomes a commodity, where you're always competing on price instead of unique value. It's a career dead-end."
✅ The Answer
My Thinking Process:
"My lesson from the cloud boom is that you don't chase the gold; you sell the shovels. Or better yet, you find the one thing the shovel-makers can't get enough of. With AI, the same logic applies. I'm not focused on becoming a better model-tuner than the next guy. I'm looking at the second-order effects—the systemic bottlenecks."
What I Did:
"Applying my 'Bottleneck Hunter's Thesis,' I see five clear areas of immense, durable value over the next decade. These are the real Plan B's:"
1. The Energy Bottleneck - Electrical Engineering: "AI data centers have an exponential appetite for energy, but the world built a linear grid. The single biggest constraint on AI's growth isn't algorithms; it's megawatts. The people who can build and maintain the power infrastructure to feed the beast will have immense leverage."
2. The Silicon Bottleneck - Semiconductor Engineering: "AI isn't software; it's silicon. The entire revolution runs on chips made in a handful of fabs in a few countries. This is the ultimate physical chokepoint. If you can design or manufacture those chips, you are operating at the very base of the new value stack."
3. The Trust Bottleneck - AI Cybersecurity: "As AI agents proliferate, the attack surface for our entire society expands infinitely. We're entering an era of AI vs. AI warfare. The human in the loop, the person who builds the defenses and has the ultimate override, becomes the most critical node in the network. This isn't a job; it's a strategic imperative."
4. The Physical Bottleneck - Nursing: "This is Moravec's paradox in action. AI can read an MRI better than a radiologist but can't comfort a patient or administer an IV. The world of atoms, empathy, and complex motor skills is surprisingly hard to automate. The economic incentive is also lower. The value is in the high-touch, physical interface to the human body."
5. The Sanity Bottleneck - AI Psychology: "AI will automate away not just jobs, but sources of identity. We're facing a societal-scale crisis of purpose and a potential epidemic of AI dependency. The most valuable service won't be debugging code; it will be debugging the human psyche. This must be done by humans, creating a massive, uniquely human profession."
What I Learned:
"The key to career survival isn't to be 'un-automatable' in the absolute sense—nothing is. The key is to find a domain where the *demand* for your human skill is growing exponentially faster than automation's ability to supply it, creating a massive, decade-long gap where you can create unique value."
🎯 The Memorable Hook
"Don't compete with AI. Don't collaborate with AI. Find the thing AI is desperately dependent on, and become its sole supplier. Find the bottleneck, and own it."
This reframes the entire 'future of work' conversation. It's not about being a victim or a collaborator; it's about being a strategic operator who understands the physics of a new system and positions themselves at a point of maximum leverage.
💭 Inevitable Follow-ups
Q: "So you're saying software engineering is a dead end?"
Be ready: "Not at all. The *nature* of the job is changing. It's shifting from being a bricklayer to being an architect. The leverage moves from writing rote code—which AI will do—to defining problems, designing systems, and making high-level decisions that AI agents then execute. The most valuable engineers will be the ones who can think clearly about systems, not just code."
Q: "But won't AI eventually solve those bottleneck problems too?"
Be ready: "Eventually, perhaps. But the timelines are mismatched. Building a nuclear power plant or a new semiconductor fab takes a decade and operates in the world of physics and geopolitics. Software moves at the speed of light. These bottlenecks represent a massive arbitrage opportunity in that timeline mismatch. You're building a career on a decade-long certainty, which is an eternity in this environment."
Launch your Graphy