Loading playbook…
THE DIARY OF A CEO · EXTRACTED
7 strategic moves for navigating the AI transition — the AI safety researcher's playbook for which careers survive, which die, and how to position yourself.
"In five years, AI will be better than humans at almost every intellectual task. This is not speculation. It's an extrapolation of rates we can already measure. The question is not whether it happens. It's what you do between now and then."
Dr. Roman Yampolskiy is a computer scientist at the University of Louisville and one of the most prominent researchers in AI safety. His warnings about artificial general intelligence are taken seriously by people who disagree about almost everything else. In this Diary of a CEO conversation, he makes claims that would have sounded outrageous three years ago and now sound merely uncomfortable: that AI will match or exceed human performance in nearly every white-collar task by 2030, that most current career paths are heading into obsolescence, and that only a narrow set of jobs will remain economically defensible. The conversation isn't panic — it's strategic positioning. Whether Yampolskiy is right about the specifics or not, his framework for which human capabilities survive AI commoditization is worth taking seriously.
Yampolskiy identifies five categories of work that remain economically valuable even as AI matches human intellectual capacity: (1) Physical labor in unstructured environments — plumbing, electrical work, elder care; (2) Creative work where novelty matters more than accuracy — original artists, scientists pushing frontiers; (3) Leadership and high-accountability roles where human responsibility is legally required; (4) Interpersonal work where the value IS human connection — therapy, high-end coaching, spiritual guidance; (5) AI specialists themselves. Everything else is at risk of commoditization. 'These five aren't immune forever. But they're the last to fall.'
THE PLAY
Audit your current work against these five categories. Which one, if any, does your role map to? If none, begin shifting immediately — either toward one of the five or toward building something that compounds independent of your labor (equity in a business, a creative asset, a brand). The adjustment is much easier while current income is stable than after disruption hits.
Yampolskiy's sharpest insight: in an AI-disrupted world, labor income becomes unreliable, but asset income becomes critical. Skills that produce equity, content, or intellectual property accumulate value over time independent of whether you keep working. A writer with a back catalog. An investor with a portfolio. A founder with equity. These positions survive AI disruption because they're rent-collecting structures, not labor streams. 'Trade labor for assets while you still have leverage. Labor gets cheaper. Assets get more valuable.'
THE PLAY
Every year, build one asset that continues earning or compounding without your direct labor: a book, a course, a small business, an investment position, a piece of intellectual property. Don't replace your income with assets — add them in parallel. By the time labor disruption arrives, you'll have multiple compounding positions that don't require your continued output.
Yampolskiy argues that the generalist middle is the most vulnerable to AI replacement. Jobs requiring moderate skill across many areas — middle-management, general-purpose analysis, most office work — get commoditized first. The extremes survive: deep specialists in narrow domains, and flexible generalists with unusual trait combinations. 'The mediocre middle dies first. Go narrow enough that AI can't replicate the specific expertise, or unusual enough that no training data captures the combination.'
THE PLAY
If your current role is broadly defined — 'marketing manager', 'business analyst', 'project coordinator' — you're in the disrupted middle. Specialize into a narrower domain that requires deep expertise: a specific industry, a specific technical area, a specific methodology. The narrower and deeper, the more defensible. Generalism sounds adaptable but is actually fragile in an AI-disrupted labor market.
Yampolskiy distinguishes between AI users (most people) and AI-native operators — those who structurally integrate AI into everything they produce. AI-native operators ship 3-5x more output than non-users at higher quality. The gap is widening. 'The productivity difference between an AI-native professional and a non-AI professional today is already as large as the gap between someone who uses a computer and someone who doesn't. In 3 years it will be larger.'
THE PLAY
For the next 90 days, commit to using AI tools for everything possible in your work: drafting, research, analysis, editing, ideation. Learn the prompting patterns for your specific domain. The investment feels expensive early and pays off immediately. Within 6 months, you'll be measurably faster than colleagues who haven't made this shift — and the gap will only grow.
Yampolskiy predicts a bifurcation: commodity services get cheaper as AI does them, while human-delivered services become premium goods. The human therapist, the human coach, the human teacher, the human advisor — these become luxury offerings precisely because they are human. 'The scarcity shifts. Machine output gets free. Human contact gets expensive. Position yourself on the expensive side.'
THE PLAY
If your work involves teaching, advising, coaching, or guiding others, you're already positioned for the premium-human-service tier. Lean into what makes your service specifically human — presence, accountability, emotional attunement, shared experience. Don't try to compete with AI on information delivery. Compete on presence. That's what the market will pay for.
Yampolskiy points out that most people underestimate disruption because they can't feel what it's like from inside. Studying past transitions — the death of telegraph operators, the decline of manufacturing towns in the 70s, the disruption of print media — gives a visceral sense of how quickly entire career fields can collapse. 'Reading about the textile workers of Manchester in 1820 is the best preparation for 2028. The arguments against their concerns were exactly the arguments used today about AI. They were wrong then. They are wrong now.'
THE PLAY
Read one book about a major technology transition in the last 200 years — the Industrial Revolution, the electrification of factories, the computerization of offices. Pay attention to how people who were 'safe' got displaced and how quickly. This historical literacy makes AI warnings land differently than any amount of abstract speculation.
Yampolskiy's unexpected recommendation: even for white-collar professionals, invest in acquiring one meaningful physical-world skill — carpentry, welding, cooking, electrical work, agriculture. Not as a primary career, but as insurance. Physical skills in unstructured environments are the most AI-resistant category of human labor, and the cost of acquiring them while employed is low. 'The tradesperson is more future-proof than the knowledge worker. That's a reversal of the last 50 years of career advice.'
THE PLAY
Identify one physical skill you'd actually enjoy learning — something hands-on that requires in-person practice. Spend 2-3 hours per week on it over the next year. Even as a hobby, you're building a fallback capability that's genuinely valuable in labor markets increasingly dominated by AI for intellectual work. Best case, you enjoy it. Worst case, you have insurance.
YOUR ACTION PLAN
All the plays, back to back. Use this as your checklist.
The Five Jobs That Survive
Audit your role against the 5 surviving categories: physical work, original creativity, high-accountability leadership, human-connection work, AI specialization. If none fit, shift now while income is stable.
Build Skills That Compound Without You
Every year, build one asset that earns without your labor: book, course, business, investment, IP. Don't replace income — add in parallel. Assets survive labor disruption.
Specialize Narrower Than Seems Safe
Broad roles are most vulnerable. Specialize into narrower domains — specific industry, technical area, methodology. The deeper you go, the more defensible against AI commoditization.
Become an AI-Native Operator
For 90 days, use AI tools for everything in your work — drafting, research, analysis, editing. Learn the prompting patterns for your domain. AI-native operators outproduce non-users 3-5x.
Move Toward Human-Connection Premium
If your work involves teaching, advising, or coaching, lean into what's specifically human — presence, accountability, emotional attunement. Don't compete with AI on information. Compete on presence.
Study the History of Technology Transitions
Read one book on a past tech transition — Industrial Revolution, electrification, office computerization. Focus on how 'safe' people got displaced. Historical literacy makes AI warnings feel real.
Hedge With Physical-World Skills
Identify one physical skill to learn — carpentry, welding, cooking, electrical. 2-3 hours/week for a year. Physical work in unstructured environments is most AI-resistant. Hobby becomes insurance.
Ep. 001
8 principles from the author of The 48 Laws of Power on charisma, influence, and becoming someone people gravitate toward.
Read playbook →Newsletter
One email per drop. No spam. Unsubscribe anytime.
THE DIARY OF A CEO · EXTRACTED BY PODEX