Thinking, language, and human value in the age of AI

Executive overview

AGI will soon surpass human capability across most domains, including creative and analytical work. The risk is not just job displacement — it is that shortcuts like using AI for schoolwork will atrophy the cognitive muscles humans need to remain useful partners.

The answer is not to avoid AI but to use it as a thinking amplifier, not a thinking replacement. The core skill that survives AGI is the ability to generate original ideas, simulate the world from other people's perspectives, and create genuine value for others.

Why writing and language still matter most

• Large language models are powerful because of language — so fluency in language is now more critical, not less.
• Using AI to do writing homework is like driving a car for exercise: the task completes, the capability never develops.
• Kids who outsource writing lose the ability to reason logically and become dependent on whatever they are fed.
• Reading, writing, communication, and logic are how thinking develops — not side effects of education, but the point of it.
• The shift: schools used to train students to do homework. Now everyone needs to learn how to grade it — to evaluate, not just execute.

The problem with test-prep culture

• Math competitions originally trained mental flexibility: every new problem was practice for an unknown situation.
• A large test-prep industry now tries to eliminate surprise by drilling students on every possible question type.
• This removes the chance for students to invent — which is the only thing that actually trains original thinking.
• The correct use of competition math: as a source of unseen problems that force students to generate their own approaches.
• Goal of Po-Shen Loh's program: finish the curriculum fast enough that students no longer need any classes from anyone.

The ecosystem model for education

• Pain point 1: students need to learn to think, not just solve.
• Pain point 2: high-achieving math students often lack interpersonal skills and emotional direction.
• Pain point 3: drama and acting graduates want flexible part-time work that supports their craft.
• The solution combines all three: math-strong high schoolers teach middle schoolers, coached by professional actors/improvisers.
• Improv training builds communication skills; the teaching role builds EQ and real-world empathy.
• Win-win-win: every participant gets something they genuinely need — the model only works if all three do.
• Scale estimate: 100,000 high school tutors in the US teaching roughly 1 million middle school students.

Simulating the world as entrepreneurial superpower

• Entrepreneurship requires playing strategies forward in your head — imagining a product or decision and its downstream effects.
• AI used correctly accelerates this: not writing reports for you, but building the internal logic you use to understand a situation.
• You cannot identify real problems without interacting directly with people — customer discovery is unavoidable.
• Talking to thousands of parents and students in parks led directly to the product insight that became the current business.
• Rural students without phones who make their own games showed more curiosity and creativity than many well-resourced classrooms.

Why human value creation is the durable edge

• As AI takes over task execution, the only reason someone employs or collaborates with you is that you create value they want.
• You cannot create value without understanding people — their needs, constraints, and motivations.
• Empathy is not soft: it is the prerequisite for problem identification, which is prior to any solution.
• A philosophy of outcompeting everyone leads to permanent dissatisfaction; a philosophy of delighting others is both addictive and correlated with conventional success.
• Authentic motivation to help others is detectable — people choose partners who genuinely want to create value, and they avoid those who don't.

Critical thinking in a world of AI-generated narrative

• AI produces highly fluent, complete-sounding responses that can appear to cover a topic exhaustively while reflecting only one perspective.
• Bias is structural: the few AI providers that exist each embed a particular worldview, narrowing what feels like the full range of views.
• The remedy is deliberate exposure to opposing sources — seeking out the points of disagreement, not the averaged middle.
• Anyone speaking to you has an agenda; the ability to detect that and evaluate it independently is a survival skill.
• Losing the capacity to reason critically means becoming vulnerable to whoever controls the most convincing narrative.