Tyler Cowen, an economist and public intellectual, discusses tariffs, academia’s structural flaws, AI’s transformative potential, and the importance of mentoring and networking in a rapidly changing world.
Tariffs and trade policy: Cowen argues that tariffs are economically inefficient, raise prices, hurt innovation, and lower living standards, with most costs borne by smaller trading partners like Canada.
He supports limited exceptions for national security (e.g., domestic ship or drone production) but opposes broad tariff policies, including those proposed by Trump in early 2025.
Historical examples like washing machine tariffs cost ~$815,000 per job created, with ripple effects increasing prices even on non-tariffed goods.
If implemented, tariffs would likely lead to a slightly lower stock market, higher inflation, and reduced real wages in the near term.
Canada as a 51st state?: Cowen dismisses the idea as unwanted by both Canadians and Americans, though he agrees Canada should meet NATO defense spending targets.
He speculates Trump’s rhetoric may be a tactic to pressure Canada into greater defense investment, especially in the Arctic, but warns this could backfire by fueling anti-American sentiment.
Academia’s strengths and weaknesses: Universities offer intellectual freedom, exposure to diverse thinkers, teaching as a tool for refining ideas, and mentorship opportunities.
However, incentives push researchers toward narrow specialization, peer approval, and risk aversion; graduate students face high rates of mental health issues; and bureaucratic burdens (grants, committees, admin) are growing rapidly.
Tenure doesn’t significantly increase risk aversion—selection effects matter more: academia attracts and trains risk-averse individuals over many years.
The grant system: Essential for large-scale empirical work (e.g., randomized controlled trials in development economics), but highly bureaucratic and time-consuming.
Researchers spend over a third of their time securing grants, skewing research toward fundable topics rather than the most valuable ones.
Cowen suggests simplifying applications and using AI to streamline refereeing, though institutional inertia slows adoption.
AI in research and publishing: Tools like OpenAI’s o1 Pro and Deep Research can produce high-quality literature reviews and referee reports comparable to human PhD-level assistants.
Cowen tested submitting an AI-generated referee report alongside his own; the journal ignored it, revealing systemic resistance to innovation.
He predicts AI will soon outperform humans in peer review, but bureaucratic adoption may take years.
Disagreement among intelligent people: Even with shared data and good intentions, smart people disagree due to self-deception, overconfidence, and circular reasoning about who counts as an epistemic peer.
Cowen co-authored a paper on this with Robin Hanson; they themselves disagreed on the paper’s implications.
Metarationality: The ability to recognize one’s limits and defer appropriately to others.
Best cultivated through activities with immediate, objective feedback (e.g., trading, chess), not abstract self-reflection.
Overconfidence in politics is common; true metarationality avoids paralysis by combining Bayesian updating with decisive action.
Stamina vs. grit: Grit is persistence on a single problem; stamina is the capacity to handle increasing workloads over decades without burnout.
Stamina is largely genetic but can be nurtured by surrounding oneself with high-stamina peers.
Academic advice flaws: Professors often advise students to pursue “solvable” problems, assuming they’ll complete their degrees—but over 50% of undergraduates and many PhD students drop out.
Cowen argues some should attempt ambitious projects early to quickly discover if academia suits them.
Interviewing style: Cowen prepares extensively (weeks to months) for podcast guests, reading their work deeply and crafting specific, challenging questions to elicit their best thinking.
He selects guests capable of rising to the occasion and avoids hostile questioning.
Worldview and truth: Cowen embraces a “patchwork” view of reality (influenced by philosopher Nancy Cartwright), where different domains (physics, economics, psychology) may not converge into a single unified theory.
He values studying grand worldviews (Plato, Hegel, Einstein) to sharpen thinking, even when disagreeing.
Free will and belief: Cowen is a determinist (“non-believer”) who rejects the label “atheist” due to its association with dogmatic anti-religion sentiment.
He assigns a 1-in-19 probability to the existence of God—up from 1-in-20—due to unexplained UAP (UFO) data, which he takes seriously after private conversations with credible insiders.
Emotional resilience: Cowen attributes his even-keeled temperament to genetics (low neuroticism), not learned behavior.
He distinguishes contextual disagreeableness (e.g., open to new ideas but hard to persuade) from general contrarianism.
Critique of Nassim Taleb: While respecting Taleb’s insights, Cowen finds some views “crazy,” particularly his claims about abnormal returns from out-of-the-money puts and his politically unconventional takes on Lebanon.
They’ve never formally debated; Cowen prefers written, refereed exchanges over public verbal debates for resolving technical disputes.
Ambition and motivation: Cowen sees ambition as often bundled with insecurity and desire for recognition—not inherently unhealthy.
He funds his own projects (blog, podcast) modestly, freeing himself from financial pressure.
Current projects: Writing a book on mentoring—how to mentor and be mentored—arguing it’s a uniquely human endeavor unlikely to be automated by AI.
Also preparing multiple podcast episodes, writing Bloomberg columns, and planning summer travel.
Advice for academics and students:
Use the best available AI tools—they’re worth the cost and transformative for learning.
Travel to places you think you’ll dislike; it broadens perspective.
Invest heavily in human networking: as AI amplifies individual impact, access to resources (funding, talent, mentors) becomes more critical.
Mentoring is undervalued and will grow in importance as AI reshapes work.
