Modern WisdomHow Long Could Humanity Continue For? - Will MacAskill
CHAPTERS
- 0:00 – 3:33
Humanity at the beginning of history: how long we could last
Will frames the core longtermist intuition: compared to cosmic timescales, humanity is extremely early, and the future could be vast if we avoid self-inflicted extinction. He contrasts species-level “life expectancy” with scenarios where humanity survives on Earth for millions of years or expands beyond it for far longer.
- •James Webb imagery as a prompt to think on cosmic scales
- •Humanity’s potential longevity: millions of years on Earth; far longer if spacefaring
- •Expected value framing: small chance of stellar expansion implies enormous stakes
- •Self-created risks (AI, engineered pathogens) loom larger than natural ones
- 3:33 – 7:35
What longtermism means (and why Toby Ord’s ‘The Precipice’ matters)
Chris and Will define longtermism as taking future generations’ interests seriously, then pivot to existential risk as the key bottleneck. They connect Will’s work to Toby Ord’s and lay out the types of events that could permanently alter civilization’s trajectory.
- •Longtermism = moral weight on future people + practical focus on high-leverage risks
- •Trajectory-level events: engineered pandemics, AGI, world war
- •Natural risks exist but are increasingly mitigable (e.g., asteroid monitoring)
- •Modern tech creates unprecedented destructive power
- 7:35 – 9:54
Two levers on the long-term future: survival vs. making the future better
Will offers a clean framework: (1) ensure we have a future by reducing extinction/collapse risk and (2) improve the quality of that future. He gives concrete examples of “defensive” tech and policies that could drastically reduce catastrophic biological risk.
- •Long-term impact via: preventing extinction/collapse + improving future quality
- •Differential tech strategy: accelerate defense, slow/avoid offense
- •Far-UVC lighting as a potential pandemic-prevention breakthrough
- •Wastewater monitoring and upgraded PPE as scalable biosecurity tools
- 9:54 – 15:46
Why now may be unusually pivotal: interconnectedness and rapid tech progress
Will argues we live in an atypical window: global interconnectedness plus fast technological change creates both outsized opportunity and danger. He uses economic growth/compound growth reasoning to argue rapid growth can’t continue indefinitely, making this era unusually ‘dense’ with pivotal inventions.
- •Global connectivity increases scale of influence—and scale of domination risk
- •Future spacefaring could reintroduce disconnection across distances
- •Tech progress has shifted from near-zero growth to modern rapid growth
- •Compound growth implies physical limits; suggests future plateau/stabilization
- 15:46 – 23:50
The urn analogy, ‘trajectory change,’ and the risk of value lock-in
Using Bostrom’s ‘urn of balls’ analogy, they discuss how fast tech development increases the rate of drawing both beneficial and harmful ‘balls.’ Will then introduces ‘trajectory change’ as improving the long-run path by shaping values and preventing dystopian lock-ins.
- •Bostrom’s urn: technologies can be ‘green’ or ‘red’ (often both)
- •Trajectory change focuses on avoiding dystopias, not just avoiding extinction
- •Value lock-in: ideologies can seize power and persist via suppression
- •AI and global governance could enable unprecedented durability of bad regimes
- 23:50 – 34:33
‘Happy Birthday’ as a lock-in parable: plasticity, ossification, and cultural power
Will uses the global dominance of the ‘Happy Birthday’ song to illustrate how widespread adoption doesn’t imply optimality—and how early ‘moments of plasticity’ can harden into long-lasting norms. The discussion broadens into how culture enforces behavior without coercive states.
- •Bad equilibrium example: universal ‘Happy Birthday’ despite being musically awkward
- •Moments of plasticity vs. later ossification (hard-to-change conventions)
- •Cultural enforcement can be stronger than formal mandates
- •Norms/memes can spread globally without being ‘best’
- 34:33 – 37:24
Extinction risk in practice: engineered pandemics and how to drive risk down
They return to the first of three ‘suboptimal futures’ (extinction) and ask how plausible it is. Will gives rough probabilities for engineered pandemics and stresses that many countermeasures are available now if society chooses to invest in them.
- •Extinction is hard but non-zero; engineered pandemics are a central concern
- •Illustrative probabilities: ~0.5% extinction risk this century from engineered pandemics; higher risk of massive casualties
- •Mitigations: far-UVC, early detection via wastewater, high-grade PPE
- •Key idea: risk is contingent—society can choose to reduce it dramatically
- 37:24 – 42:49
Vulnerable World Hypothesis and the democracy vs. surveillance-state tension
Chris asks about Bostrom’s Vulnerable World Hypothesis: future tech could make catastrophe easy for individuals, seemingly requiring extreme surveillance. Will is more optimistic, arguing humanity often bans taboo tech and can build strong norms, while warning against sacrificing liberal democracy for marginal risk reduction.
- •Vulnerable World Hypothesis: some tech could grant ‘doomsday’ power broadly
- •Will’s skepticism: societies ban/limit tech (e.g., human cloning) and can set norms
- •Authoritarian safety ‘solutions’ may destroy the value we’re trying to protect
- •Prefer large risk reduction via non-authoritarian means when possible
- 42:49 – 48:25
Is Earth over-populated? People as both risk and resilience
They explore whether more people reduce extinction risk by diversifying survival chances, and whether climate/resource concerns imply we should have fewer children. Will argues additional people bring innovation and moral value, and that population size differences (8B vs 10B) likely matter less for x-risk than technology and governance.
- •Common worry: more people worsen climate/resource depletion; counterpoint: people also innovate and create value
- •Carbon impacts can be offset relatively cheaply compared to childrearing costs
- •Extinction risk comparison across centuries is ambiguous (new weapons vs more resilience)
- •Population may matter most through sustaining R&D capacity and avoiding stagnation
- 48:25 – 59:17
Technological stagnation as a hidden existential risk (and why ‘slow down everything’ fails)
Will explains why stagnation is dangerous: if civilization gets stuck with powerful offensive capabilities but weak defenses, risk accumulates over centuries. They discuss differential technological progress and why global degrowth/blanket tech-slowing is likely futile due to coordination and competitive dynamics.
- •Stagnation isn’t slower growth—it’s a stop (or reversal) in innovation
- •Risk accumulates if dangerous tech persists without improved defenses (e.g., bioweapons)
- •Need to pass through a ‘time of perils’ toward ‘existential security’ (Ord)
- •Differential tech progress: speed up defense, slow offensive capabilities; global degrowth is hard to coordinate
- 59:17 – 1:07:14
Climate change attention: valuable movement-building, but not the whole x-risk picture
Will praises climate activism for normalizing concern for future generations, arguing that such concern is historically rare and culturally contingent. Chris worries climate has ‘captured’ the entire existential risk conversation; Will responds that movement change takes decades and x-risk fields may be in an earlier stage analogous to climate in the mid-20th century.
- •Environmentalism as a major moral breakthrough: representing future generations
- •Climate concern has deep roots (Arrhenius 1896; growing consensus mid-century)
- •Risk prioritization: other x-risks (bio, AI) may be larger yet less salient
- •Movement-building is slow; AI/bio risk discourse may be at an earlier ‘climate-1950s’ phase
- 1:07:14 – 1:13:14
Extinction vs. collapse: would civilization recover after a global setback?
They distinguish extinction from near-total collapse and ask whether humanity would ‘bounce back’ after losing industrial capacity. Will reports his research conclusion: recovery is likely (>90%), citing historical resilience (Black Death, Hiroshima), the availability of stored knowledge, and remaining accessible fossil fuels—while noting fuel accessibility is one of the key constraints.
- •Collapse scenario: 99%+ die-off and reversion to pre-industrial tech; ‘unrecoverable’ collapse could be near as bad as extinction
- •Historical resilience: massive shocks rarely cause permanent collapse
- •Knowledge and imitation: libraries/records make reindustrialization easier than the first time
- •Key bottleneck: accessible fossil fuels; still likely enough to reach today’s tech level again (for now)
- 1:13:14 – 1:17:23
Coal mines, clean tech, and preserving resources for recovery scenarios
Will explains a less intuitive longtermist argument for clean energy: preserving fossil fuels could matter for reindustrialization after a catastrophe. They explore (and partially debunk) the idea of buying coal mines to shut them down due to regulatory and market substitution challenges, reinforcing that clean tech investment is the robust strategy.
- •Clean tech is ‘robustly good’ for health, climate, and resilience reasons
- •Longtermist twist: coal may be valuable for reindustrialization after collapse
- •Buying/shuttering coal mines faces legal ‘use-it-or-lose-it’ rules and substitution effects
- •Better leverage: accelerate clean alternatives (solar, nuclear, geothermal)
- 1:17:23 – 1:22:02
Civilizational backups: sealed refuges, seed banks, and ‘air-gapped’ humanity
Chris proposes an ‘air-gapped’ backup civilization; Will strongly endorses the idea as practical insurance against worst-case pandemics or other global catastrophes. They discuss design considerations—population size, rotating residency, scientific capacity, and stockpiles for rebuilding—arguing it’s reasonable given global resources and stakes.
- •Concept: hermetically sealed refuge(s) to survive doomsday scenarios
- •Include key skills (scientists/engineers) plus supplies for years and for rebuilding
- •Rotating vs. permanent residency models; multiple asynchronous sites
- •High expected value: tiny fraction of world GDP could preserve civilization’s continuity
- 1:22:02 – 1:31:42
What should humanity optimize for? Optionality, the ‘long reflection,’ and speeding up wisdom
Will argues the goal isn’t a fixed blueprint; humanity should aim for existential safety plus open options so future people can decide wisely. He introduces the ‘morally exploratory society’ and Bostrom’s ‘long reflection’—delaying irreversible lock-in (e.g., world government, space settlement) until moral understanding matures—and suggests investing far more in moral reasoning capacity.
- •Primary near-term aim: reach a safe, stable state with many futures open
- •‘Morally exploratory society’ and the ‘long reflection’ before irreversible commitments
- •Technological progress has incentives; moral progress does not—so it needs deliberate investment
- •Possible accelerants: value diversity, institutional experimentation, more humanities/moral philosophy capacity
- 1:31:42 – 1:34:19
How to help: donations, careers, and where to find Will’s work
Will closes with actionable paths: read his book, donate effectively, and choose high-impact careers aligned with safeguarding the long-term future. He points listeners to Giving What We Can and 80,000 Hours as on-ramps for practical engagement.
- •Read: *What We Owe The Future*
- •Donate via Giving What We Can toward high-impact interventions (biosecurity, AI safety, war prevention)
- •Career focus via 80,000 Hours (research, policy, communication, technical work)
- •Longtermist framing: small actions can affect civilization’s trajectory for millennia
