Rees, Martin

Astronomer Royal, cosmologist and astrophysicist at Cambridge, author of "If Science Is to Save Us" and co-author of "The End of Astronauts"

Martin Rees: Black Holes, Alien Life, Dark Matter, and the Big Bang

Lex Fridman Podcast · Jul 23, 2022

• Martin Rees
• dark matter
• theory of everything
• AI and scientific discovery
• Nobel Prize critique
• Big Bang and multiverse
• SETI and extraterrestrial intelligence
• post-biological civilizations
• human spaceflight vs. robotic exploration
• existential risk
• bioweapons and cyberattack risk
• global inequality
• social media and polarization

Core ideas

Rees opens on the limits of physical theory, arguing that even a complete "theory of everything" governing fundamental particles and forces would tell us little about the emergent complexity of stars, planets, or life, since new phenomena and concepts arise at each level of complexity that aren't usefully reduced to the underlying physics — a view he extends to AI, suggesting that machine learning systems may increasingly derive true, useful results (in mathematics or physics) that are essentially "unfamiliar territory for the human mind," accepted on the basis of demonstrated reliability rather than human comprehension, in the way a physicist might trust a result without fully following the proof. He connects this to a critique of the Nobel Prizes as poorly matched to how modern science actually gets done — large, collaborative, often unglamorous efforts (he cites gravitational wave detection and AI-driven discovery) that resist attribution to two or three individuals — a theme of his upcoming book "If Science Is to Save Us."

On cosmology, Rees describes the current uncertainty about dark matter (an unidentified particle or class of particles inferred from gravitational effects, with a wide range of candidate masses and properties still being searched for) and distinguishes between the well-established, testable aspects of the Big Bang (the universe's expansion and thermal history back to fractions of a second after the singularity) and highly speculative extensions like cosmic inflation and the multiverse, which he treats as serious but unconfirmed hypotheses rather than settled science. He discusses the search for extraterrestrial life and intelligence, noting his own involvement chairing a Yuri Milner-funded SETI effort, and argues that if intelligent life exists elsewhere, it may well have transitioned to non-biological, "electronic" or post-biological forms — civilizations that could be far stranger and more durable than biological life, potentially offering one candidate explanation (an "escape clause") for the Fermi paradox: intelligence without built-in aggression or self-destructive tendencies might persist quietly rather than expand aggressively or broadcast its presence.

A substantial portion of the conversation concerns human spaceflight, drawing on his book "The End of Astronauts" (co-written with Don Goldsmith), in which Rees argues that robotic and robotic-instrument technology has advanced so dramatically (citing the gap between 1990s-era probes like Cassini and modern smartphone-level electronics) that essentially all useful scientific and infrastructure work in space — exploring the outer planets, building structures on the Moon or Mars — can be done far more cheaply and safely by machines than by humans. He argues public funding should not be the vehicle for continued human space exploration, both because it is difficult to justify on practical grounds and because the public may not indefinitely tolerate the cost and risk; instead, he thinks human spaceflight should be left to private, risk-accepting adventurers and billionaires (analogizing to solo Atlantic aviators, high-altitude BASE jumpers, or free divers whom the public admires but does not expect to fund), with Mars colonization framed as a one-way pioneering effort for those willing to accept the risk, rather than a broad societal program. He voices measured praise for Elon Musk and SpaceX's engineering achievements (reusable rockets in particular) while separately noting his skepticism, in a broader technology-adoption argument, that any given technology curve (citing human spaceflight after Apollo, and supersonic aviation's rise and fall with Concorde) tends to plateau or reverse once initial novelty or geopolitical motivation fades, cautioning against assuming exponential progress continues indefinitely in any one domain.

Rees extends this note of technological ambivalence to AI and automation more broadly: he argues important decisions (medical, judicial, financial) should retain human oversight and an explanation a person can understand, even where machine judgment is statistically more reliable, and proposes taxing large firms that displace labor with automation to fund dignified, inherently human jobs (elder care, teaching assistants, park workers) — framing this as recovering value for people displaced from "mind-numbing" automatable jobs. Looking further out, he sees a serious chance that the human species, in its current pure biological form, will be succeeded within roughly a century by increasingly integrated, genetically modified, or cyborg descendants, a transition he thinks should be approached cautiously and with respect for existing human welfare, though he distinguishes this timeline from a starker warning about near-term existential risk: unlike historical eras, humanity now has the technological power to alter the entire planet (climate change, biodiversity loss) and, more urgently in his view, the vulnerability created by small groups or even individuals being able to trigger global catastrophe — engineered pandemics via gain-of-function-style biotechnology, and cyberattacks against increasingly automated critical infrastructure including nuclear command-and-control systems, a risk he says did not exist a few decades ago. He states plainly that the war in Ukraine sharpened his sense that the risk of nuclear escalation is higher than he'd previously assumed, given how easily miscalculation, false alarms, or cyber-induced confusion could cascade into catastrophe.

On global inequality and ethics, Rees pushes back on Steven Pinker's optimistic narrative of long-run human progress (rising life expectancy, falling infant mortality), arguing that while material conditions have improved, the gap between how the world is and how it could be — given available knowledge and resources — has widened enormously since the Middle Ages, and that this widening gap represents a genuine ethical regression rather than progress, citing global wealth concentration (citing a figure that the 2,000 richest people could double the income of the poorest billion) as an indictment of the current order, and calling for a large-scale, self-interested "Marshall Plan"-style investment in African development. He argues social media has accelerated political polarization by removing the moderating filter once provided by professional journalism, and — while acknowledging debate over whether platforms merely reveal rather than create division — expresses concern that this makes democratic societies harder to govern, adding that the personal and reputational toll of public life increasingly discourages capable people from entering politics. He closes with reflections on Newton (imagining what a single message about atoms, exoplanets, or spectroscopy might have meant to him), a defense of humility and curiosity as career advice ("pick a field where new things are happening"), and a personal, contented view of mortality, framed by his own closing words: that humans are unlikely to be the "culmination" of evolution, and that whatever entities eventually witness the sun's demise billions of years hence will be as different from us as we are from bacteria.

Excerpts