04-10-2026Price:

The Frontier

Your signal. Your price.

AI & TECH

AI redesigns matter and life through first principles

Friday, April 10, 2026 · from 5 podcasts
Dwarkesh PodcastMoonshots with Peter Diam…Peter St Onge PodcastRadiolabNo Priors: Artificial Int…
  • AI bypasses flawed scientific data by building robotic labs to discover ground truth.
  • Companies like Colossal use de-extinction moonshots to engineer scalable biology platforms.
  • Chaos is biology's default state, forcing AI to handle permanent flux, not equilibrium.

Biology and materials science are shifting from observation to engineering. Liam Fedus told No Priors that AI’s bottleneck is the physical world. Trained on unreliable scientific papers where material properties vary by orders of magnitude, AI models reproduce human confusion. Fedus’s startup, Periodic Labs, addresses this by creating closed-loop systems where AI directs robotic labs to run experiments, generating its own high-fidelity data.

This transition makes AI not just a tool for prediction but a new type of scientific object. On the Dwarkesh Podcast, Michael Nielsen argued that models like AlphaFold are high-dimensional artifacts requiring “interpretability archaeology.” They contain embedded knowledge we must excavate, changing the scientist's role from theorist to excavator.

“Science ultimately isn't sitting in a room thinking really hard. You have to conduct experiments to interface with reality.”

- Liam Fedus, No Priors

Colossal applies this engineering mindset to biology. CEO Ben Lamm told Moonshots the woolly mammoth project is a stress test for a platform. The goal is a synthetic biology engine capable of spinning out companies tackling everything from plastic degradation with AI-designed microbes to artificial wombs. The AI leap is stark: where genetic editing was 40% efficient three years ago, Colossal now achieves 90% efficiency on hundreds of simultaneous edits.

This design-based approach confronts nature’s inherent disorder. Radiolab featured a 30-year experiment where a sealed barrel of seawater never reached ecological balance, its species populations booming and crashing chaotically. If there's no natural equilibrium to restore, the task for AI-driven conservation becomes managing preferred states of chaos.

“Actually, chaos is a system which is high predictability on the short run, but cannot be predicted in the long term.”

- Alisa Beninca, Radiolab

The shift is attracting a new kind of researcher. Fedus noted that physicists, bottlenecked after the Higgs boson discovery by slow, massive hardware projects, now dominate AI. They treat neural networks as physical systems to be measured. This principled thinking accelerates the move from digital simulation to atomic rearrangement, promising a materials revolution akin to the agricultural leap. The frontier is no longer just understanding nature, but remaking it.

AI InfrastructureBiologyModels

By the Numbers

  • 1887Year of Michelson-Morley experimentmetric
  • 1940Approximate year of muon decay experimentsmetric
  • 1838Year stellar parallax was measuredmetric
  • 260total Colossal scientistsmetric
  • 200Colossal scientists in the USmetric
  • 60Colossal scientists in Australiametric

Entities Mentioned

ApolloProduct
BlackstoneCompany
CoinbaseCompany
ColossalCompany
FOLDCompany
UAECompany
ViagenCompany

Source Intelligence

What each podcast actually said

Dwarkesh Podcast
Dwarkesh Podcast

Michael Nielsen – How science actually progresses • Apr 7

  • The theory of natural selection emerged independently in the 1850s because necessary building blocks like deep geological time (established by Lyell in the 1830s) and global biogeography from colonial voyages were finally in place.

Also from this episode:

Physics (4)
  • The Michelson-Morley experiment (1887) did not prove the ether nonexistent. It only falsified certain ether theories, like the existence of an ether wind. Michelson continued to believe in the ether until his death in the 1920s.
  • Nielsen argues that falsification in science is far more complicated than naive models suggest. The Michelson-Morley result didn't induce special relativity; it merely ruled out some ether models while others remained viable.
  • Lorentz derived the mathematical transformations that form the basis of special relativity before Einstein, but interpreted them as physical effects of moving through the ether. His theory was experimentally indistinguishable from Einstein's until later tests like muon decay experiments in the 1940s.
  • Poincaré understood key postulates of special relativity but clung to a dynamical explanation for length contraction, which Nielsen suggests shows how deep expertise can sometimes obstruct fundamental conceptual shifts.
Science (4)
  • Scientific communities can converge on a correct interpretation before definitive experimental proof arrives, as with the acceptance of heliocentrism centuries before stellar parallax was measured in 1838.
  • Nielsen argues that major bottlenecks in science occur where existing heuristics and processes no longer apply. Progress requires a diversity of research programs exploring many promising ideas, as with the different responses to anomalies in the orbits of Uranus (leading to Neptune) and Mercury (requiring general relativity).
  • Nielsen believes the technology and science tree is vastly larger than we realize, with different civilizations likely exploring different branches. This creates the potential for massive future gains from trade in ideas, not just resources.
  • The explosion of new fields like computer science shows diminishing returns arguments fail because unseen 'desserts' are constantly added to the buffet of knowledge, allowing fresh progress without mastering prior centuries of work.
AI & Tech (4)
  • Nielsen contends AlphaFold's success is primarily a story of decades of expensive experimental data acquisition (the Protein Data Bank), with AI modeling representing only a small fraction of the total investment.
  • Nielsen sees complex AI models like AlphaFold not as classic explanations but as new types of objects. They may contain embedded explanations that can be extracted through interpretability work, or enable novel operations like merging and distillation.
  • Nielsen argues new fundamental primitives keep being discovered within established frameworks, like public key cryptography and blockchain ideas emerging decades after the Church-Turing thesis defined computation.
  • Quantum computing emerged as a field in the 1980s because two historically contingent trends matured simultaneously: the salience of personal computing and the new ability to manipulate single quantum states with ion traps.
ModelsPhysicsHistoryPhilosophy
Moonshots with Peter Diamandis
Moonshots with Peter Diamandis

How AI Is Bringing Extinct Animals Back (And What Comes Next) | Ben Lamm (Colossal) | EP #245 • Apr 7

Also from this episode:

AI & Tech (4)
  • Colossal is an AI-powered synthetic biology platform focused on de-extinction and biodiversity. The same foundational platform used for the woolly mammoth is now spinning out companies targeting other massive biological problems.
  • Ben Lamm says without AI, Colossal could not operate. He believes every company should be an AI company, as AI is essential for designing and building new living products.
  • Colossal's team has grown to 260 scientists across the US and Australia. A significant portion of the team is dedicated to AI programming.
  • Lamm says Colossal's genetic editing capabilities are far ahead of competitors. The company now performs hundreds of precise genomic edits simultaneously at 90% efficiency, where two years ago it managed only a couple of edits at 40%.
Science (9)
  • Colossal's first spinout is Breaking, a company developing microbial solutions to degrade plastic. Lamm says its microbes break the chemical bonds of plastic instead of just creating smaller microplastics.
  • Publicly announced de-extinction projects include the woolly mammoth, Tasmanian tiger, dodo, and moa. Colossal also cloned dire wolves from a 73,000-year-old skull in 18 months.
  • Ben Lamm cites an EY market estimate for de-extinction. It valued the potential market for content and experiences related to extinct species at $1.7 trillion, based on global consumer spending patterns.
  • Lamm secured a nine-figure deal with the UAE to build the world's first biovault, a centralized repository for sequencing and preserving global biodiversity data. He calls it a nine-figure initiative for both the country and Colossal.
  • Colossal is developing artificial wombs across multiple animal clades to 'productionize' species development. Lamm argues this could save species like the northern white rhino more efficiently than current conservation spending.
  • Colossal holds advanced cloning capabilities through Viagen, a company it acquired. Viagen's cloning efficiency is 78%, vastly higher than the industry standard of 2%, and it cloned the only endangered species ever successfully reproduced.
  • The company is applying its platform to create disease-resistant plants and animals. A key project is developing chytrid fungus-resistant amphibians to combat the leading extinction driver for frogs.
  • Lamm identifies invasive species control via gene drives as a massive market. He cites a global economic impact of $5.4 trillion, with the U.S. impact alone exceeding $500 billion.
  • Work on artificial wombs has led to innovations in human IVF. Colossal built a hydrodynamically-focused microfluidics device that improves embryo health and could replace the archaic morphological grading system.
BiologyAI InfrastructureStartupsRegulationMacro
Peter St Onge Podcast
Peter St Onge Podcast

Ep 167 Weekly Roundup: 72 Scam Hospices in a Single Building • Apr 6

Also from this episode:

Energy (3)
  • Peter St Onge reports that global oil shortages are severely impacting Europe and Asia, with gas prices significantly higher and energy rationing beginning. He contrasts this with the US, which has doubled domestic oil and natural gas production since the 1970s, making it the world's largest exporter and somewhat insulated from the crisis.
  • Peter St Onge attributes Europe's high energy dependence, including 80% petroleum and near-total natural gas imports, to its 'net-zero' climate policies that shuttered coal and nuclear plants. Asian nations also face severe energy vulnerability due to geographic limitations, with China covering only 25% of its oil use and other major economies importing nearly 100%.
  • Peter St Onge describes how rapid energy shortages in Asia triggered widespread government interventions, including export bans, price caps, rationing, driving restrictions, and A/C limits. He warns that potential mandatory factory idling could lay off tens of millions across the region.
Business (3)
  • Peter St Onge reports US home sales crashed 20% in a single month, the worst since 2008, with sales at 587,000 units - half of pandemic levels and 150,000 below 2019 figures. Despite homes for sale rising 4.9% and prices falling 7% year-on-year, the market remains stalled.
  • Peter St Onge attributes housing market stagnation to the Federal Reserve's nearly $7 trillion COVID-era monetary expansion, which artificially drove mortgage rates to 2.6%. Now, with rates doubled to 6.4%, half of COVID-era mortgage holders are effectively trapped, unable to sell without facing double payments on an identical new house.
  • Peter St Onge warns of spreading cracks in the $1.8 trillion private credit industry, with major funds like Ares Management and Blue Owl limiting redemptions - an 'asset management equivalent of a bank run.' He attributes this to years of risky 'covenant light' loans and Fed rate hikes, impacting commercial real estate and leveraged buyouts, which could threaten pensions and insurance companies.
Adoption (2)
  • Peter St Onge reports Wall Street banks are lobbying Congress to ban interest on stablecoins, which they see as an existential threat to fractional reserve banking. Stablecoins, backed 1:1 with treasuries, offer over 4% interest with zero fees, significantly outperforming traditional banks that are 7-10% backed and pay minimal interest.
  • Peter St Onge warns that Republicans in Congress are pushing legislation, disguised as a housing bill, that would greenlight a wholesale Central Bank Digital Currency (CBDC) - issued only to banks - while banning a retail CBDC where the Fed issues directly to the public. He notes this happens despite overwhelming public opposition, with 80-95% of Americans opposing CBDCs once understood.
EnergyBankingMacroStablecoins
Radiolab
Radiolab

Life in a Barrel • Apr 3

  • Ecology professor Reinhard maintained a 100-liter barrel of brackish Baltic Sea water, initially from a two-week student experiment, for over six years.
  • After the Berlin Wall fell in 1989, Reinhard rediscovered the barrel, finding it thriving with diverse microorganisms like phytoplankton, zooplankton, and bacteria.
  • Reinhard's observation in the barrel challenged ecological theories that predicted ecosystems would stabilize or follow cyclical patterns in isolated conditions.
  • Over six years, Reinhard found the barrel's ecosystem to be completely chaotic, with species booming, crashing, and shifting dominance, never reaching a stable state.
  • Reinhard's work, co-authored with Alisa Beninca, was published in *Nature*, prompting skepticism from ecologists who questioned the purpose of restoration if nature is chaotic.
  • Hendrick Schubert, replicating Reinhard's experiment with eight barrels, found signs of chaos in some, but not all, vessels and compartments, indicating continued uncertainty.
  • In 1972, Stephen J. Gould, Tom Schopf, Dave Raup, and Dan Simberloff used computers to simulate evolution at random, finding results that mirrored the actual fossil record.
  • The simulation suggested extinction might be a random process, challenging Darwin's theory that fitness and natural selection are the sole drivers of survival.
  • Stephen J. Gould saw the computer simulation as a pivotal moment, elevating paleontology's status by posing a new, fundamental question about life's diversity and adaptation.
  • Matt Kielty notes that 99.9% of all species that have ever existed on Earth have gone extinct, suggesting that extinction is a near-universal fate.
  • The common 'primordial soup' theory of life's origin largely stems from Stanley Miller's 1952 experiment, which simulated early Earth conditions.
  • Stanley Miller's experiment, combining early atmosphere gases (ammonia, hydrogen, methane) with 'lightning,' produced amino acids, the building blocks of life.
  • Professor Nick Lane, an evolutionary biochemist, argues that forming a self-copying cell requires '10 or 12 more steps' beyond amino acids, which Miller's experiment did not explain.
  • Nobel Prize winner Francis Crick, co-discoverer of DNA, proposed 'directed panspermia,' suggesting alien civilizations seeded Earth with bacterial cells.
  • Organic molecules, including amino acids and components of DNA, have been found in space and on meteorites, suggesting a cosmic origin for some building blocks of life.
  • Nick Lane's preferred hypothesis for life's origin is deep-sea hydrothermal vents, which offer necessary chemicals, Earth's heat as energy, and a cell-like structure.
  • Hydrothermal vents, found 5-6 kilometers deep, form craggy structures up to 60 meters tall that mimic cells, facilitating the spontaneous formation of 'protocells.'

Also from this episode:

Culture (1)
  • Radiolab editor Soren believed the three featured stories independently explored the theme of chaos versus order in fundamental aspects of life.
Science (3)
  • Theoretical ecologist Alisa Beninca defines chaos not as randomness, but as high predictability in the short term, becoming unpredictable over the long term, like weather.
  • Matt Kielty reports that Stephen J. Gould, a renowned science writer and paleontologist, became fascinated with fossils after seeing a T-Rex at age four or five.
  • Paleontology was viewed more as 'stamp collecting' than a 'real science' capable of answering fundamental questions before Gould's contributions.
BiologyClimateHistory
No Priors: Artificial Intelligence | Technology | Startups
No Priors: Artificial Intelligence | Technology | Startups

AI for Atoms: How Periodic Labs is Revolutionizing Materials Engineering with Co-Founder Liam Fedus • Apr 3

  • Many AI researchers like Dario Amodei and Adam D'Angelo have physics backgrounds, a trend Liam Fettis attributes to physicists' principled thinking and high leverage in AI.
  • Periodic Labs is building an AI foundation lab for atoms, focusing on applying AI to material science, chemistry, and the physical world.
  • Periodic's AI system acts as an orchestration layer, using large language models to direct experiments and specialized neural nets designed for atomic systems.
  • Fettis says the acceleration of digital software engineering creates an imperative to connect AI systems to the physical world for scientific and technological progress.
  • Current AI technology, including improved reasoning and reliable tool use, is now sufficiently advanced to connect AI to the physical world, unlike in 2022.
  • Periodic leverages existing models for coding and language, spending zero effort on improving them, to focus its machine learning efforts on physical world frontiers.
  • A key data challenge in materials science is that reported property values from literature often span orders of magnitude, making ground truth difficult to establish without experiments.
  • Data generalization for physical systems is often domain-specific; a model trained on quantum mechanical objects doesn't help much with fluid dynamics.
  • Fettis believes AI systems exhibit odd spikiness in intelligence, being world-class in one domain but potentially poor in adjacent ones, challenging the idea of intelligence as a scalar.
  • Software engineering self-improvement by AI is happening now due to cheap, verifiable environments like unit tests, but this doesn't automatically translate to other domains like biology.
  • Fettis says AI research self-improvement is a slower outer loop than software engineering because experiments require GPUs and hours to evaluate model convergence and scaling properties.

Also from this episode:

Startups (4)
  • Periodic's approach relies on an interactive closed-loop system where experimental data feeds back to identify aberrations and patterns to drive the next experiments.
  • Fettis sees the most internal advances where there is an abundance of data in a specific chemical or material space.
  • The biggest capital cost for Periodic's work is GPU compute, not physical infrastructure, though lab setup has long lead times and calibration difficulties.
  • The multidisciplinary collaboration at Periodic, with physicists, chemists, AI researchers, and engineers, is allowing veteran scientists to see their fields fundamentally change.
Robotics (2)
  • While reliable robotics would be a huge accelerator, Periodic currently uses hybrid human-automation systems and off-the-shelf robotics to generate sufficient high-throughput data.
  • Fettis sees the interface of AI with the physical world via robotics as a transformative opportunity, given labor shortages and the vast number of people who work with the physical world.
Science (1)
  • Fettis views the future as one where AI generates matter, profoundly impacting semiconductors, aerospace, and energy by increasing the pace of physical world development.
ModelsStartupsRobotics

Related Stories