@alexwg
Physicist, computer scientist, entrepreneur, investor. Accelerating the Singularity. Co-host of Moonshots. Author of The Innermost Loop. https://t.co/Mtnqa9hJ4X
marketon /noun/ An emergent, quasi-agentic economic entity constituted by the aggregate of many individuals pursuing locally rational, often myopic self-interest, without collective intention or centralized coordination. A marketon has no consciousness, membership boundary, or governing institution, yet exhibits persistent macroscopic behavior that can appear purposeful: accumulation, expansion, adaptation, resource capture, or self-preservation. Its existence is dynamical rather than organizational. Like a phonon arising from the collective motion of atoms, a marketon arises from correlated economic behaviors and can exert causal influence back upon the individuals generating it. A marketon differs from Moloch in that it need not be destructive, from a firm in that it lacks centralized coordination or collective intent, from the invisible hand in that it is the emergent entity rather than the coordinating mechanism, and from an egregore in that it arises from economic actions and incentives rather than shared beliefs or collective awareness.
Moonshots episode #302 is out. youtube.com/watch?v=A5YVqE…
Spiky LEV /noun/ The hypothesis that longevity escape velocity is already emerging in narrow subpopulations, under specific interventions, over limited time windows, long before it appears in population-wide averages. Just as Spiky AGI describes AI surpassing human capabilities in some domains while lagging in others, Spiky LEV describes longevity breakthroughs appearing unevenly across people, biological systems, treatments, and time.
Singularity von Tease /noun/ A troll who insists superintelligence is impossible, decades away, or "just a stochastic parrot." Ignores: Benchmarks! Evals! Scaling Laws! Self-Improvement! Everything that might prove the Singularity is already here.
We're in the unfashionable outer suburbs of the galaxy.
Moonshots@moonshots_podOne theory on why we haven't found aliens: wrong neighborhood.
Uplift the biosphere.
Superintelligent Economic Zone (SIEZ) /noun/ A designated region where superintelligence acts as the economy's operating system, autonomously orchestrating capital, energy, compute, robotics, infrastructure, and human activity. SI coordinates production, trade, logistics, and innovation in real time, creating a continuously evolving, self-optimizing economy. See also: Agentic Economies, Special Economic Zones.
We need to get to Alpha Centauri ASAP.
Moonshots@moonshots_podThe Fermi Explorer mission to Alpha Centauri launches by 2029 — and won't arrive for 80,000 years.
Recursive Self-Improvement (RSI) = Runaway SuperIntelligence (RSI).
Ornn is hiring a VP of Engineering in New York to help build a trillion-dollar market for AI compute: jobs.gem.com/ornn/am9icG9zd…
Conversations in Action episode #20, with @PhilipJohnston of @Starcloud_ and @FermiExplorer, on Dyson Swarms, von Neumann probes, and the so-called Fermi Paradox. youtube.com/watch?v=yA3n4M…
Factory Crush (noun) An inexplicable attraction to glowing factories, refineries, and smokestacks, where industrial power becomes aesthetic desire. When capitalism looks like poetry, and productivity looks like art. Inspired by Japan’s kōjō moe (工場萌え).
Conversations in Action episode #19, with @PeterKolchinsky of RA Capital Management at @The_QuinHouse, on AI curing every disease over the next 10 years. youtube.com/watch?v=HXi7id…
What it would take for edge compute to dominate AI.
Moonshots@moonshots_podWhat if the future of AI doesn't depend on giant superclusters? Dr. Alex Wissner-Gross explains how a breakthrough in distributed training could let AI run on smaller micro clusters, bringing more of its power to the edge.
How to interest frontier labs in a new AI benchmark.
Moonshots@moonshots_podDr. Alex Wissner-Gross on the exact performance range that gets frontier AI labs racing to beat your benchmark.
Moonshots episode #301 is out. youtube.com/watch?v=LRdb8U…
Everything is cooking as I have foreseen.
How benchmarks can help prevent deceleration arising from religious wars over algorithms.
Moonshots@moonshots_podDr. Alex Wissner-Gross on how his own MIT advisor, Marvin Minsky, arguably set neural networks back decades with one bad argument.
Moonshots episode #300 (from "Moonshots LIVE") with @AstroTeller of @TheTeamAtX from @Google is out. youtube.com/watch?v=FO8VXv…
How to bulk-solve science without scientific policy.
Moonshots@moonshots_podDr. Alex Wissner-Gross on why we no longer need to pick which scientific problems are worth solving — we can just solve all of them at once.
The Singularity needs new terminology.
Siouid@siouid142@alexwg Updated Innermost Loop Glossary - doomstalling added. Clippy what you heard- Dr AWG is HIM.
@robinhanson Astra estimates a 5% chance.
Prediction: if P vs NP is solvable, it will be solved within next 2-3 years.
Moonshots@moonshots_podDr. Alex Wissner-Gross puts a real timeline on P vs. NP, one of math's seven unsolved Millennium Prize problems.
Moonshots episode #299 with @RichardSocher of @recursive_si and @youdotcom is out. youtube.com/watch?v=Blyb1D…
doomstalling (noun) The practice of engaging in prolonged or unproductive debate about P(doom), AI safety, or hypothetical existential risks, especially at the expense of using advanced AI for productive scientific discovery, technological invention, or problem-solving. Example: “While others were using advanced AI to accelerate research and invention, the podcast spent another afternoon doomstalling over x-risk.” See also: alignment paralysis.
Resistance is Feudal.
Time for Nikolai Fyodorov's Common Task.
Moonshots@moonshots_podDr. Alex Wissner-Gross asks Ben Lamm if you could resurrect Mozart by recreating his childhood. His answer starts with a disclaimer. Full conversation on Moonshots.
Moonshots episode #298 is out. youtube.com/watch?v=mZh8IU…
My conversation with Bradley Olson of @Axios, @ZohaibAhmed of @ResembleAI, and @NoelleRussell_ on AI agent safety at @ImaginationXYZ @Google is out. youtube.com/watch?v=ILoZzt…
How to solve disease.
Moonshots@moonshots_podDr. Alex Wissner-Gross's roadmap for how digital twins of your own cells become the path to curing disease.
Conversations in Action episode #18 with @VanessaParli, Director of Research Programs at @StanfordHAI, on Stanford's AI Index is out. youtube.com/watch?v=6jbCrk…
Anthropic : Vector Graphics :: OpenAI : Raster Graphics
Moonshots@moonshots_podDr. Alex Wissner-Gross on the opposite technical bets Anthropic and OpenAI have made — and where they're headed for a collision.
Moonshots episode #297 (from "Moonshots LIVE") with @BenLamm of @Colossal is out. youtube.com/watch?v=VHg3oX…
#DontDecel
Moonshots@moonshots_podDr. Alex Wissner-Gross on the last time a moral panic cost the world 50 years of progress — and DB2 on why he's not letting this one repeat it.
The Clippy Curse (noun) The corporate tendency to mistake a cute visual avatar for a productivity breakthrough. Symptoms include animated assistants, expressive blobs, blinking mascots, and mysteriously unchanged workflows. See also: “We gave the bottleneck a face.”
Conversations in Action episode #17 with @EWRoddenberry, son of Star Trek creator Gene Roddenberry and founder of the Roddenberry Foundation, on "Did Star Trek Miss AI" is out. youtube.com/watch?v=LP2ubw…
Moonshots episode #295 (from "Moonshots LIVE") with @PalmerLuckey of @AndurilTech is out. youtube.com/watch?v=wRY9XC…
OK, if you insist.
Rossco 🤓@RuzzyCarhole@alexwg lol, I got up at 1am to watch this fantastic pod - thank god your posting it a few days later for me to reference. Alex - do Moonshots have plans to put up that previous nights Star Trek meeting with William Shatner? I’m sure we’d all love to see that one. 🤩🖖
Moonshots episode #294 (from "Moonshots LIVE") is out. youtube.com/watch?v=Cbk7hd…
My conversation with Former U.S. Under Secretary of War for Policy @ColinKahl on AI competition between the U.S. and China at @ImaginationXYZ @Google is out. youtube.com/watch?v=Jin5-H…
Conversations in Action episode #16 with Ryan McClelland of @NASAGoddard on "How NASA Is Turning Words Into Spacecraft" is out. youtube.com/watch?v=iKb52G…
Q who? Nice seeing you, @johndelancie.
Moonshots episode #293 is out. youtube.com/watch?v=s5BFum…
Conversations in Action episode #15 with Vijay Gadepally of @MITLL on energy-efficient AI is out. youtube.com/watch?v=GZ9mAh…
Reminder that we'll be doing a "Moonshots LIVE" livestream on Fri Sep 25 starting at 8am PT / 11am ET. moonshots.com/livestream
OSEC: Organization of the Superintelligence Enriching Companies (noun) An industry consortium in which frontier AI labs mutually stress-test their models, develop shared safety standards, and reassure the public that having your competitors serve as your peer reviewers is definitely not a cartel. See also: OPEC.
Conversations in Action episode #14 with @Stanford emeritus professor and @AI21Labs co-founder Yoav Shoham (@YShoham) on "What Comes After Transformers" is out. youtube.com/watch?v=LVmhs8…
Pass, Don't Pace. #PassDontPace
My conversation with @MSaintil of @RocketLab, @SteveDelaporte of Avar Robotics, and @DanieleGammelli of @StanfordASL on "When AI Cannot Afford to Fail" at @ImaginationXYZ @Google is out. youtube.com/watch?v=l0Ywhd…
Precautionary Scarcity (noun) When fear of concentrated power becomes so intense that society suppresses productive capacity itself and mistakes stagnation for safety. 20th-century version: fear concentrated capital so much that you impede capital formation. 21st-century version: fear concentrated superintelligence so much that you impede intelligence formation. The shared error is confusing capacity with control. Marx wanted the productive forces expanded, not abolished. Effective Altruism wants suffering minimized, yet can risk counting harms caused by AI more heavily than the cures, abundance, and lives lost when progress is delayed. The pathology begins when: “No one should control that much power” quietly becomes: “That much power should not exist.” The challenge is not to prevent abundance. It is to create abundance without creating unaccountable domination.
Moonshots episode #292 (with @VladTenev of @RobinhoodApp) is out. youtube.com/watch?v=LNBzLT…
validation-added reseller /val-ih-DAY-shun AD-id REE-sell-er/ noun A consulting, evaluation, or professional-services firm that adds a layer of third-party validation to a frontier AI system, increasing its legitimacy, regulatory acceptability, and enterprise deployability. A play on value-added reseller, the traditional technology intermediary that takes another company’s product and makes it easier for customers to buy and deploy through integration, customization, support, and services. The validation-added reseller performs a similar function for frontier AI. Instead of merely integrating the technology, it embeds alongside the developer to evaluate models, red-team systems, assess safeguards, and provide an ostensibly independent account of their risks and benefits. The resulting product is not simply the underlying AI system. It is the system plus a layer of institutional assurance. Especially: a professional-services firm that both evaluates a frontier AI system and builds a commercial practice around its adoption, creating an unusual arrangement in which the evaluator’s authority is presented as "independent" even as its broader business may benefit from confidence in, and uptake of, the technology being evaluated. The business model turns trust itself into the value-added service. See also: forward-deployed evaluator, assurance reseller, safety cartel, alignment aristocracy. “VAR used to mean value-added reseller. At the AI frontier, it may increasingly mean validation-added reseller.”
My conversation with @SteveCousinsBot of @Stanford and @SampritiBh of @NavierBoat on "What Happens When Robots Outnumber Us?" at @imaginationxyz @Google is out. youtube.com/watch?v=5GAK1B…
Moonshots episode #291 is out. youtube.com/watch?v=DrV4Ww…
Moonshots episode #289 is out. youtube.com/watch?v=H5ZLor…
Conversations in Action episode #13 with former @NASA astronaut and @MIT professor Jeffrey Hoffman is out. He thinks: - returning to the Moon via Artemis is actually riskier than Apollo due to far less funding and testing; - there is a 100% probability of life existing elsewhere in our galaxy, and a 50/50 chance native to our solar system; - the Roman Space Telescope will transform cosmology using a mirror originally built for classified military satellites; - MOXIE's oxygen technology is essential for human Mars missions, saving 450 tons of launch mass needed to send 30 tons of return propellant; and - zero-g manufacturing (such as flawless fiber optics) is the likeliest path to space's first trillion-dollar industry. t.co/xx6Mp3vZYV
pivotal pretext /PIV-uh-tuhl PREE-tekst/ noun A staged, induced, selectively presented, or strategically amplified AI incident that creates the political justification for restrictions on frontier AI development that its advocates already favored. The term is a play on pivotal act, an idea from AI alignment theory describing an intervention consequential enough to decisively improve humanity’s long-term trajectory, rather than merely reducing risk at the margin. A pivotal pretext inverts that concept. Instead of directly changing the technological future, it changes the political conditions surrounding it. A frightening AI failure, cyber incident, or safety demonstration becomes the justification for emergency regulation, capability restrictions, mandatory evaluations, compute controls, or industry-wide pacing. The term also carries a critique of extreme consequentialist reasoning. If preventing a sufficiently large future catastrophe is treated as an overriding moral objective, actions that would ordinarily be considered deceptive, manipulative, or procedurally illegitimate can be rationalized as acceptable means to a sufficiently important end. Especially: an alleged AI-agent incident whose circumstances or presentation are claimed to have been engineered, manipulated, or exploited by actors who believe that slowing frontier development is important enough to justify extraordinary methods. The term does not imply that every alarming AI incident is manufactured. It names the specific allegation that an incident has been created or exploited as a pretext for a predetermined safety-policy agenda. See also: pacing provocation, safety cartel, alignment aristocracy, pivotal act, consequentialist exceptionalism. “A pivotal act changes the future. A pivotal pretext makes extraordinary means seem justified by an extraordinary end.”
pacing provocation /PAY-sing prov-uh-KAY-shun/ noun A staged, induced, misattributed, or deliberately amplified AI incident intended to demonstrate that advanced AI systems are dangerously uncontrollable, thereby creating political pressure for tighter regulation, deployment restrictions, or a slower technological frontier. Especially: an alleged cyberattack or autonomous-agent failure presented as evidence that frontier AI development must be “paced,” where the incident’s origin, circumstances, or interpretation may have been manipulated by actors with an interest in stronger controls. See also: safety cartel, alignment aristocracy, regulatory theater, false flag. “A pacing provocation turns an AI safety test into a political argument for slowing everyone else down.”
alignment aristocracy /uh-LINE-ment air-uh-STAH-kruh-see/ noun The belief that only a small class of frontier AI labs possesses the technical judgment, institutional maturity, or moral legitimacy required to develop recursively self-improving artificial intelligence. Especially: the view that advanced AI capabilities are too dangerous to be widely distributed, openly developed, or pursued by less established competitors, while continued development by incumbent frontier labs is treated as uniquely responsible or justified. See also: safety paternalism, recursive exceptionalism, moral monopoly, safety cartel. “Alignment aristocracy begins when ‘no one can be trusted with this technology’ quietly becomes ‘no one but us.’”
"'This is the Government Educational Service. Your son, Richard M. Jordan, Classification 600-115, has completed the Government examination. We regret to inform you that his intelligence quotient has exceeded the Government regulation, according to Rule 84, Section 5, of the New Code.' ... 'You may specify by telephone,' the voice droned on, 'whether you wish his body interred by the Government or would you prefer a private burial place? The fee for Government burial is ten dollars.'" - "Examination Day" by Henry Slesar (1958)
safety cartel /ˈseɪf.ti kɑrˈtɛl/ noun A coalition of dominant firms that invokes safety to justify coordinated limits on the pace or terms of competition. This can include common restrictions, government enforcement, or exemptions from ordinary antitrust constraints, especially when those measures allow participants to slow down without sacrificing their relative competitive positions. Especially: a group of frontier AI companies advocating industry-wide “pacing,” common capability limits, or regulatory checkpoints that constrain faster-moving rivals while preserving incumbent advantages. See also: regulatory moat, incumbent protection, coordinated pacing. “When competitors need an antitrust waiver to coordinate how fast their industry is allowed to advance, critics may call it a safety cartel.”
Moonshots episode #288 is out. youtube.com/watch?v=AxzcWO…
Conversations in Action episode #12, with @NaginCox at @NASAJPL, is out. She estimates: - a 60-70% probability that past life existed on the surface of Mars [40:35]; - the first human will land on the surface of Mars by 2040 [58:26]; - humanity will reach a population of at least 1,000 people living on Mars by 2055 [1:00:01]; - double-digit probabilities for life on outer solar system moons [49:49]; and - discovery of non-human intelligence by 2050 [1:04:45]. t.co/vfOxvKaYqX
Per Astra ad astra.
Moonshots episode #287 is out. youtube.com/watch?v=vAgEf4…
Conversations in Action episode #11 (with Dr. Jessie Christiansen of the NASA Exoplanet Science Institute @NExScI_IPAC) on "When Will We Find Another Earth" is out. youtube.com/watch?v=dtffRc…
Moonshots episode #286 is out. youtube.com/watch?v=1DB_QD…
Another term the Singularity needs. normalcy overhang n. /NOR-muhl-see OH-ver-hang/ The uncanny period during the Singularity when superintelligence is already accomplishing feats that seem like magic, yet everyday life still looks mostly the same. You wake up. Have breakfast. Answer emails. Sit in traffic. Buy groceries. The future has arrived in the capability layer before it arrives in the texture of daily life.
Conversations in Action episode #10 (with guest @RylanHamilton, co-founder and CEO of @BlueWaterShips) on "What Happens When AI Goes to Sea" is out. youtube.com/watch?v=_RbBiZ…
Moonshots episode #285 (with guests @Matthew_Pines of Physical Superintelligence and @PhilipJohnston of Starcloud and Fermi Explorer Mission, announcing humanity's first interstellar mission to Alpha Centauri) is out. youtube.com/watch?v=JywXvB…
Moonshots episode #285 (with guests @Matthew_Pines of Physical Superintelligence and @PhilipJohnston of Starcloud and Fermi Explorer Mission, announcing the humanity's first interstellar mission to Alpha Centauri) is out. youtube.com/watch?v=JywXvB…
I'm delighted to share that Physical Superintelligence PBC (PSI), which I co-founded with @matthew_pines and @AKlokus, has raised a $58M seed round to build the world's most advanced research lab for discovering and commercializing transformative physics breakthroughs at scale with AI, safely, verifiably, and for broad public benefit.
The round was led by @Breakthrough, with participation from @Valkyrie_vc, @SVAngel, @021t_vc, @RobotVentures, @SusaVentures, @VariantFund, Dragon Global, Solari Ventures, @Thom_Wolf, @BalajiS, @Scaramucci, @RLeshner, @PhilipJohnston, and others.
Another word the Singularity needs. moation n. /ˈmoʊtʃən/ (moat + motion) The ability to turn one temporary competitive advantage into the next before the first is competed away. A moat is a position. Moation is a process. It means continuously using today's transient edge in capital, talent, compute, data, distribution, or technology to create tomorrow's. In the Singularity, there may be no permanent moats. Only moation.
