Return-Path: Delivered-To: alvarochf@traceseis.com Received: from cs510.bluehost.com by cs510.bluehost.com with LMTP id gOsVECLJ32lj+gEAWiBQxw (envelope-from ) for ; Wed, 15 Apr 2026 11:21:38 -0600 Return-path: Envelope-to: alvarochf@traceseis.com Delivery-date: Wed, 15 Apr 2026 11:21:38 -0600 Received: from v556.v5375b7fa.use4.send.mailgun.net ([159.112.244.56]:50788) by cs510.bluehost.com with esmtps (TLS1.2) tls TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 (Exim 4.98.2) (envelope-from ) id 1wD3vk-00000000YM4-3OlQ for alvarochf@traceseis.com; Wed, 15 Apr 2026 11:21:38 -0600 DKIM-Signature: a=rsa-sha256; v=1; c=relaxed/relaxed; d=mg2.substack.com; q=dns/txt; s=mailo; t=1776273680; x=1776280880; h=List-Unsubscribe-Post: List-Unsubscribe: List-Post: List-Id: List-Archive: List-Owner: Reply-To: In-Reply-To: References: Sender: Sender: Date: Message-Id: To: To: From: From: Subject: Subject: Content-Type: Mime-Version; bh=IR2y6EVwMpd2cQIE+LkBFrWRGrwES+S41IZ7LhY8MkM=; b=hY8MNAM49hlAYT21P8dQ0UEVW/KAqt9wlG5rQxly1n3RIuPyAWi8QtyCkDbKNjZ/Ye/uaaVL5/w3d5Yj1yFwFz+RPQlPsXbbDJifgkMk50SnIG0ejBWMQs6YnaYz7qPk/H492O7yp1QemmyxHYUsXCgx9+K6GhrmHbAByQeyQh4= X-Mailgun-Sid: WyI3MDdlYyIsImFsdmFyb2NoZkB0cmFjZXNlaXMuY29tIiwiNGQxODU4Il0= Received: by 26b11f4b8191dcc21056faf62aec7d1d02d603f507af344723210188b5e4dc4e with HTTP id 69dfc90fb35d0202afb54d5c; Wed, 15 Apr 2026 17:21:19 GMT X-Mailgun-Sending-Ip: 159.112.244.56 X-Mailgun-Batch-Id: 69dfc90f114670a6a3bf0eb2 Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="a5c2ad23665ebd8657b8089bca91d63d4c4b21bf3aaef79b9915a6738176" Subject: Can Old Ideas Survive the AI Age? From: Cosmos Institute To: alvarochf@traceseis.com X-Mailgun-Tag: post X-Mailgun-Track-Clicks: false Message-Id: <20260415162052.3.6877936a27ffa4e5@mg2.substack.com> Date: Wed, 15 Apr 2026 16:20:52 +0000 Feedback-Id: post-194306299:cat-post:pub-2225794:substack Sender: Cosmos Institute References: In-Reply-To: Reply-To: Cosmos Institute List-Owner: List-Url: List-Archive: List-Id: List-Post: List-Unsubscribe: List-Unsubscribe-Post: List-Unsubscribe=One-Click X-Mailgun-Variables: {"category": "post", "email_generated_at": "1776273678316", "is_freemail": "true", "post_audience": "everyone", "post_id": "194306299", "post_type": "newsletter", "pub_community_enabled": "true", "publication_id": "2225794", "subdomain": "cosmosinstitute", "user_id": "347258985"} X-Spam-Status: No, score=-1.5 X-Spam-Score: -14 X-Spam-Bar: - X-Spam-Flag: NO --a5c2ad23665ebd8657b8089bca91d63d4c4b21bf3aaef79b9915a6738176 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable View this post on the web at https://blog.cosmos-institute.org/p/can-old-id= eas-survive-the-ai-age Last week, we marked 20,000 subscribers by opening the floor. Your question= s were often genuinely challenging. Thank you to everyone who took the time= =2E So you don=E2=80=99t have to trawl through the comments, we=E2=80=99ve comp= iled all the responses in one place. Apologies to anyone whose question did= n=E2=80=99t make the cut. There were a few I=E2=80=99m still chewing over a= nd we=E2=80=99ll revisit many of these topics at greater length in future e= ssays. Let us know in the comments if this is a format you=E2=80=99d like u= s to repeat. Anna Lisa asks: Are there specific non-academic experiences or =E2=80=9Ccontainers=E2=80=9D= of formation that you think hold a lot of promise? (e.g. ones that were me= aningful to you or ones that you seek out for yourself/your children?) As a dad, I love this question. A lot of the important formation happens in places that do not look educati= onal at all, and are not primarily about instruction. They are about habitu= ation, responsibility, emulation, and contact with reality. The first was the household I grew up in. My mother was a Catholic conserva= tive historian from a military family who taught special needs kids for 36 = years. My father was a left-leaning physicist turned environmental lawyer, = and a pacifist. They agreed on almost nothing politically. But they agreed = on something deeper: that you should care about something beyond yourself a= nd that how you act matters more than what you know (and that knowing is bo= und up in doing!). That gave me, without anyone naming it, a kind of virtue= culture, and I think it=E2=80=99s the kind of thing that=E2=80=99s very ha= rd to manufacture deliberately but very obvious in its absence. For childre= n, what seems to matter is not ideological uniformity or even ideological s= ophistication, but a home in which seriousness, duty, and moral aspiration = are normal. The second was the submarine. I spent 610 days underwater, including under = ice. In a steel tube, you do not get to opt out of reality. You can=E2=80= =99t leave, and your mistakes could get someone killed. That kind of enviro= nment forms you because it imposes standards that are not negotiable. It te= aches service, competence, and mutual reliance in a way that is hard to sim= ulate. I think containers of formation are often places with real stakes, s= hared discipline, and demands that do not bend to your preferences. The third, and probably deepest, has been fatherhood. I had two kids and so= ld two companies in close succession, and while both changed my life, the c= hildren changed it more. I remember sitting in the corner of my room one ni= ght after putting Arden and Pierce down and asking myself whether I could w= rite what I believed on a single sheet of paper. I couldn=E2=80=99t. That w= as the moment I started reading seriously, beginning with the ancients, who= thought about these questions deeply and genuinely (my original list is he= re [ https://substack.com/redirect/14763042-057e-4ba5-977c-ee87cb33c866?j= =3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ]). Havi= ng children makes the question of what you=E2=80=99re actually =E2=80=9Cfor= =E2=80=9D impossible to defer. I now look for opportunities in my own children=E2=80=99s lives for contain= ers that place them in contact with reality, responsibility, and admirable = adults. The hard part is that the best formation is often a byproduct rathe= r than something you can engineer directly. You can build the conditions fo= r it, but you usually cannot force it. Turner Halle asks: You argue that philosopher-builders need explicit moral commitments to avoi= d optimizing for the wrong things. But your three pillars (truth-seeking, a= utonomy, decentralization) are themselves a normative framework that not ev= eryone shares. China=E2=80=99s AI strategy is still coherent, explicit, and= philosophical, it just starts from different premises. So how do you argue= for your philosophy without just replacing one set of defaults with anothe= r? What makes Cosmos=E2=80=99s values the right foundation rather than just= a well-packaged preference? Hi Turner, you=E2=80=99re right that truth-seeking, autonomy, and decentral= ization are substantive commitments. I think they matter less as one moral = doctrine than as conditions that keep moral life from collapsing into force= or drift. If you consider moral frameworks from Confucianism to Christianity to Marxi= sm, for them to have legitimate force over a person, that person has to be = able to genuinely endorse it (otherwise, you don=E2=80=99t have a moral com= mitment). That endorsement depends on human autonomy =E2=80=93 which is to = say, the capacity to reflect, evaluate, and take something on as your own r= ather than merely inheriting it or obeying it. So autonomy is not just one = preference among others. It is the deep substratum that makes moral commitm= ent possible at all. Take utilitarianism: Jeremy Bentham devised a system that, in my mind, diss= olves individual judgment into aggregate utility. This is in conflict with = autonomy-as-an-end. And yet building this system was itself a radical exerc= ise of autonomous reason. Every person who adopts utilitarianism is exercis= ing the same capacity. You can=E2=80=99t be a utilitarian in any meaningful= sense unless you=E2=80=99ve freely taken it on. So even a framework that s= ubordinates individual judgment to aggregate welfare requires individual ju= dgment to get off the ground. Now, someone could say that only makes autono= my instrumentally necessary, not foundationally important. I think that vie= w is unstable, because the goods autonomy is supposedly serving only become= moral goods for a person if they can in some real sense take them on as th= eir own. There are hard cases. In the Ash=E2=80=99ari tradition in Islamic theology,= divine command constitutes moral value rather than being something reason = independently discovers and then endorses. That=E2=80=99s a genuine challen= ge to autonomy as foundational. But even there, the person who freely choos= es submission is doing something categorically different from the person wh= o never had the choice. And a secular collectivist can make a parallel argu= ment: that harmony or collective flourishing is the true precondition, beca= use no individual life goes well outside a stable social order. I think tha= t is partly right. But unless people can participate in judging the terms o= f that order, harmony becomes coordination imposed on them rather than a go= od they share in shaping. Truth-seeking has a similar status. Any framework worthy of allegiance has = to remain in contact with reality. It has to be open, at least in principle= , to its own refutation. If someone could show me that decentralization pro= duces a worse outcome in a domain I care about, I=E2=80=99d have to take th= at seriously and I would. Systems that suppress truth-seeking can be intern= ally coherent in the way that closed systems are coherent. But they can=E2= =80=99t tolerate the mechanisms that would let them find out they are mista= ken. That=E2=80=99s a serious defect, especially if you think we=E2=80=99re= all operating under real uncertainty about what AI is going to do to human= life. And decentralization follows from the same logic at the institutional level= =2E If no person or committee is wise e= nough to determine the good for everyo= ne, then we should be wary of a small number of actors hard-coding their an= thropology into the substrate of society. Decentralization is valuable beca= use it preserves room for people to try things, get them wrong, and leave, = keeping mistakes from becoming total. So to your question about China: yes, their AI strategy is coherent, explic= it, and philosophical. But coherence purchased by foreclosing the capacity = for self-correction is brittle in exactly the way that matters most right n= ow, though that is still a bet, not a proof. I=E2=80=99m also skeptical tha= t it preserves the standing of human beings as agents capable of judgment r= ather than increasingly treating them as objects of coordination. A framewo= rk can be philosophically serious and still be wrong about what a person is= =2E So I would not say Cosmos is advancing a one true final doctrine that every= civilization must affirm. That would be too strong, and it would collapse = into exactly the kind of totalizing move you are warning against. I would s= ay instead that Cosmos is trying to defend the conditions under which free = people can genuinely seek truth, make judgments, form commitments, and buil= d different kinds of lives together, without having orthodoxy imposed on th= em by default. Salvador Duarte asks: Will the Cosmos Grants ever open again? We=E2=80=99re planning to re-open these in the next 60 days. We=E2=80=99ve = just had the demos from our latest batches of winners. Stay tuned for updat= es!=20 Bert Clements asks: Assuming frontier large language models, together with their multimodal and= agentic extensions, are trained to effective saturation on an exhaustive c= orpus that represents the totality of digitized human knowledge including a= ll scientific publications, books, patents, archival records, cultural arti= facts, and recorded conversations, will these systems be capable of transce= nding the statistical manifold of their training distribution to autonomous= ly discover, validate, and iteratively expand novel knowledge beyond the cu= rrent human frontier? I=E2=80=99m not sure scientific knowledge is a kind of territory where data= defines a bounded region that a model may or may not be able to venture be= yond. The affirmative view of this picture strikes me as broadly empiricist= insofar as knowledge comes from data, and scientists make discoveries by e= xtrapolating beyond what they already know. Your specific examples, though,= are actually the strongest case for the affirmative: theorem provers, simu= lators, multi-agent workflows, and verifiable rewards are exactly the kinds= of feedback-rich settings where I would expect systems to extend the front= ier. But that is not the only picture of science, and I do not think it is the d= eepest one. Science also advances by reorganizing what researchers take to = be meaningful in the first place: which anomalies matter, which questions a= re worth asking, and which explanations count as illuminating rather than m= erely predictive. Systems can already generate novel candidate hypotheses, and in domains wit= h strong automated verification, they may well extend the frontier. Formal = mathematics looks especially promising, because conjecture can be paired wi= th proof or disproof inside a relatively crisp evaluative architecture. In = such cases, I expect AI systems to produce results that are genuinely new t= o humanity. Just this week a constellation of agents improved a math proble= m that=E2=80=99s been open since Newton (Kissing Number in dimension 11: 59= 3 =E2=86=92 604). That is impressive. It is also, I think, a good example o= f the distinction I=E2=80=99m drawing: a real extension of an existing line= of inquiry, but still closer to powerful normal science than to scientific= revolution. But that does not settle the larger question. There is a difference between= producing novelty within an existing framework and generating a new framew= ork altogether. A system may help prove a theorem, optimize a search, or id= entify that drug X affects disease Y, all without altering our understandin= g of why the problem is structured as it is. That is a real scientific cont= ribution, but it is not reorganizing the conceptual landscape. The harder question is whether these systems can exercise scientific judgme= nt in the richer sense: whether they can tell which anomalies are significa= nt, which inconsistencies are fertile, which explanations deepen understand= ing rather than merely extend prediction, and which questions are worth reo= rganizing inquiry around. That is a higher bar than novelty, and I am not y= et convinced we know how to evaluate it well. Part of what makes this hard = is that frameworks are underdetermined by data. The same body of results ca= n often support multiple lines of inquiry, and judgment is what tells you w= hich one is worth building a field around. That remains, to my mind, the de= eper open question. tappert asks: Most of the current work on =E2=80=98AI, collective epistemic structures an= d decision-making=E2=80=99 focuses on filling gaps: more participants, fast= er information exchange, more efficient decision-making. This will help wit= h many problems, but certainly not with the most complex ones, because it j= ust accelerates the practical execution of the same thought styles that led= to the problems. Therefore: How can we use future AI to foster new thought= styles that are currently not supported by our existing social structures? Yes, I think the intuition that better collective decisions will emerge if = we simply gather more data from more people more efficiently breaks down at= the limit. That can improve performance within an existing paradigm, but i= t does much less when the paradigm itself is the problem. What groups develop over time are not just bodies of knowledge, but epistem= ic constitutions: implicit rules about what counts as evidence, which quest= ions are legitimate, who gets to propose, who gets to criticize, and on wha= t terms. Mill saw part of this in his account of the tyranny of prevailing = opinion and the epistemic importance of dissent. But the problem runs deepe= r than opinion alone. Entire institutions decide in advance what counts as = serious thought. So one promising use of AI would be to make those constitutions more visibl= e. A good system might show a research community, an organization, or a pol= ity where its methods systematically exclude certain questions, place some = assumptions beyond criticism, or discount certain voices before the argumen= t even begins. In medicine, for example, it might reveal a field that privi= leges what is easily measurable while sidelining patient testimony or long-= horizon effects that do not fit the dominant method. But diagnosis is only the beginning. I like this direction because the prob= lem is often not that new thought styles do not exist. It is that they rema= in stranded at the margins because the reigning structures of legitimacy su= ppress them. And sometimes the deeper problem is that the social conditions= required for a new thought style have not yet been built. New thought styl= es need protected spaces, alternative standards, and enough provisional leg= itimacy to develop before the dominant paradigm dismisses them. In that cas= e, the most useful contribution AI could make to collective epistemics is n= ot novelty on demand, but widening the space in which criticism, recombinat= ion, and intellectual minority formation can occur. Thomas Yiu asks: What is your definition of intelligence? When AI reaches ASI in the future,= do you think it will be safe and aligned? As a species, what is our purpos= e after ASI world? How can thrive as a species? I=E2=80=99d resist the standard definition of intelligence as raw problem-s= olving horsepower. For me, intelligence is the capacity to learn from reali= ty, inquire into it well, and let it correct you. Part of what I like about Fran=C3=A7ois Chollet=E2=80=99s work [ https://su= bstack.com/redirect/2a4820bb-b654-4c20-b140-63add99fc0ef?j=3DeyJ1IjoiNXFxeX= F4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ], and why ARC Prize [ ht= tps://substack.com/redirect/7e894c63-53b1-41bf-85f9-89987f7a7176?j=3DeyJ1Ij= oiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] has mattered, i= s the insistence that intelligence is not the same thing as accumulated ski= ll. A system can look impressive because it has absorbed an enormous amount= , or because the task has been made easy for it. The more interesting quest= ion is how much it can learn from limited experience, under real constraint= s, and still generalize well. But I do not think that is enough on its own. Leslie Valiant=E2=80=99s idea= of educability gets closer to the human picture (see his Cosmos Lecture fr= om last year here [ https://substack.com/redirect/0f43e68c-e0a0-4a93-ba67-4= 3eabb7b3dbf?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-= HyoA ]). Human intelligence includes the capacity to learn from experience,= receive instruction, integrate both, and apply them in new circumstances. = What distinguishes the human mind is not only that it learns, but that it c= an be taught and formed. And I would add one more layer. Drawing on HAI Lab director Philipp Koralus= [ https://substack.com/redirect/b6572fce-d9de-468d-8211-6fa54fc441ec?j=3De= yJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ], I think r= easoning is fundamentally question-directed. Minds are shaped by the questi= ons they pursue. They go wrong through shallow questions, premature closure= , and a failure to inquire far enough, just as much as through false conclu= sions. That matters for AI because a system can become very good at answeri= ng questions while still narrowing the range of questions humans ask, or re= warding closure where inquiry ought to stay open. That is why I=E2=80=99m less interested in arguing about whether AI will co= unt as =E2=80=9Csuperintelligent=E2=80=9D than in asking what it does to hu= man intelligence. A system can be extraordinarily capable and still erode o= ur capacity for inquiry, judgment, and self-government. That is the danger = I worry about most. On whether ASI will be safe and aligned: I do not assume that can be taken = for granted. I would trust highly capable systems only to the extent that t= hey remain corrigible, contestable, and embedded in institutions that prese= rve human judgment rather than replacing it. The problem is not just gettin= g the objective right once. It is making sure people can still question, re= vise, and refuse the system=E2=80=99s guidance when it matters most. As for human purpose after ASI, I do not think our purpose changes. If anyt= hing, it comes into clearer view. We are not here to compete with machines = at speed or scale. We are here to exercise judgment, form character, build = institutions, love particular people, and deliberate about the good. That l= ast point matters more than it may sound. Love is not interchangeable, and = responsibility is not abstract. A more powerful machine does not make our o= bligations to particular human beings less central. In a world of highly ca= pable AI, those things become even more important. Todd Enkhbat asks: Is it possible to carry on our learning from humanity up until now and jump= start a new society with the help of AI, assuming that we can concentrate a= nd utilize all the data we accumulated up until now? At what point does the= need for a new constitution or a new world order arise and how do we know = it? In short, no. Firstly, I don=E2=80=99t think =E2=80=9Call the data we accumulated up unti= l now=E2=80=9D is the same thing as the total weight of human knowledge. Mu= ch of the knowledge that keeps a society functioning is tacit, dispersed, a= nd unwritten. Some of it lives in practiced judgment: an ICU nurse sensing = that a patient is about to crash before the monitor shows it. Some of it li= ves in inherited forms: the habits of trust, restraint, and association on = which a free society depends, even when no one can fully specify them. As M= ichael Polanyi put it, we know more than we can tell. More importantly, I=E2=80=99d push back on the idea that we can jumpstart a= society at all. Societies aren=E2=80=99t machines that you design to a blu= eprint. Tocqueville saw this in the institutions of local self-government. = Hayek saw it in the way social orders carry dispersed knowledge that no pla= nner can gather in full. A free society is learned in practice through thin= gs like townships, juries, churches, and associations. Those are the ordina= ry disciplines by which people become capable of governing themselves. The question is whether our institutions can still sustain a free people ca= pable of self-government under new technological conditions. And that does = leave open the question you raise about constitutional inadequacy: how do w= e know when inherited arrangements are no longer enough? I do not think the= re is a clean threshold. Usually the signs are visible first in practice, w= hen institutions that once formed judgment begin producing passivity, depen= dence, or elite insulation instead. When they cannot, the answer is not a tabula rasa redesign of =E2=80=9Cthe = new world order.=E2=80=9D I would look to renewal through institution-build= ing, and Benjamin Franklin is the example I keep returning to. He took an E= nlightenment conviction =E2=80=94 that access to knowledge should not remai= n under the custody of church, state, or a narrow elite =E2=80=94 and embod= ied it in an institution. The subscription library made a philosophy of fre= edom socially real. That is why Franklin still matters to me here. He shows= what it looks like to translate a philosophy into civic machinery. We need= the AI-age equivalent: institutions that widen access to knowledge and jud= gment without concentrating them in a few hands. We need philosopher-builde= rs [ https://substack.com/redirect/cd1dae35-e87b-4470-babe-8402ee77e925?j= =3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] in tha= t spirit again. Miss Zanarkand asks: How can we motivate our children to learn at school? Should we try to motiv= ate them or find rather a way out of the system? (e.g. reading more classic= al books, rather than encouraging them to read what school nowadays gives?) Young people have a natural longing to be seized by something greater than = themselves. To be captivated. The promise of liberal education, going back = to the Greeks, is that there are magnificent ways of living, and magnificen= t questions about how to live, and that encountering them through great min= ds and great books can awaken a desire that organizes everything else. The disaster of modern education is that it has taught young people their l= onging is naive. That no book is really better than another, that no life i= s really higher than another, and that the hunger to be drawn upward by som= ething extraordinary is itself a kind of error. So I would say motivation is the right place to focus, but we should be pre= cise about what we mean. There is a kind of motivation that is intrinsic: t= he eros I just described, the desire to encounter greatness because it call= s to something real inside you. And there is extrinsic motivation: incentiv= es, structure, well-designed systems that make it easier to do the work. Bo= th matter. The best schools I=E2=80=99ve seen, including Alpha where my kid= s go, are serious about the extrinsic architecture. They=E2=80=99ve built a= n environment where children actually want to show up and work. Extrinsic design clears the path, but then you have to light the fire. The = fire is eros, and it=E2=80=99s fed by contact with things worthy of love: b= ooks, questions, lives, guides who still care about these things enough to = take them seriously in front of children. Whether that happens at school or at home is incidental. What matters is th= at a child sees adults who are genuinely stirred by ideas, who return to ce= rtain books not because they were assigned but because they can=E2=80=99t l= eave them alone. A six-year-old can learn a lot about what seriousness look= s like by watching someone practice it. Eugene Yiga asks: The accelerationist world still seems to dominate the public narrative by c= ommunicating in everyday language on everyday platforms in a way that meets= people where they actually are. Meanwhile, even the most accessible AI eth= ics content tends to assume familiarity with Mill, Tocqueville, or Heidegge= r. The philosopher-builder framing is compelling to people already inside t= he tent. How does Cosmos think about the people outside it? Is philosophica= l depth a feature for the community you=E2=80=99re building, or a barrier t= o the broader cultural shift you want to see? The honest answer is that depth is the point. If we watered down the philos= ophy so we could meet everyone where they are, we=E2=80=99d be producing th= e same frictionless content you see elsewhere. Philosophical seriousness cr= eates a negative selection gradient, and we want that. The people who do th= e reading are the people most likely to build something different. But =E2=80=9Cdepth=E2=80=9D and =E2=80=9Cjargon=E2=80=9D aren=E2=80=99t the= same thing. A lot of AI ethics writing assumes you=E2=80=99ve already read= Heidegger or whomever, which risks filtering out precisely the builders wh= o might be transformed by reading him. I know this because I=E2=80=99ve mad= e the mistake myself. When I started writing this Substack I leaned on more= jargon than I needed to, and I=E2=80=99ve had to learn over time how to ma= ke the ideas more accessible without making them thinner. The people outside the tent aren=E2=80=99t who you might think. I sold two = companies and wrote a national AI strategy, and I couldn=E2=80=99t write wh= at I believed on a single sheet of paper. There are a lot of capable builde= rs out there who never had anyone hand them the books or sit with them thro= ugh the hard parts. Cosmos partly exists because I was one of them. The aud= ience for this is bigger than it looks. Where I=E2=80=99d push back on your framing is the implicit suggestion that= the accelerationists win because they=E2=80=99re more accessible. They hav= e their own jargon. Try reading about negentropy, Kardashev III, and thermo= dynamic civilizational substrate for the first time. What they=E2=80=99ve d= one well is compress a real conceptual core into memes that travel. I respe= ct that. The challenge for us is that some ideas compress more easily than others. = =E2=80=9CBuild faster=E2=80=9D is more memeable than =E2=80=9Ccultivate jud= gment.=E2=80=9D =E2=80=9CTechnology goes up=E2=80=9D fits on a poster. =E2= =80=9CThe conditions under which free people can exercise genuine choice re= quire institutional renewal=E2=80=9D does not. This logic holds for political movements more generally: the larger the aud= ience you try to build, the cruder the message has to become. The lowest co= mmon denominator wins by default, not because it=E2=80=99s right but becaus= e it compresses. I don=E2=80=99t think the answer is to compete on that ter= rain. I think it=E2=80=99s to make the longer argument compelling enough th= at people seek it out, and to be honest that not everyone will. The harder truth is that we live in a culture of secondary orality where th= e long coherent essay is increasingly marginal. That=E2=80=99s a loss. It m= akes what we do at Cosmos more countercultural than it would have been fift= y years ago, but it also makes it more necessary. The essay, the book, the = salon: these are the forms where ideas actually get tested rather than just= transmitted. We=E2=80=99re not going to stop producing them because the cu= lture has moved on. If anything, the fact that sustained argument is now un= usual is exactly why it matters. Emily Kittley asks:=20 For someone coming to AI without a technical background but with a strong i= nterest in understanding its societal and philosophical implications, what = foundational books or resources would you recommend?=20 Second, as a parent, I=E2=80=99m thinking about how to prepare my kids for = a world where AI is increasingly embedded in everyday life. Beyond basic di= gital literacy, what kinds of skills, habits, or ways of thinking do you be= lieve will matter most for the next generation? Are there age-appropriate t= ools or frameworks you=E2=80=99d recommend for introducing AI concepts earl= y in a thoughtful, not just utilitarian, way? Hi Emily :) I=E2=80=99ll take the kids question first because it=E2=80=99s closer to my= heart. The risk I think about most is what I=E2=80=99ve called =E2=80=9Cautocomple= te for life=E2=80=9D: the possibility that AI systems will increasingly sha= pe not just what our children do but how they deliberate about what=E2=80= =99s worth doing. Each small delegation of judgment seems harmless. But tog= ether, they habituate a person away from self-governance and toward depende= nce. The question for parents is how you build resistance to that drift bef= ore your child is old enough to name it. Our ancestors needed to know how to make bread. We need to know where to fi= nd the recipe. The next generation will need something different again: the= capacity to think about how they think, in relation to systems that could = do the thinking for them. In our household, the main way we work on this is Socratic conversation. Ar= den and Pierce do weekly sessions with Michael Strong [ https://substack.co= m/redirect/d2d40d6e-3785-449f-85e2-69a715d8d971?j=3DeyJ1IjoiNXFxeXF4In0.h9d= EDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] built entirely around questions.= =E2=80=9CWhat=E2=80=99s the difference between a bird and a plane?=E2=80= =9D =E2=80=9CWhat does it mean for something to be alive?=E2=80=9D =E2=80= =9CWhen mommy and daddy disagree, who is right? What about daddy vs. AI? Wh= at about AI vs. AI?=E2=80=9D A child who has practiced working out what the= y believe, and who has had to think about whether to trust their own judgme= nt or defer to an external authority, is better prepared for a world of alg= orithmic suggestion than a child who has learned to code. I also want my kids to be entrepreneurial. When America was founded, around= 80% of free workers were self-employed on farms or in small crafts. Today = that number is about 10%. We became a society of employees, and something a= trophied. As the economy changes again, the ability to know yourself, act o= n what you believe, and build something from that conviction will matter mo= re than any technical skill we could teach them now. On resources for someone coming to AI without a technical background: I=E2= =80=99d start with the question of what AI does to us rather than how AI wo= rks. A couple of recent pieces that I=E2=80=99d recommend are S=C3=A9b Krie= r =E2=80=99s Musings on Self-Recursive Improvement [ https://substack.com/r= edirect/b0ec1789-b8a8-4b28-bb92-d7a0b6ae452a?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDb= Gj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] and Alex Imas =E2=80=99s What Will = Be Scarce [ https://substack.com/redirect/4b67d596-42ea-481e-a407-93236a7cc= 8d1?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ]. = For ongoing reading, Jack Clark , Azeem Azhar , and Ethan Mollick regularl= y write about AI and society. Jasmine Sun and Henrik Karlsson have a wide= r aperture and I often find them thought-provoking. For anyone interested i= n AI=E2=80=99s effects on democracy and self-governance, Harvey Mansfield= =E2=80=99s Tocqueville: A Very Short Introduction is the best ~100 pages yo= u could spend. Tocqueville saw the drift toward comfortable dependence comi= ng two centuries ago. The application to AI is left to the reader, but it i= sn=E2=80=99t hard to find. Substack Joe asks: My sense is that the vision animating Cosmos has deep predecessors not just= in classical philosophy but, in my impression, religious eschatology. Teil= hard de Chardin=E2=80=99s Omega Point or Augustine=E2=80=99s City of God, a= nd even secular variants like Condorcet=E2=80=99s perfectibilism all share = your orientation toward civilizational-scale transformation in service of h= uman flourishing. More explicitly, your pillars of reason, autonomy, and decentralization als= o echo the long Aristotelian and classical liberal tradition from Mill to T= ocqueville. So, what does Cosmos contribute that is genuinely novel in its normative ar= chitecture, rather than a restatement of those traditions in the presence o= f AI? And if it is largely a restatement, is that a problem?=20 I think you=E2=80=99re closer to the mark with some of these influences tha= n others. Teilhard, Augustine, Condorcet: I share their impulse toward civilizational= -scale thinking, and I take it seriously. But for all their differences, th= ey are ultimately teleological writers. They saw history as the unfolding o= f a determined, directional arc. At Cosmos, we want to keep the conditions = open that allow people to find their own path. We=E2=80=99re not about to g= et into eschatology. You are, of course, completely right about Aristotle, Mill, and Tocqueville= , and we regularly acknowledge our intellectual debt to them. I don=E2=80= =99t think the pillars need to be new to be worth defending, and I=E2=80=99= d be suspicious of anyone claiming to have invented a wholly new account of= human flourishing in 2026. For me, the interesting question isn=E2=80=99t whether Cosmos has discovere= d a value nobody thought of before. Instead, it=E2=80=99s whether an old se= t of commitments can survive as a living practice. Mill didn=E2=80=99t have= to ask whether the harm principle could be encoded in a model=E2=80=99s tr= aining objective. Tocqueville didn=E2=80=99t have to think about what decen= tralization looks like when the substrate is compute rather than townships,= when the everyday infrastructure of life anticipates your choices rather t= han forcing you to deliberate, associate, and decide alongside your neighbo= rs. When your community is mediated by algorithmic curation and your civic = life is shaped by systems you never consented to and cannot inspect, the To= cquevillian question of how free people learn to govern themselves together= doesn=E2=80=99t disappear. It becomes harder, and the institutional forms = it requires don=E2=80=99t exist yet. That=E2=80=99s where your last point lands, and I think it=E2=80=99s the ri= ght one. The proudest achievement of the eighteenth century was the transla= tion of philosophy into law: Enlightenment commitments about liberty, conse= nt, and the rights of individuals became encoded in constitutions and legal= systems that gave them institutional force. The challenge of the twenty-fi= rst century is the translation of philosophy into code. The commitments are= old. The work of making them operative in the infrastructure that actually= governs daily life is new, and it is the work Cosmos exists to do. But I wouldn=E2=80=99t call what we=E2=80=99re doing a restatement. Restate= ment is what you do in a seminar. Institutional embodiment is what you do w= hen you think the ideas actually matter and must be operative in the AI age= =2E Thomas Dias asks: What do you think of the prospects for a stable, left-right coalition on AI= in favor of sensible regulation and general cautious optimism that include= s religious conservatives and secular social democrats? Or will this get po= larized across political lines like everything else? On the coalition point, I can already see signs of this. Religious conserva= tives and secular social democrats agree on little, but they intuitively gr= asp some things that many accelerationists don=E2=80=99t: that people are f= ormed by their communities, work and dignity are connected, and that we sho= uldn=E2=80=99t try to optimize society into passivity. I=E2=80=99d also thr= ow old school liberals into that coalition too. In the coming years, I=E2= =80=99m sure there=E2=80=99ll be scope for productive, broad-based conversa= tions about kids, loneliness, work, and communities. Where I=E2=80=99d push back is the idea that any future coalition should co= alesce around =E2=80=9Csensible regulation.=E2=80=9D I don=E2=80=99t think = regulation is the best tool for addressing most of these concerns. Treating= it as the default is how you end up with something like the EU AI Act, a c= lassic example of doctor-induced illness. It created a compliance moat that= only the largest companies can afford to cross, while doing essentially no= thing to address the risks it was supposed to mitigate. The more productive ground is further upstream. What are we building? What = do we fund? What should we teach? What institutions do we need to form? A c= oalition focused on those questions would look less like a regulatory body = and more like a network of individuals doing the building, teaching, and fu= nding that no regulation can mandate. I=E2=80=99m less worried about polarization acting as an obstacle here. Muc= h of this work sits outside electoral politics at the moment, and as far as= I=E2=80=99m concerned the longer that remains the case the better. Partisa= n dynamics reward exactly the kind of simplification that makes these quest= ions worse. The moment AI becomes a left-right issue, the entire conversati= on becomes about how much to regulate, and the question of what to build fo= r never gets asked. Alina asks: Here is my question: Your three pillars (truth-seeking, autonomy, and decen= tralisation) are compelling at the individual level. I am curious how you t= hink about them when the actors are states rather than individuals. The US-= China AI dynamic, for instance, seems to run against all three: opacity rat= her than truth-seeking, control rather than autonomy, and concentration rat= her than decentralisation. Does Cosmos=E2=80=99s framework extend to the qu= estion of how countries could potentially cooperate on AI, or does that req= uire a different philosophical foundation entirely? Thanks Alina, great question. The pillars were designed with individuals and institutions in mind, so ext= ending them to the state level requires real philosophical work. Fichte took the Kantian account of individual autonomy and argued that it a= pplied to nations: a people that cannot determine its own form of life is u= nfree in the same sense an individual under tutelage is unfree. The autonom= y pillar, taken seriously, has a national analogue. So does truth-seeking: = a polity that can=E2=80=99t inquire openly into its own condition is in the= same trap as a closed mind. And so does decentralization: a world of self-= governing peoples is the international expression of the same instinct that= makes you wary of concentrated power inside a country. But Fichte also shows you what happens when you scale autonomy alone. His a= ttempt to extend individual self-determination to the collective ended in a= rguments for the unique world-historical mission of the German nation, an a= utarkic closed state, and the exclusion of those who didn=E2=80=99t fit the= national community. The lesson isn=E2=80=99t just =E2=80=9Cbe careful.=E2= =80=9D It=E2=80=99s that the three pillars need to travel together. Autonom= y without truth-seeking becomes self-righteousness. Autonomy without decent= ralization becomes domination. What checks national self-determination is t= he same thing that checks individual self-determination: openness to correc= tion and the refusal to concentrate power beyond what can be held accountab= le. On US-China, the goal isn=E2=80=99t a single global regime that imposes one= model of AI governance on everyone, because that would violate the decentr= alization commitment at the international scale. The better question is: wh= at conditions allow distinct political communities to develop AI in line wi= th their own forms of life without crushing each other in the process? And what happens when a community=E2=80=99s =E2=80=9Cform of life=E2=80=9D = involves suppressing the autonomy of its own citizens? The pillars can come= into tension here. Respect for national self-determination and respect for= individual autonomy pull in opposite directions. This is where Tocqueville matters most. The meaningful unit of self-governm= ent is rarely the nation-state on its own. It=E2=80=99s the dense layer of = associations, communities, firms, religious groups, and local institutions = that sit between the individual and the state. Any serious thinking about i= nternational AI governance has to make room for those middle layers. Tocque= ville saw that democratic freedom doesn=E2=80=99t live in declarations from= the center, but in the practice of self-government at the local and associ= ational level. Mark Frazier asks: Can you set up a path for crowdfunding projects or contests to realize idea= s that the Cosmos Institute seeds? Interesting. Not something we=E2=80=99ve considered, but we=E2=80=99ll thin= k about whether there=E2=80=99s a model that works for us. George asks: Do you plan to have online cohorts? No plans right now, but we may consider it in the future! Sarthak D asks: I see all these wonderful essays and people doing great work. Honestly, I w= ould love to interact with the community + become part of it in some capaci= ty. Is there a channel where people who are interested in the ideas that Co= smos is working towards but not necessarily are academics or builders can c= ommunicate with the fellows and the team?=20 Not right now, but we are thinking about whether there=E2=80=99s something = we can do here! Kevin Cutright asks: I=E2=80=99m persuaded by the concern about cognitive risks and the need for= =E2=80=9CAI for epistemics,=E2=80=9D =E2=80=9Cdeliberative AI,=E2=80=9D et= c. Do you know of organizations developing benchmarks around the goal of bo= lstering critical thinking and improving epistemic processes and outcomes? We have some grant projects that have focused on this. Two that come to min= d are DeliberationBench [ https://substack.com/redirect/1815bb6c-576c-40e1-= b6cb-6759b323defd?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoF= v6eWS-HyoA ], which assesses AI persuasion in comparison with diverse human= discussion, and Priori [ https://substack.com/redirect/4ae4da77-7222-46be-= b5f4-085f5348e6a3?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoF= v6eWS-HyoA ], a tool that surfaces hidden assumptions when you are interact= ing with an AI model. Two of our grantees (Steven Molotnikov [ https://subs= tack.com/redirect/a1c6b8ab-0a46-41d0-9cac-790fad7ab0b5?j=3DeyJ1IjoiNXFxeXF4= In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] and Cathy Fang [ https://= substack.com/redirect/2dd99ba2-c50e-4d76-9724-50ba1aa0a13a?j=3DeyJ1IjoiNXFx= eXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ]) are running a resear= ch study on how Priori and related human oversight interfaces work in pract= ice. I think there is a wave of energy in this area. Various orgs are thinking m= ore about AI for Human Reasoning (with Future of Life Foundation funding [ = https://substack.com/redirect/180ba751-b63f-4d0a-ae6a-8d83bccaa6c6?j=3DeyJ1= IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] work in this = area, Forethought writing [ https://substack.com/redirect/5788a1f6-c547-4be= 6-a76b-dab79e249850?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKG= oFv6eWS-HyoA ] about it, and Elicit [ https://substack.com/redirect/a211203= e-40b2-4570-9093-6c3e79928420?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbM= YWklSu8UKGoFv6eWS-HyoA ] working on directly in the for-profit space). Also= anecdotally I hear researchers thinking more about ideas like =E2=80=9Cepi= stemic security=E2=80=9D or =E2=80=9Ccognitive security=E2=80=9D or =E2=80= =9Ccognitive sovereignty=E2=80=9D as well as ways to improve information en= vironments without restricting speech and expression. I share your enthusiasm for more work in this area =E2=80=93 both on benchm= arking but also technology that better enables open contestation of ideas (= inspired by classical liberal premises, and Mill=E2=80=99s ideas on this). = If readers are working on this please do reach out! Cosmos Institute [ https://substack.com/redirect/dc165246-e784-41fa-852d-aa= bdb8c9a3b1?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-H= yoA ] is the Academy for Philosopher-Builders, technologists building AI fo= r human flourishing. We run fellowships, fund AI prototypes, and host semin= ars with institutions like Oxford, Aspen Institute, and Liberty Fund. Unsubscribe https://substack.com/redirect/2/eyJlIjoiaHR0cHM6Ly9ibG9nLmNvc21= vcy1pbnN0aXR1dGUub3JnL2FjdGlvbi9kaXNhYmxlX2VtYWlsP3Rva2VuPWV5SjFjMlZ5WDJsa0= lqb3pORGN5TlRnNU9EVXNJbkJ2YzNSZmFXUWlPakU1TkRNd05qSTVPU3dpYVdGMElqb3hOemMyT= Wpjek5qYzNMQ0psZUhBaU9qRTRNRGM0TURrMk56Y3NJbWx6Y3lJNkluQjFZaTB5TWpJMU56azBJ= aXdpYzNWaUlqb2laR2x6WVdKc1pWOWxiV0ZwYkNKOS55ckhDVV8yblRYZGtmbnlLRV81bmpWWC1= VcWdRWjlrd2xIWUZJTHZ1RUwwIiwicCI6MTk0MzA2Mjk5LCJzIjoyMjI1Nzk0LCJmIjp0cnVlLC= J1IjozNDcyNTg5ODUsImlhdCI6MTc3NjI3MzY3NywiZXhwIjoyMDkxODQ5Njc3LCJpc3MiOiJwd= WItMCIsInN1YiI6ImxpbmstcmVkaXJlY3QifQ.G_92-2yNth_2vsx4794AuEomoHmqWzo0hxm6u= -DSLKc? --a5c2ad23665ebd8657b8089bca91d63d4c4b21bf3aaef79b9915a6738176 Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable = Can Old Ideas Survive the AI Age?
Your questions answered: on philosophy, chil= dren, and China
͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­͏ =     ­͏     ­͏     = 3;͏     ­͏     ­͏   = 199; ­͏     ­͏     ­͏ &= nbsp;   ­͏     ­͏     ­= ;͏     ­͏     ­͏   Q= 99; ­͏     ­͏     ­͏ &n= bsp;   ­͏     ­͏     ­= ͏     ­͏     ­͏   ̳= 9; ­͏     ­͏     ­͏ &nb= sp;   ­͏     ­͏     ­&= #847;     ­͏     ­͏    = ; ­͏     ­͏     ­͏ &nbs= p;   ­͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­͏ =     ­͏     ­͏     = 3;͏     ­͏     ­͏   = 199; ­͏     ­͏     ­͏ &= nbsp;   ­͏     ­͏     ­= ;͏     ­͏     ­͏   Q= 99; ­͏     ­͏     ­͏ &n= bsp;   ­͏     ­͏     ­= ͏     ­͏     ­͏   ̳= 9; ­͏     ­͏     ­͏ &nb= sp;   ­͏     ­͏     ­&= #847;     ­͏     ­͏    = ; ­͏     ­͏     ­͏ &nbs= p;   ­͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­͏ =     ­͏     ­͏     = 3;͏     ­͏     ­͏   = 199; ­͏     ­͏     ­
Forwarded this email? Subscribe here for more
Can Old Ideas Survive the AI Age? <= /h1>

= Your questions answered: on philosophy, children, and China

=
Brendan McCord
=
=
 
=
3D""
3D""
3D""
3D""
READ IN APP
 
3D""
An Unidentified Classical Subject: A Fight by Antonio Z= ucchi (1767)

Last week, we m= arked 20,000 subscribers by opening the floor. Your questions were often ge= nuinely challenging. Thank you to everyone who took the time.

So you don’t have to trawl through the comments, we’ve com= piled all the responses in one place. Apologies to anyone whose question di= dn’t make the cut. There were a few I’m still chewing over and = we’ll revisit many of these topics at greater length in future essays= =2E Let us know in the comments if this= is a format you’d like us to re= peat.

Upgrade to paid
=

Anna Lisa asks:

Are there specific= non-academic experiences or “containers” of formation that you= think hold a lot of promise? (e.g. ones that were meaningful to you or one= s that you seek out for yourself/your children?)

A= s a dad, I love this question.

A lot of the important forma= tion happens in places that do not look educational at all, and are not pri= marily about instruction. They are about habituation, responsibility, emula= tion, and contact with reality.

The first was the household= I grew up in. My mother was a Catholic conservative historian from a milit= ary family who taught special needs kids for 36 years. My father was a left= -leaning physicist turned environmental lawyer, and a pacifist. They agreed= on almost nothing politically. But they agreed on something deeper: that y= ou should care about something beyond yourself and that how you act matters= more than what you know (and that knowing is bound up in doing!). That gav= e me, without anyone naming it, a kind of virtue culture, and I think it= 217;s the kind of thing that’s very hard to manufacture deliberately = but very obvious in its absence. For children, what seems to matter is not = ideological uniformity or even ideological sophistication, but a home in wh= ich seriousness, duty, and moral aspiration are normal.

The= second was the submarine. I spent 610 days underwater, including under ice= =2E In a steel tube, you do not get to=20= opt out of reality. You can’t le= ave, and your mistakes could get someone killed. That kind of environment f= orms you because it imposes standards that are not negotiable. It teaches s= ervice, competence, and mutual reliance in a way that is hard to simulate. = I think containers of formation are often places with real stakes, shared d= iscipline, and demands that do not bend to your preferences.

my original list = is here). Having children makes the question of what you’re= actually “for” impossible to defer.

I n= ow look for opportunities in my own children’s lives for containers t= hat place them in contact with reality, responsibility, and admirable adult= s. The hard part is that the best formation is often a byproduct rather tha= n something you can engineer directly. You can build the conditions for it,= but you usually cannot force it.

Turner Halle asks:

= You argue that philosopher-builders need explicit moral commitments= to avoid optimizing for the wrong things. But your three pillars (truth-se= eking, autonomy, decentralization) are themselves a normative framework tha= t not everyone shares. China’s AI strategy is still coherent, explici= t, and philosophical, it just starts from different premises. So how do you= argue for your philosophy without just replacing one set of defaults with = another? What makes Cosmos’s values the right foundation rather than = just a well-packaged preference?

Hi Turner, you= 217;re right that truth-seeking, autonomy, and decentralization are substan= tive commitments. I think they matter less as one moral doctrine than as co= nditions that keep moral life from collapsing into force or drift.

If you consider moral frameworks from Confucianism to Christianity = to Marxism, for them to have legitimate force over a person, that person ha= s to be able to genuinely endorse it (otherwise, you don’t have a mor= al commitment). That endorsement depends on human autonomy – which is= to say, the capacity to reflect, evaluate, and take something on as your o= wn rather than merely inheriting it or obeying it. So autonomy is not just = one preference among others. It is the deep substratum that makes moral com= mitment possible at all.

Take utilitarianism: Jeremy = Bentham devised a system that, in my mind, dissolves individual judgment in= to aggregate utility. This is in conflict with autonomy-as-an-end. And yet = building this system was itself a radical exercise of= autonomous reason. Every person who adopts utilitarianism is exercising th= e same capacity. You can’t be a utilitarian in any meaningful sense u= nless you’ve freely taken it on. So even a framework that subordinate= s individual judgment to aggregate welfare requires individual judgment to = get off the ground. Now, someone could say that only makes autonomy instrum= entally necessary, not foundationally important. I think that view is unsta= ble, because the goods autonomy is supposedly serving only become moral goo= ds for a person if they can in some real sense take them on as their own.

There are hard cases. In the Ash’ari trad= ition in Islamic theology, divine command constitutes= moral value rather than being something reason independently discovers and= then endorses. That’s a genuine challenge to autonomy as foundationa= l. But even there, the person who freely chooses submission is doing someth= ing categorically different from the person who never had the choice. And a= secular collectivist can make a parallel argument: that harmony or collect= ive flourishing is the true precondition, because no individual life goes w= ell outside a stable social order. I think that is partly right. But unless= people can participate in judging the terms of that order, harmony becomes= coordination imposed on them rather than a good they share in shaping.

Truth-seeking has a similar status. Any framework worthy o= f allegiance has to remain in contact with reality. It has to be open, at l= east in principle, to its own refutation. If someone could show me that dec= entralization produces a worse outcome in a domain I care about, I’d = have to take that seriously and I would. Systems that suppress truth-seekin= g can be internally coherent in the way that closed systems are coherent. B= ut they can’t tolerate the mechanisms that would let them find out th= ey are mistaken. That’s a serious defect, especially if you think we&= #8217;re all operating under real uncertainty about what AI is going to do = to human life.

And decentralization follows from the= same logic at the institutional level. If no person or committee is wise e= nough to determine the good for everyone, then we should be wary of a small= number of actors hard-coding their anthropology into the substrate of soci= ety. Decentralization is valuable because it preserves room for people to t= ry things, get them wrong, and leave, keeping mistakes from becoming total.=

So to your question about China: yes, their AI strategy is= coherent, explicit, and philosophical. But coherence purchased by foreclos= ing the capacity for self-correction is brittle in exactly the way that mat= ters most right now, though that is still a bet, not a proof. I’m als= o skeptical that it preserves the standing of human beings as agents capabl= e of judgment rather than increasingly treating them as objects of coordina= tion. A framework can be philosophically serious and still be wrong about w= hat a person is.

So I would not say Cosmos is advancing a o= ne true final doctrine that every civilization must affirm. That would be t= oo strong, and it would collapse into exactly the kind of totalizing move y= ou are warning against. I would say instead that Cosmos is trying to defend= the conditions under which free people can genuinely seek truth, make judg= ments, form commitments, and build different kinds of lives together, witho= ut having orthodoxy imposed on them by default.

<= span class=3D"mention-pnpTE1 mentionStatic-liMklk" data-component-name=3D"M= entionStatic" style=3D"white-space: nowrap;cursor: pointer;color: #25384c;b= order-radius: 4px;transition: color .25scubic-bezier(.19,1,.22,1),backgroun= d-color .25scubic-bezier(.19,1,.22,1),box-shadow .25scubic-bezier(.19,1,.22= ,1),border .25scubic-bezier(.19,1,.22,1),border-radius .25scubic-bezier(.19= ,1,.22,1),opacity .25scubic-bezier(.19,1,.22,1),filter .25scubic-bezier(.19= ,1,.22,1),stroke .25scubic-bezier(.19,1,.22,1),transform .25scubic-bezier(.= 19,1,.22,1),scale .25scubic-bezier(.19,1,.22,1),outline .25scubic-bezier(.1= 9,1,.22,1);margin: 0 -.125em;padding: 0 .125em .0625em;text-decoration: non= e;">Salvador Duarte asks:

Will the Cosmos Grants ever open again?

We’re planning to re-open these in the next 60 day= s. We’ve just had the demos from our latest batches of winners. Stay = tuned for updates!

Bert Cl= ements asks:

Assu= ming frontier large language models, together with their multimodal and age= ntic extensions, are trained to effective saturation on an exhaustive corpu= s that represents the totality of digitized human knowledge including all s= cientific publications, books, patents, archival records, cultural artifact= s, and recorded conversations, will these systems be capable of transcendin= g the statistical manifold of their training distribution to autonomously d= iscover, validate, and iteratively expand novel knowledge beyond the curren= t human frontier?

I’m not sure scientific kn= owledge is a kind of territory where data defines a bounded region that a m= odel may or may not be able to venture beyond. The affirmative view of this= picture strikes me as broadly empiricist insofar as knowledge comes from d= ata, and scientists make discoveries by extrapolating beyond what they alre= ady know. Your specific examples, though, are actually the strongest case f= or the affirmative: theorem provers, simulators, multi-agent workflows, and= verifiable rewards are exactly the kinds of feedback-rich settings where I= would expect systems to extend the frontier.

But that is n= ot the only picture of science, and I do not think it is the deepest one. S= cience also advances by reorganizing what researchers take to be meaningful= in the first place: which anomalies matter, which questions are worth aski= ng, and which explanations count as illuminating rather than merely predict= ive.

Systems can already generate novel candidate hypothese= s, and in domains with strong automated verification, they may well extend = the frontier. Formal mathematics looks especially promising, because conjec= ture can be paired with proof or disproof inside a relatively crisp evaluat= ive architecture. In such cases, I expect AI systems to produce results tha= t are genuinely new to humanity. Just this week a constellation of agents i= mproved a math problem that’s been open since Newton (Kissing Number = in dimension 11: 593 → 604). That is impressive. It is also, I think,= a good example of the distinction I’m drawing: a real extension of a= n existing line of inquiry, but still closer to powerful normal science tha= n to scientific revolution.

But that does not settle the la= rger question. There is a difference between producing novelty within an ex= isting framework and generating a new framework altogether. A system may he= lp prove a theorem, optimize a search, or identify that drug X affects dise= ase Y, all without altering our understanding of why the problem is structu= red as it is. That is a real scientific contribution, but it is not reorgan= izing the conceptual landscape.

The harder question is whet= her these systems can exercise scientific judgment in the richer sense: whe= ther they can tell which anomalies are significant, which inconsistencies a= re fertile, which explanations deepen understanding rather than merely exte= nd prediction, and which questions are worth reorganizing inquiry around. T= hat is a higher bar than novelty, and I am not yet convinced we know how to= evaluate it well. Part of what makes this hard is that frameworks are unde= rdetermined by data. The same body of results can often support multiple li= nes of inquiry, and judgment is what tells you which one is worth building = a field around. That remains, to my mind, the deeper open question.

tappert asks:

Most of the current work on ‘AI, col= lective epistemic structures and decision-making’ focuses on filling = gaps: more participants, faster information exchange, more efficient decisi= on-making. This will help with many problems, but certainly not with the mo= st complex ones, because it just accelerates the practical execution of the= same thought styles that led to the problems. Therefore: How can we use fu= ture AI to foster new thought styles that are currently not supported by ou= r existing social structures?

Yes, I think the int= uition that better collective decisions will emerge if we simply gather mor= e data from more people more efficiently breaks down at the limit. That can= improve performance within an existing paradigm, but it does much less whe= n the paradigm itself is the problem.

What groups develop o= ver time are not just bodies of knowledge, but epistemic constitutions: imp= licit rules about what counts as evidence, which questions are legitimate, = who gets to propose, who gets to criticize, and on what terms. Mill saw par= t of this in his account of the tyranny of prevailing opinion and the epist= emic importance of dissent. But the problem runs deeper than opinion alone.= Entire institutions decide in advance what counts as serious thought.

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">So one promising use of AI would be to make those constitutions= more visible. A good system might show a research community, an organizati= on, or a polity where its methods systematically exclude certain questions,= place some assumptions beyond criticism, or discount certain voices before= the argument even begins. In medicine, for example, it might reveal a fiel= d that privileges what is easily measurable while sidelining patient testim= ony or long-horizon effects that do not fit the dominant method.

But diagnosis is only the beginning. I like this direction because th= e problem is often not that new thought styles do not exist. It is that the= y remain stranded at the margins because the reigning structures of legitim= acy suppress them. And sometimes the deeper problem is that the social cond= itions required for a new thought style have not yet been built. New though= t styles need protected spaces, alternative standards, and enough provision= al legitimacy to develop before the dominant paradigm dismisses them. In th= at case, the most useful contribution AI could make to collective epistemic= s is not novelty on demand, but widening the space in which criticism, reco= mbination, and intellectual minority formation can occur.

Thomas Yiu asks:

What is your definition of intelligence? When = AI reaches ASI in the future, do you think it will be safe and aligned? As = a species, what is our purpose after ASI world? How can thrive as a species= ?

I’d resist the standard definition of inte= lligence as raw problem-solving horsepower. For me, intelligence is the cap= acity to learn from reality, inquire into it well, and let it correct you.<= /p>

Part of what I like about François Chollet’s = work, and why ARC Prize has mattered, is the insistence th= at intelligence is not the same thing as accumulated skill. A system can lo= ok impressive because it has absorbed an enormous amount, or because the ta= sk has been made easy for it. The more interesting question is how much it = can learn from limited experience, under real constraints, and still genera= lize well.

But I do not think that is enough o= n its own. Leslie Valiant’s idea of educability gets closer to the hu= man picture (see his Cosmos Lecture from last year here). Human inte= lligence includes the capacity to learn from experience, receive instructio= n, integrate both, and apply them in new circumstances. What distinguishes = the human mind is not only that it learns, but that it can be taught and fo= rmed.

And I would add one more layer. Drawing = on = HAI Lab director Philipp Koralus, I think reasoning is fundamenta= lly question-directed. Minds are shaped by the questions they pursue. They = go wrong through shallow questions, premature closure, and a failure to inq= uire far enough, just as much as through false conclusions. That matters fo= r AI because a system can become very good at answering questions while sti= ll narrowing the range of questions humans ask, or rewarding closure where = inquiry ought to stay open.

That is why I’m le= ss interested in arguing about whether AI will count as “superintelli= gent” than in asking what it does to human intelligence. A system can= be extraordinarily capable and still erode our capacity for inquiry, judgm= ent, and self-government. That is the danger I worry about most.

On whether ASI will be safe and aligned: I do not assume that can be = taken for granted. I would trust highly capable systems only to the extent = that they remain corrigible, contestable, and embedded in institutions that= preserve human judgment rather than replacing it. The problem is not just = getting the objective right once. It is making sure people can still questi= on, revise, and refuse the system’s guidance when it matters most.

As for human purpose after ASI, I do not think our purpose ch= anges. If anything, it comes into clearer view. We are not here to compete = with machines at speed or scale. We are here to exercise judgment, form cha= racter, build institutions, love particular people, and deliberate about th= e good. That last point matters more than it may sound. Love is not interch= angeable, and responsibility is not abstract. A more powerful machine does = not make our obligations to particular human beings less central. In a worl= d of highly capable AI, those things become even more important.

Todd Enkhbat asks:

Is it possible to carry on our lear= ning from humanity up until now and jumpstart a new society with the help o= f AI, assuming that we can concentrate and utilize all the data we accumula= ted up until now? At what point does the need for a new constitution or a n= ew world order arise and how do we know it?

In sho= rt, no.

Firstly, I don’t think “all the data we= accumulated up until now” is the same thing as the total weight of h= uman knowledge. Much of the knowledge that keeps a society functioning is t= acit, dispersed, and unwritten. Some of it lives in practiced judgment: an = ICU nurse sensing that a patient is about to crash before the monitor shows= it. Some of it lives in inherited forms: the habits of trust, restraint, a= nd association on which a free society depends, even when no one can fully = specify them. As Michael Polanyi put it, we know more than we can tell.

=

More importantly, I’d push back on the idea that we can = jumpstart a society at all. Societies aren’t machines that you design= to a blueprint. Tocqueville saw this in the institutions of local self-gov= ernment. Hayek saw it in the way social orders carry dispersed knowledge th= at no planner can gather in full. A free society is learned in practice thr= ough things like townships, juries, churches, and associations. Those are t= he ordinary disciplines by which people become capable of governing themsel= ves.

The question is whether our institutions can still sus= tain a free people capable of self-government under new technological condi= tions. And that does leave open the question you raise about constitutional= inadequacy: how do we know when inherited arrangements are no longer enoug= h? I do not think there is a clean threshold. Usually the signs are visible= first in practice, when institutions that once formed judgment begin produ= cing passivity, dependence, or elite insulation instead.

When they cannot, the answer is not a tabula rasa redesign of “th= e new world order.” I would look to renewal through institution-build= ing, and Benjamin Franklin is the example I keep returning to. He took an E= nlightenment conviction — that access to knowledge should not remain = under the custody of church, state, or a narrow elite — and embodied = it in an institution. The subscription library made a philosophy of freedom= socially real. That is why Franklin still matters to me here. He shows wha= t it looks like to translate a philosophy into civic machinery. We need the= AI-age equivalent: institutions that widen access to knowledge and judgmen= t without concentrating them in a few hands. We need philosopher-builders<= span> in that spirit again.

Miss Zanarkand asks:

How can we motivate our children to learn at school? Should= we try to motivate them or find rather a way out of the system? (e.g. read= ing more classical books, rather than encouraging them to read what school = nowadays gives?)

Young people have a natural longi= ng to be seized by something greater than themselves. To be captivated. The= promise of liberal education, going back to the Greeks, is that there are = magnificent ways of living, and magnificent questions about how to live, an= d that encountering them through great minds and great books can awaken a d= esire that organizes everything else.

The disaster of moder= n education is that it has taught young people their longing is naive. That= no book is really better than another, that no life is really higher than = another, and that the hunger to be drawn upward by something extraordinary = is itself a kind of error.

So I would say motivation = is the right place to focus, but we should be precise about what we mean. T= here is a kind of motivation that is intrinsic: the eros I just described, the desire to encounter greatness because it calls to= something real inside you. And there is extrinsic motivation: incentives, = structure, well-designed systems that make it easier to do the work. Both m= atter. The best schools I’ve seen, including Alpha where my kids go, = are serious about the extrinsic architecture. They’ve built an enviro= nment where children actually want to show up and work.

Extrinsic design clears the path, but then you have to light the= fire. The fire is eros, and it’s fed by contac= t with things worthy of love: books, questions, lives, guides who still car= e about these things enough to take them seriously in front of children.

Whether that happens at school or at home is incidenta= l. What matters is that a child sees adults who are genuinely stirred by id= eas, who return to certain books not because they were assigned but because= they can’t leave them alone. A six-year-old can learn a lot about wh= at seriousness looks like by watching someone practice it.

Eugene Yiga asks:

The accelerationist world still seems to domi= nate the public narrative by communicating in everyday language on everyday= platforms in a way that meets people where they actually are. Meanwhile, e= ven the most accessible AI ethics content tends to assume familiarity with = Mill, Tocqueville, or Heidegger. The philosopher-builder framing is compell= ing to people already inside the tent. How does Cosmos think about the peop= le outside it? Is philosophical depth a feature for the community you’= ;re building, or a barrier to the broader cultural shift you want to see?

The honest answer is that depth is the point. If we= watered down the philosophy so we could meet everyone where they are, we&#= 8217;d be producing the same frictionless content you see elsewhere. Philos= ophical seriousness creates a negative selection gradient, and we want that= =2E The people who do the reading are t= he people most likely to build somethi= ng different.

But “depth” and “jargonR= 21; aren’t the same thing. A lot of AI ethics writing assumes youR= 17;ve already read Heidegger or whomever, which risks filtering out precise= ly the builders who might be transformed by reading him. I know this becaus= e I’ve made the mistake myself. When I started writing this Substack = I leaned on more jargon than I needed to, and I’ve had to learn over = time how to make the ideas more accessible without making them thinner.

=

The people outside the tent aren’t who you might think. = I sold two companies and wrote a national AI strategy, and I couldn’t= write what I believed on a single sheet of paper. There are a lot of capab= le builders out there who never had anyone hand them the books or sit with = them through the hard parts. Cosmos partly exists because I was one of them= =2E The audience for this is bigger th= an it looks.

Where I= 217;d push back on your framing is the implicit suggestion that the acceler= ationists win because they’re more accessible. They have their own ja= rgon. Try reading about negentropy, Kardashev III, and thermodynamic civili= zational substrate for the first time. What they’ve done well is comp= ress a real conceptual core into memes that travel. I respect that.

The challenge for us is that some ideas compress more easily than = others. “Build faster” is more memeable than “cultivate j= udgment.” “Technology goes up” fits on a poster. “T= he conditions under which free people can exercise genuine choice require i= nstitutional renewal” does not.

This logic holds for = political movements more generally: the larger the audience you try to buil= d, the cruder the message has to become. The lowest common denominator wins= by default, not because it’s right but because it compresses. I don&= #8217;t think the answer is to compete on that terrain. I think it’s = to make the longer argument compelling enough that people seek it out, and = to be honest that not everyone will.

The harder truth is th= at we live in a culture of secondary orality where the long coherent essay = is increasingly marginal. That’s a loss. It makes what we do at Cosmo= s more countercultural than it would have been fifty years ago, but it also= makes it more necessary. The essay, the book, the salon: these are the for= ms where ideas actually get tested rather than just transmitted. We’r= e not going to stop producing them because the culture has moved on. If any= thing, the fact that sustained argument is now unusual is exactly why it ma= tters.

Emily Kittley asks:

For someone com= ing to AI without a technical background but with a strong interest in unde= rstanding its societal and philosophical implications, what foundational bo= oks or resources would you recommend?

Sec= ond, as a parent, I’m thinking about how to prepare my kids for a wor= ld where AI is increasingly embedded in everyday life. Beyond basic digital= literacy, what kinds of skills, habits, or ways of thinking do you believe= will matter most for the next generation? Are there age-appropriate tools = or frameworks you’d recommend for introducing AI concepts early in a = thoughtful, not just utilitarian, way?

Hi Emily :)=

I’ll take the kids question first because it’s= closer to my heart.

The risk I think about most is what I&= #8217;ve called “autocomplete for life”: the possibility that A= I systems will increasingly shape not just what our children do but how the= y deliberate about what’s worth doing. Each small delegation of judgm= ent seems harmless. But together, they habituate a person away from self-go= vernance and toward dependence. The question for parents is how you build r= esistance to that drift before your child is old enough to name it.

Our ancestors needed to know how to make bread. We need to know wh= ere to find the recipe. The next generation will need something different a= gain: the capacity to think about how they think, in relation to systems th= at could do the thinking for them.

In our household, = the main way we work on this is Socratic conversation. Arden and Pierce do = weekly sessions with Michael Strong built entirely around questions.= “What’s the difference between a bird and a plane?” R= 20;What does it mean for something to be alive?” “When mommy an= d daddy disagree, who is right? What about daddy vs. AI? What about AI vs. = AI?” A child who has practiced working out what they believe, and who= has had to think about whether to trust their own judgment or defer to an = external authority, is better prepared for a world of algorithmic suggestio= n than a child who has learned to code.

I also want = my kids to be entrepreneurial. When America was founded, around 80% of free= workers were self-employed on farms or in small crafts. Today that number = is about 10%. We became a society of employees, and something atrophied. As= the economy changes again, the ability to know yourself, act on what you b= elieve, and build something from that conviction will matter more than any = technical skill we could teach them now.

On resources= for someone coming to AI without a technical background: I’d start w= ith the question of what AI does to us rather than ho= w AI works. A couple of recent pieces that I’d recommend are <= span class=3D"mention-pnpTE1 mentionStatic-liMklk" data-component-name=3D"M= entionStatic" style=3D"white-space: nowrap;cursor: pointer;color: #25384c;b= order-radius: 4px;transition: color .25scubic-bezier(.19,1,.22,1),backgroun= d-color .25scubic-bezier(.19,1,.22,1),box-shadow .25scubic-bezier(.19,1,.22= ,1),border .25scubic-bezier(.19,1,.22,1),border-radius .25scubic-bezier(.19= ,1,.22,1),opacity .25scubic-bezier(.19,1,.22,1),filter .25scubic-bezier(.19= ,1,.22,1),stroke .25scubic-bezier(.19,1,.22,1),transform .25scubic-bezier(.= 19,1,.22,1),scale .25scubic-bezier(.19,1,.22,1),outline .25scubic-bezier(.1= 9,1,.22,1);margin: 0 -.125em;padding: 0 .125em .0625em;text-decoration: non= e;">Séb Krier’s Musings on Self-R= ecursive Improvement and Alex Imas’s What Will Be Scarce. For ongoing re= ading, Jack Clark, Azeem Azhar, and Ethan= Mollick regularly write about AI and society. Jasmine Sun and Henrik Karlsson have a wider aperture and I often find = them thought-provoking. For anyone interested in AI’s effects on demo= cracy and self-governance, Harvey Mansfield’s Tocqueville:= A Very Short Introduction is the best ~100 pages you could spen= d. Tocqueville saw the drift toward comfortable dependence coming two centu= ries ago. The application to AI is left to the reader, but it isn’t h= ard to find.

Substack= Joe asks:

My sen= se is that the vision animating Cosmos has deep predecessors not just in cl= assical philosophy but, in my impression, religious eschatology. Teilhard d= e Chardin’s Omega Point or Augustine’s City of God, and even se= cular variants like Condorcet’s perfectibilism all share your orienta= tion toward civilizational-scale transformation in service of human flouris= hing.

More explicitly, your pillars of rea= son, autonomy, and decentralization also echo the long Aristotelian and cla= ssical liberal tradition from Mill to Tocqueville.

So, what does Cosmos contribute that is genuinely novel in its nor= mative architecture, rather than a restatement of those traditions in the p= resence of AI? And if it is largely a restatement, is that a problem?

I think you’re closer to the mark with some of t= hese influences than others.

Teilhard, Augustine, Condorcet= : I share their impulse toward civilizational-scale thinking, and I take it= seriously. But for all their differences, they are ultimately teleological= writers. They saw history as the unfolding of a determined, directional ar= c. At Cosmos, we want to keep the conditions open that allow people to find= their own path. We’re not about to get into eschatology.

You are, of course, completely right about Aristotle, Mill, and Tocque= ville, and we regularly acknowledge our intellectual debt to them. I don= 217;t think the pillars need to be new to be worth defending, and I’d= be suspicious of anyone claiming to have invented a wholly new account of = human flourishing in 2026.

For me, the interesting question= isn’t whether Cosmos has discovered a value nobody thought of before= =2E Instead, it’s whether an old=20= set of commitments can survive as a li= ving practice. Mill didn’t have to ask whether the harm principle cou= ld be encoded in a model’s training objective. Tocqueville didn’= ;t have to think about what decentralization looks like when the substrate = is compute rather than townships, when the everyday infrastructure of life = anticipates your choices rather than forcing you to deliberate, associate, = and decide alongside your neighbors. When your community is mediated by alg= orithmic curation and your civic life is shaped by systems you never consen= ted to and cannot inspect, the Tocquevillian question of how free people le= arn to govern themselves together doesn’t disappear. It becomes harde= r, and the institutional forms it requires don’t exist yet.

That’s where your last point lands, and I think it’s the= right one. The proudest achievement of the eighteenth century was the tran= slation of philosophy into law: Enlightenment commitments about liberty, co= nsent, and the rights of individuals became encoded in constitutions and le= gal systems that gave them institutional force. The challenge of the twenty= -first century is the translation of philosophy into code. The commitments = are old. The work of making them operative in the infrastructure that actua= lly governs daily life is new, and it is the work Cosmos exists to do.

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">But I wouldn’t call what we’re doing a restatement.= Restatement is what you do in a seminar. Institutional embodiment is what = you do when you think the ideas actually matter and must be operative in th= e AI age.

Thomas Dias asks:

What do you think o= f the prospects for a stable, left-right coalition on AI in favor of sensib= le regulation and general cautious optimism that includes religious conserv= atives and secular social democrats? Or will this get polarized across poli= tical lines like everything else?

On the coalition= point, I can already see signs of this. Religious conservatives and secula= r social democrats agree on little, but they intuitively grasp some things = that many accelerationists don’t: that people are formed by their com= munities, work and dignity are connected, and that we shouldn’t try t= o optimize society into passivity. I’d also throw old school liberals= into that coalition too. In the coming years, I’m sure there’l= l be scope for productive, broad-based conversations about kids, loneliness= , work, and communities.

Where I’d push back is the i= dea that any future coalition should coalesce around “sensible regula= tion.” I don’t think regulation is the best tool for addressing= most of these concerns. Treating it as the default is how you end up with = something like the EU AI Act, a classic example of doctor-induced illness. = It created a compliance moat that only the largest companies can afford to = cross, while doing essentially nothing to address the risks it was supposed= to mitigate.

The more productive ground is further upstrea= m. What are we building? What do we fund? What should we teach? What instit= utions do we need to form? A coalition focused on those questions would loo= k less like a regulatory body and more like a network of individuals doing = the building, teaching, and funding that no regulation can mandate.

I’m less worried about polarization acting as an obstacle he= re. Much of this work sits outside electoral politics at the moment, and as= far as I’m concerned the longer that remains the case the better. Pa= rtisan dynamics reward exactly the kind of simplification that makes these = questions worse. The moment AI becomes a left-right issue, the entire conve= rsation becomes about how much to regulate, and the question of what to bui= ld for never gets asked.

Alina= asks:

Here is my que= stion: Your three pillars (truth-seeking, autonomy, and decentralisation) a= re compelling at the individual level. I am curious how you think about the= m when the actors are states rather than individuals. The US-China AI dynam= ic, for instance, seems to run against all three: opacity rather than truth= -seeking, control rather than autonomy, and concentration rather than decen= tralisation. Does Cosmos’s framework extend to the question of how co= untries could potentially cooperate on AI, or does that require a different= philosophical foundation entirely?

Thanks Alina, = great question.

The pillars were designed with individuals = and institutions in mind, so extending them to the state level requires rea= l philosophical work.

Fichte took the Kantian account of in= dividual autonomy and argued that it applied to nations: a people that cann= ot determine its own form of life is unfree in the same sense an individual= under tutelage is unfree. The autonomy pillar, taken seriously, has a nati= onal analogue. So does truth-seeking: a polity that can’t inquire ope= nly into its own condition is in the same trap as a closed mind. And so doe= s decentralization: a world of self-governing peoples is the international = expression of the same instinct that makes you wary of concentrated power i= nside a country.

But Fichte also shows you what happe= ns when you scale autonomy alone. His attempt to exte= nd individual self-determination to the collective ended in arguments for t= he unique world-historical mission of the German nation, an autarkic closed= state, and the exclusion of those who didn’t fit the national commun= ity. The lesson isn’t just “be careful.” It’s that = the three pillars need to travel together. Autonomy without truth-seeking b= ecomes self-righteousness. Autonomy without decentralization becomes domina= tion. What checks national self-determination is the same thing that checks= individual self-determination: openness to correction and the refusal to c= oncentrate power beyond what can be held accountable.

And what happens when a community’s “form of life&#= 8221; involves suppressing the autonomy of its own citizens? The pillars ca= n come into tension here. Respect for national self-determination and respe= ct for individual autonomy pull in opposite directions.

Th= is is where Tocqueville matters most. The meaningful unit of self-governmen= t is rarely the nation-state on its own. It’s the dense layer of asso= ciations, communities, firms, religious groups, and local institutions that= sit between the individual and the state. Any serious thinking about inter= national AI governance has to make room for those middle layers. Tocquevill= e saw that democratic freedom doesn’t live in declarations from the c= enter, but in the practice of self-government at the local and associationa= l level.

<= span class=3D"mention-pnpTE1 mentionSta= tic-liMklk" data-component-name=3D"MentionStatic" style=3D"white-space: now= rap;cursor: pointer;color: #25384c;border-radius: 4px;transition: color .25= scubic-bezier(.19,1,.22,1),background-color .25scubic-bezier(.19,1,.22,1),b= ox-shadow .25scubic-bezier(.19,1,.22,1),border .25scubic-bezier(.19,1,.22,1= ),border-radius .25scubic-bezier(.19,1,.22,1),opacity .25scubic-bezier(.19,= 1,.22,1),filter .25scubic-bezier(.19,1,.22,1),stroke .25scubic-bezier(.19,1= ,.22,1),transform .25scubic-bezier(.19,1,.22,1),scale .25scubic-bezier(.19,= 1,.22,1),outline .25scubic-bezier(.19,1,.22,1);margin: 0 -.125em;padding: 0= .125em .0625em;text-decoration: none;">Mark Frazier asks:

Can you set up a p= ath for crowdfunding projects or contests to realize ideas that the Cosmos = Institute seeds?

Interesting. Not something weR= 17;ve considered, but we’ll think about whether there’s a model= that works for us.

George asks:

Do you plan to ha= ve online cohorts?

No plans right now, but we may = consider it in the future!

Sa= rthak D asks:

I s= ee all these wonderful essays and people doing great work. Honestly, I woul= d love to interact with the community + become part of it in some capacity.= Is there a channel where people who are interested in the ideas that Cosmo= s is working towards but not necessarily are academics or builders can comm= unicate with the fellows and the team?

Not right now, but we are thinking about whether there’s somethi= ng we can do here!

Kevin C= utright asks:

I&#= 8217;m persuaded by the concern about cognitive risks and the need for R= 20;AI for epistemics,” “deliberative AI,” etc. Do you kno= w of organizations developing benchmarks around the goal of bolstering crit= ical thinking and improving epistemic processes and outcomes?

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">We have some grant projects that have focused on this. Tw= o that come to mind are DeliberationBench, which assesses AI persuas= ion in comparison with diverse human discussion, and Priori, a tool = that surfaces hidden assumptions when you are interacting with an AI model.= Two of our grantees (Steven Molotnikov and Cathy Fang) are r= unning a research study on how Priori and related human oversight interface= s work in practice.

I think there is a wave of= energy in this area. Various orgs are thinking more about AI for Human Rea= soning (with Future of Life Foundation funding work in this area, Fo= rethought writing about it, and Elicit working on directly in= the for-profit space). Also anecdotally I hear researchers thinking more a= bout ideas like “epistemic security” or “cognitive securi= ty” or “cognitive sovereignty” as well as ways to improve= information environments without restricting speech and expression.=

I share your enthusiasm for more work in this area –= both on benchmarking but also technology that better enables open contesta= tion of ideas (inspired by classical liberal premises, and Mill’s ide= as on this). If readers are working on this please do reach out!

C= osmos Institute is the Academy for Philosopher-Builders, technolo= gists building AI for human flourishing. We run fellowships, fund AI protot= ypes, and host seminars with institutions like Oxford, Aspen Institute, and= Liberty Fund.

Upgrade to paid

 
3D""Like
3D"=Comme= nt
3D""Restack
 

= 9; 2026 Cosmos Institute
500 W 2nd Street, Suite 1900, Aust= in, TX 78701
Unsubscribe

=3D"Get3D"Start=

3D""3D"" --a5c2ad23665ebd8657b8089bca91d63d4c4b21bf3aaef79b9915a6738176--