What the function is
The Foundation provides considered analysis of cognitive automation’s consequences to political, religious, educational, and civil society institutions. The function is operational from launch in limited form rather than deferred; the disruption begins immediately, and political institutions trying to make sense of it will need analysis from credible sources from day one.
The Foundation has a perspective on the transition that almost no other actor will have: direct visibility into how the technology is being used, technical understanding of capability and limits, values orientation distinguishing it from commercial actors, and credibility from public-domain release with anti-capture provisions. This perspective is the function’s substantive contribution. The function’s operational form is the work of producing analysis that bears the perspective into venues where institutions making decisions can use it.
Five structural commitments
The advisory function operates under five structural commitments, stated publicly and reviewed by the board annually. The commitments protect the function from political capture by establishing the conditions under which the analysis the function produces is legible as analysis rather than as advocacy.
Public, not private. Analysis goes into the public record. It does not go into private briefings that allow influence without accountability. A briefing whose contents are not visible to the public, given to a decision-maker whose decision will affect the public, is the structural form that captured advisory functions take. The Foundation does not produce that form. Analysis the Foundation produces is published — on the Foundation’s site, in published reports, in venues where the analysis is independently citable by anyone who encounters it.
Analytical, not advocacy-based. The Foundation describes consequences and trade-offs. It does not advocate for particular policies. The line is concrete: “concentrated power produces bad outcomes systematically” is the kind of structural claim the Foundation makes; “Congress should pass bill X” is the kind of claim it does not. The first is a generalizable analytical observation; the second is a policy recommendation. The Foundation produces the first kind because the first kind preserves the analysis’s usefulness across changes in the political landscape. The second kind would tie the Foundation’s credibility to the fortunes of specific political coalitions, which is exactly the capture the function is designed to resist.
Bounded by mission scope. The Foundation has a perspective on cognitive automation, knowledge infrastructure, and frameworks. It does not extend to areas where its expertise does not apply. A mission scope that includes “AI’s effects on knowledge work” does not authorize the Foundation to opine on monetary policy, military strategy, or global health. The bound is structurally important because expansion beyond expertise is how advisory functions become political actors with opinions on everything; the discipline of staying within scope is what makes the analysis the Foundation does produce hold up.
Mission-loyal staffing. Staff whose primary loyalty is to the Foundation, not to political careers. The advisory function is not a stepping stone to government positions; that pattern produces capture. People who join the function do so because they believe in the mission and intend to do the work for the duration of their commitment, not because they see the function as a credential for their next political role.
Modest in resources. A small high-quality analytical capacity is worth more than a large well-funded one. The function is not a think tank competing with the Brookings Institutions and AEIs of the field; it is a small focused capacity producing analysis on the specific question the Foundation has expertise in. Smallness is a feature: it constrains scope, prevents drift toward generalist political analysis, and makes the function affordable on the Foundation’s modest revenue model.
Why operational from launch
The advisory function is operational from launch in limited form rather than deferred until the Foundation’s incorporation. The reason is structural.
The disruption begins immediately. Political institutions trying to make sense of it will need analysis from credible sources from day one. The labs will be producing analysis from their own perspective, with their own incentives shaping what they say and how they say it. Industry-funded think tanks will be producing analysis with funder-shaped framing. Advocacy organizations will be producing analysis with advocacy-shaped framing. A perspective grounded in public-domain release, with anti-capture provisions, with technical depth, with values orientation distinct from commercial actors, will not exist by default; it has to be produced deliberately.
Deferring the function for years means the Foundation is silent during the period when its analysis is most needed and other actors are filling the analytical space with their own framings. The framings that establish themselves first become the defaults; the framings that arrive later have to dislodge defaults already in place. Pre-incorporation operation is the Foundation’s commitment to not be late.
Pre-incorporation, the function operates under the founder’s name with the same structural commitments that will eventually be encoded in Foundation governance. The analysis is published on the Foundation’s site or under the founder’s byline at venues where the analysis is independently citable. The five structural commitments apply to the pre-incorporation work; they do not depend on the Foundation having corporate existence to be operative.
Post-incorporation, the function transitions to Foundation operation with formal staffing as funding permits. The transition does not change the substance of the work; it changes the institutional locus and the staffing model. The same five commitments continue to govern.
What the function produces
The function’s primary output is published analytical material. The form varies; the substance is consistent. Published reports on specific aspects of the cognitive transition. Analytical essays on questions of public consequence. Position papers (analytical, not advocacy-based) on questions where the Foundation has perspective worth contributing. Cross-references to other analytical work where another organization has produced something the Foundation thinks is important to amplify.
Two specific framings appear in the published analysis with deliberate frequency.
The Hari Seldon framing works for technically literate readers, policy circles, and those who read serious science fiction. It frames the work as a considered response to anticipated civilizational stress: psychohistory looking ahead, building an institution that addresses what’s coming rather than what’s currently visible. Government officials who read science fiction get this framing immediately. The framing’s analytical posture is that the cognitive transition has shape and that the shape is partially predictable, and that institutions that prepare for the predictable parts produce better outcomes than institutions that wait to be surprised.
The Lord of the Rings framing works for broader audiences. The corruption of Frodo, the genuinely good-hearted hero who carries the ring as far as he can but ultimately succumbs at the very end, is the apt metaphor for what concentrated AI capability does to those who hold it. The point of public-domain release is that no single holder has the power to be corrupted by it. The framing’s analytical posture is that concentrated power produces systematic distortion of the people who hold it, regardless of the holders’ intentions, and that the structural answer is distribution rather than oversight.
Both framings should appear in published materials. The Hari Seldon framing in analytical documents and the long-form essay; the ring-of-power framing in more accessible communications. Together they signal both depth of consideration and moral seriousness behind the public-domain decision.
The published analysis of long-term effects (the “psychohistory document”) is live before government officials and journalists begin contacting the founder. This means the launch sequence includes the analytical material being available before the inevitable contact begins. By the time officials and journalists reach out, the foundational analysis is already in the public record and can be referenced; the Foundation does not have to produce the analysis under deadline pressure.
Audience
The function’s audience is institutional rather than individual. Political institutions of all alignments. Religious institutions. Educational institutions. Civil society organizations. Government agencies. Disability advocacy organizations. Labor organizations. Anyone who needs analysis to think clearly about how cognitive automation is reshaping their domain.
The Foundation does not curate its audience. It provides analysis to anyone who requests it, on the same terms. A request from a conservative think tank gets the same analysis a request from a progressive advocacy organization gets. A request from a federal agency gets the same analysis a state government gets. The analysis is the analysis; the Foundation does not tailor the content to the audience’s political alignment, because tailoring is exactly the kind of capture the structural commitments rule out.
The non-curation matters because the cognitive transition affects every institution roughly the same way structurally, even if the institutions are politically opposed. A conservative think tank trying to understand the cognitive transition needs the same analytical work a progressive advocacy organization needs. Both will use the analysis differently — the conservative think tank will fold it into its own analytical outputs that recommend conservative policy responses; the progressive advocacy organization will fold it into outputs that recommend progressive policy responses. The Foundation is upstream of those policy responses; the Foundation’s job is to make the structural analysis available so the downstream policy work has solid ground to stand on.
The Foundation does engage with media. The analysis is meant to be read by journalists and to inform journalism on the cognitive transition. The engagement follows the same structural commitments: analysis goes into the public record (a published report a journalist quotes is structurally different from a private briefing a journalist incorporates without attribution), the analysis is analytical not advocacy-based, the analysis is bounded by mission scope. The engagement is meant to make journalism on the transition better-informed, not to use journalism as a delivery vehicle for Foundation positions.
What the function does not do
It does not produce position papers on policy specifics. The Foundation does not say “Congress should pass bill X” or “the EU AI Act should be revised in such-and-such direction” or “the FTC should investigate company Y.” Policy recommendations are advocacy, not analysis. The Foundation produces analysis that makes policy work easier; it does not do the policy work itself.
It does not endorse parties, candidates, political alignments, or coalitions. The Foundation does not have a partisan posture. The analytical work is structurally calibrated to be useful across alignments because the alignments will use the analysis differently and the Foundation’s job is to make the analysis itself solid.
It does not do private briefings. A meeting with a decision-maker that produces no public artifact is structurally inappropriate to the function. If the Foundation has a perspective worth a decision-maker hearing, the perspective is published; the decision-maker is welcome to read the publication and to be in conversation with the Foundation about it; what the Foundation does not do is produce content that exists only in the decision-maker’s hearing of it.
It does not lobby. The Foundation does not have a government-affairs department; it does not have lobbyists; it does not engage in the activities that distinguish lobbying from analytical work. The line between policy analysis and policy advocacy is not always sharp, but the line is real, and the Foundation operates on the analytical side of it.
It does not advise on questions outside its expertise. A request for analysis on cognitive automation gets analysis on cognitive automation. A request for analysis on monetary policy, on military strategy, on global health, on any domain where the Foundation does not have demonstrated expertise gets a polite redirect to organizations whose mission covers the domain.
What the function is for, ultimately
The advisory function exists because the cognitive transition will produce decisions that affect populations who will not be in the room when the decisions are made. School districts will decide what to do about AI in classrooms. State legislatures will decide what to do about AI in employment. Federal agencies will decide what to do about AI in healthcare and finance and criminal justice. International bodies will decide what to do about AI in cross-border data flows and military applications. These decisions will be made; the question is what analytical work will be available to the people making them.
The Foundation’s posture is that public-interest analysis — analysis whose authorship is not for hire, whose framing is not aligned with commercial actors, whose substance is not shaped by funder incentives — is the kind of analytical work decisions of this consequence deserve. The function is the operational form of that commitment. The function is small, modest, focused, and durable across decades. It is one specific contribution to a transition that will produce many decisions; it is not the totality of public-interest analytical work; it is the Foundation’s specific role in a larger ecosystem of organizations whose missions overlap with parts of the work.
The summary
The advisory function provides considered analysis of cognitive automation’s consequences to institutions making decisions that affect populations who will not be in the room. Five structural commitments — public not private, analytical not advocacy, mission-bounded, mission-loyal staffing, modest resources — protect the function from political capture and make the analysis legible as analysis. The function is operational from launch in limited form because the disruption begins immediately and the analytical space gets filled by other actors if the Foundation is silent. The function does not endorse parties, recommend policies, conduct private briefings, lobby, or advise outside its expertise. What it does is produce published analysis grounded in the Foundation’s specific perspective — public-domain release, anti-capture commitments, technical depth, values orientation distinct from commercial actors — for any institution that needs the analysis. That perspective is the substantive contribution; the operational discipline is what makes the perspective land where it can do good.