No rules match that filter.
The rules apply to all presenters and all participating researchers. GEA is bound by the rules in the same way external presenters are, including the grounding criteria and the review process.
Rule 1
Participation Requirements
Statement connection. Every presentation must be explicitly tied to at least one statement from the event's statement pool. If the presentation bears on multiple statements, the presenter must identify a primary statement and name any additional statements.
Position on each statement. For each statement addressed, the presenter must state their position: supporting the statement, opposing the statement, or approaching the statement orthogonally.
Explicit connection in the presentation. The statement connection and position must be stated clearly at the opening of the presentation itself. It is not sufficient to state the connection only in pre-event materials.
Acceptance of peer review. All presentations are subject to evaluation by the Independent Review Board. Participation in the event constitutes acceptance of the review process. Presenters may not opt out of review as a condition of speaking.
Rule 2
Submissions and Materials
Presenters must submit the following materials to the event coordinator by the deadlines specified in the coordinator's invitation materials.
Profile photo. A high-resolution photograph suitable for marketing. Headshot preferred.
Bio. A biographical statement of up to 200 words.
Presentation title. A working title is acceptable for marketing purposes. The final title may be updated in coordination with the event coordinator.
Marketing one-liner. Completion of the sentence "This Is AGI: ___" in the presenter's own words. More than one completion is welcome. These are used in pre-event social media.
Rule 3
Presentation Conduct
Length. Presentations are 10 to 15 minutes in length unless otherwise agreed in writing with the event coordinator.
Audience challenge follows. Each presentation is followed by live audience challenge under the Scientific Code of Conduct defined in Rule 12. Challenge time is allocated at roughly parity with presentation time unless otherwise announced.
Closing panel. Participation in the closing panel is optional. Presenters who wish to participate should indicate so when submitting materials.
Remote participation. In-person presentation is strongly preferred. Remote presentation may be arranged in advance with the event coordinator when in-person participation is not possible.
Conceptual focus. Presentations should focus on concepts, pathways, and grounding rather than deep technical detail. Presentations heavy on engineering specifics without scientific grounding do not review well. See Rule 4 for the criteria the board applies.
No marketing content. Presentations may not be used for commercial promotion of products, services, or organizations. Affiliation statements and institutional attribution are permitted and appropriate; sales content is not.
Rule 4
Presentation Tracks
Every presentation is classified under one of the tracks defined in this Rule. The track determines how the Independent Review Board evaluates the presentation and how the presentation appears in the City AGI Report.
Track classification
Single primary track. Every presentation is classified under a single primary track. If the presentation spans multiple tracks, the primary track is the one best reflecting the presentation's principal contribution.
Cross-references. If the presentation bears on tracks beyond its primary track, the additional tracks are noted as cross-references. The review board applies the criteria of the primary track to the presentation as a whole, and applies the criteria of any cross-referenced tracks to the specific claims within the presentation that fall in those tracks.
Track classification is finalized before the event. Presenters propose a track classification under Rule 2.6. GEA and the event coordinator review classifications and confirm them before the event. Disputed classifications are resolved by GEA in consultation with the review board chair.
The five sub-criteria. The review board evaluates every presentation against five sub-criteria of scientific grounding. The sub-criteria apply to all tracks, with adaptations as defined in Rules 4.3 through 4.11.
Foundations identified. The presentation clearly identifies the accepted science — or, in non-science tracks, the established body of thought — it builds on.
Foundations engaged correctly. The presentation uses its foundations accurately. Misrepresentation or misapplication of the cited foundations fails this criterion even if the novel claims built on them are interesting.
Extensions acknowledged. The presentation is clear about where its foundations end and its novel claims begin. Conflating speculation with established findings fails this criterion on grounds of intellectual honesty.
Coherent reasoning. The path from foundations to novel claims is internally consistent. The conclusion follows from what precedes it.
Falsifiable in principle. The novel claims are stated in a form that could be shown wrong by empirical observation, logical counter-argument, or demonstrable counter-example. Where current technology or available evidence does not permit falsification today, the presentation must describe the specific steps by which the claim could be falsified in the future — what observation, experiment, demonstration, or counter-argument would refute it. Claims that cannot fail under any conceivable conditions are not evaluated as scientific claims.
Scientific Frameworks — Mature Research.
Presentations proposing or defending a scientifically grounded framework for intelligence, cognition, or a path to AGI, with substantial prior development: articulated theoretical structure, extended body of work, published treatment, or demonstrated implementations.
Examples of work in this track:
- Theoretical frameworks with extensive prior publication and development
- Cognitive architectures with implemented systems and published literature
- Empirically grounded research programs in biology, neuroscience, or cognitive science applied to intelligence
- Mathematical frameworks that are explicitly grounded in empirical science and make falsifiable predictions about physical or biological systems
Foundations identified: explicit citation of the accepted scientific foundations the framework anchors to (physics, thermodynamics, causality, neuroscience, computational theory, or other established fields).
Foundations engaged correctly: the framework uses its foundations accurately. The board may consult specialist reviewers for the relevant discipline where needed.
Extensions acknowledged: clear distinction between established science and the framework's novel extensions.
Coherent reasoning: reasoning from foundations to claims is internally consistent and follows logically.
Falsifiable in principle: the framework makes predictions or claims that could be shown wrong. Where testing awaits technology that does not yet exist, the framework must specify what observations or demonstrations would falsify it when the technology becomes available.
Scientific Frameworks — Ideas (Early-Stage).
Early-stage proposals, new hypotheses, thought experiments, or preliminary frameworks. Less developed than mature research but scientifically serious. The distinction between this track and Rule 4.3 is development stage, not scientific seriousness.
Examples of work in this track:
- Novel hypotheses about the nature of intelligence with initial theoretical development
- Preliminary frameworks that identify a research program but have not yet produced extensive published work
- Thought experiments proposing new approaches and arguing for their scientific grounding
- Extensions of existing frameworks into new domains
Foundations identified: applied strictly. Early-stage work must still anchor to accepted science.
Foundations engaged correctly: applied strictly. Misuse of foundations is not excused by early development stage.
Extensions acknowledged: particularly central for this track. Most of the work is extension beyond accepted science; acknowledging this clearly and honestly matters more, not less, at the early stage.
Coherent reasoning: applied to the reasoning as far as it has been developed. The framework need not be fully worked out, but what has been developed must hold together.
Falsifiable in principle: applied strictly. The proposal must be stated in a form that could in principle be shown wrong. Where testing is deferred, the falsification path must be specified.
Meta-Analysis.
Presentations making claims about the state of a field — for example, that mainstream AGI research is scienceless, that a given approach conflates engineering with science, or that the field lacks a falsifiable theory. These are arguments about the field rather than scientific frameworks themselves.
Examples of work in this track:
- Arguments that a field or approach lacks scientific foundation
- Critiques of research programs for methodological or theoretical gaps
- Claims about what is missing from the current state of a discipline
- Historical or sociological analyses of how a field has developed
Claim clearly defined: what is being claimed about the field, and what would count as evidence for or against it. Vague generalizations fail this criterion.
Evidence substantial: specific cited work, published claims, or concrete examples support the meta-claim. Generalizations without specifics fail.
Counter-positions engaged: a rigorous meta-claim acknowledges the best case for the opposite view. Straw-man versions of the field's defenders fail this criterion.
Reasoning coherent: the cited evidence actually supports the claim. The argument from evidence to conclusion follows logically.
Falsifiable in principle: what observation, development, or discovery would change the claim. Where the meta-claim depends on future developments, the falsification path must be specified. Meta-claims that cannot fail under any conditions are not evaluated as scientific claims.
Economics.
Presentations on the economic implications of AGI and ASI — labor markets, capital, production, distribution, the future of capitalism, or proposals for alternative economic arrangements under AGI conditions.
Examples of work in this track:
- Analyses of AGI's impact on employment, wages, or labor markets
- Arguments about the viability of capitalism under AGI or ASI conditions
- Proposals for new economic frameworks adapted to post-AGI conditions
- Empirical or theoretical claims about economic trajectories under AGI development
Foundations identified: the economic theory or empirical base the presentation builds on is explicitly named — classical, neoclassical, behavioral, institutional, heterodox, or specific empirical literatures.
Foundations engaged correctly: theory and data are represented accurately. Misrepresentation of economic frameworks or misuse of empirical findings fails this criterion.
Extensions acknowledged: clear distinction between established economics and the novel extensions into AGI-era territory.
Reasoning coherent: arguments from foundations to claims about AGI-era economies follow logically, with honest treatment of uncertainty and edge cases.
Falsifiable in principle: empirical claims are stated in testable form. Theoretical claims are stated such that coherent counter-argument could refute them. Where testing is deferred, the falsification path must be specified.
Governance.
Presentations on how AGI and ASI affect political structures, democratic institutions, state capacity, international coordination, or proposals for new governance models under AGI conditions.
Examples of work in this track:
- Analyses of AGI's impact on democratic institutions or state power
- Proposals for new governance frameworks adapted to AGI-era conditions
- Arguments about structural incompatibility between AGI and particular governance forms
- International coordination or regulatory proposals
Foundations identified: the political theory, institutional analysis literature, or historical precedents the presentation builds on are explicitly named.
Foundations engaged correctly: theoretical and historical sources are represented accurately.
Extensions acknowledged: clear distinction between what established theory or precedent supports and what the presentation proposes beyond it.
Reasoning coherent: arguments from foundations to conclusions about AGI-era governance hold together, with honest treatment of institutional dynamics and counter-cases.
Falsifiable in principle: claims are stated such that historical precedent, logical counter-argument, or empirical development could refute them. Where testing is deferred, the falsification path must be specified.
Law.
Presentations on legal doctrine, liability, regulation, rights, and other legal questions raised by AGI development and deployment.
Examples of work in this track:
- Analyses of existing legal doctrine applied to AGI systems
- Proposals for new legal frameworks, liability regimes, or regulatory structures
- Arguments about constitutional, statutory, or common-law implications of AGI
- Comparative or international legal analysis of AGI governance
Foundations identified: the doctrine, precedent, statute, or legal-theoretical tradition the presentation builds on is explicitly named.
Foundations engaged correctly: doctrine and precedent are represented accurately. Misstatement of the law fails this criterion.
Extensions acknowledged: clear distinction between existing law and the presentation's proposed extensions, amendments, or new frameworks.
Reasoning coherent: legal reasoning is internally consistent. Arguments applying doctrine to novel AGI scenarios contain no logical gaps.
Falsifiable in principle: claims about what the law does or should do are stated such that legal counter-argument, precedent, or empirical legal observation could refute them.
Philosophy.
Presentations on philosophical questions raised by AGI — consciousness, moral status, free will, personal identity, epistemology, and related topics.
Examples of work in this track:
- Analyses of consciousness in AGI systems
- Arguments about the moral status of AGI or ASI entities
- Philosophical critiques of AGI claims (for example, about understanding, agency, or selfhood)
- Epistemological analyses of what counts as intelligence, knowledge, or comprehension
Foundations identified: the philosophical tradition, literature, or body of prior argument the presentation builds on is explicitly named.
Foundations engaged correctly: philosophical sources and existing arguments are represented accurately.
Extensions acknowledged: clear distinction between the existing philosophical literature and the presentation's novel contribution.
Reasoning coherent: the argument is logically valid. Premises and conclusions are clearly related.
Falsifiable in principle: claims are stated such that logical counter-argument, counter-example, or empirical development (where relevant) could refute them. Purely definitional or tautological claims must be labeled as such.
Ethics, Morals, and Safety.
Presentations on AGI safety, alignment, ethical implications of AGI development, moral status of AGI systems, existential risk, and related normative and practical concerns. Presentations in this track commonly rest on empirical claims about AGI systems, philosophical claims about values or moral status, and normative claims about what should be done. The presentation must be clear about which kind of claim is being made at each step.
Examples of work in this track:
- AI safety and alignment research and arguments
- Ethical analyses of AGI development trajectories
- Existential risk arguments concerning AGI or ASI
- Proposals for ethical constraints, safety measures, or development norms
Foundations identified: the ethics, safety, or risk-analysis literature the presentation builds on is explicitly named. Empirical claims about AGI systems identify their scientific foundations separately.
Foundations engaged correctly: literature is represented accurately. Empirical claims are grounded in work that itself meets scientific grounding standards.
Extensions acknowledged: clear distinction between established argument and novel contribution. The distinction between empirical, normative, and philosophical claims is made explicit.
Reasoning coherent: the argument from premises to conclusions is logically valid. The strongest counter-arguments are engaged.
Falsifiable in principle: empirical claims are stated in testable form. Normative claims are stated with premises explicit enough that disagreement can be located. Unfalsifiable claims must be labeled as such.
Visions.
Presentations offering coherent speculative scenarios for an AGI-era future — applied proposals built on grounded theoretical foundations for what society, economics, governance, or human life could look like when AGI or ASI conditions obtain. Visions are explicitly proposals about possibility, not predictions about likelihood. This track is for applied work that extends grounded theoretical foundations into speculative scenarios. It is not a track for free-floating speculation.
Examples of work in this track:
- Applied proposals for post-AGI civilizational design built on grounded theoretical foundations
- Scenarios that take specific theoretical frameworks and extend them into coherent future pictures
- Constructive visions of AGI-era transitions with explicit grounding in existing research
Foundations identified: the theoretical foundation(s) the vision applies must be named explicitly, and those foundations must themselves meet the grounding standard in their native track. A vision grounded in nothing does not qualify.
Foundations engaged correctly: the theoretical foundation is represented accurately. Claims in other domains (economics, governance, science) within the vision are evaluated against the criteria of those domains.
Extensions acknowledged: the work is clearly labeled as a vision, not a prediction or empirical claim. Explicit distinction between what the foundational theory supports and what the vision proposes as application.
Internal consistency: the imagined future hangs together as a coherent scenario. Contradictions between elements of the vision fail this criterion.
Falsifiable in principle: the vision's claims about possibility are stated such that coherent counter-argument could refute them. A vision that asserts a future is possible must be clear about what conditions would make it impossible.
Rule 5
The Independent Review Board
Role. The Independent Review Board evaluates every presentation against the criteria defined in Rule 4, applied according to the presentation's track, and produces structured evaluations that feed into the City AGI Report.
Composition. The core board consists of independent researchers from scientific disciplines relevant to the event, including but not limited to physics, biology, neuroscience, cognitive science, complexity science, and information theory. Philosophy of science is included for its role in evaluating the scientific status of claims. Specialist reviewers are added for tracks outside the core scientific disciplines — economists for the economics track, legal scholars for the law track, philosophers for the philosophy track, and so on.
AI researcher limit. At most two members of the core board may be AI researchers. The remaining members must be scientists from other disciplines. This composition protects the focus on underlying science rather than industry-specific framing.
Independence from GEA. No presenters, members of presenters' organizations, or members of GEA may sit on the board. This prohibition applies regardless of capacity.
Conflict rules. Board members who have publicly taken strong positions on a specific event statement must recuse from evaluating presentations addressing that statement. Other conflicts — financial, personal, or institutional — are disclosed and handled by the board chair.
Term and rotation. Board membership is per event, per city, or longer as specified at time of appointment. No member serves indefinitely without review. Rotation protects against the board's drift toward mainstream gatekeeping.
Rule 6
Evaluation Process
The board's evaluation operates at two levels. Level one evaluates each presentation individually for scientific grounding under Rule 4. Level two evaluates each of the event's statements in light of all grounded contributions addressing it. Only grounded contributions feed into statement-level evaluation.
Level one — per-presentation grounding evaluation
Per-presentation evaluation. For each presentation, the board evaluates against the five sub-criteria as adapted for the presentation's track under Rule 4. The board produces, for each presentation, a structured record that includes: the scientific or disciplinary foundations the presentation anchors to, the novel extensions beyond those foundations, the internal coherence of the reasoning, the falsifiability assessment, and any serious concerns.
Grounding verdict categories. Each presentation receives one of three grounding verdicts:
Meets the grounding standard. The presentation satisfies all five sub-criteria as adapted for its track. The work is scientifically serious. It qualifies as input to the statement-level evaluation for the statements it addresses.
Partially meets the standard. The presentation satisfies most sub-criteria but has clearly identified gaps. It qualifies as input to the statement-level evaluation with the gaps noted, so that the statement-level analysis can weigh the contribution appropriately.
Does not meet the standard. The presentation fails one or more sub-criteria substantially. It does not qualify as input to the statement-level evaluation. It is noted in the Report with the board's reasoning so that readers understand what was presented and why it did not meet the grounding bar.
Pro-ticket submissions. Pro-ticket submissions that clear the Scientific Code of Conduct under Rule 11 are evaluated at level one against the five sub-criteria adapted for commentary. Submissions that meet or partially meet the standard qualify as input to statement-level evaluation. Submissions that do not meet the standard are published in the Report with the board's reasoning but do not feed into statement-level verdicts.
Level two — per-statement scientific state evaluation
Unit of evaluation. For each statement addressed at the event, the board evaluates the scientific state of the statement given all grounded presentations and grounded pro-ticket commentary addressing it. The statement is the unit of evaluation; the grounded contributions are inputs.
Comparative analysis required. Where two or more grounded contributions bear on the same statement, the board evaluates them comparatively. Comparative analysis identifies the nature of any disagreement (empirical, definitional, methodological, or foundational), assesses how the disagreement is grounded on each side, and states what would be required to resolve it.
Scientific state verdict categories. Each statement receives one of five scientific state verdicts:
Scientifically supported. Grounded contributions support the statement and no grounded opposition was presented. The statement has grounded scientific support as addressed at this event.
Scientifically opposed. Grounded contributions oppose the statement and no grounded support was presented. The statement has grounded scientific opposition as addressed at this event.
Scientifically contested. Grounded contributions are presented on both sides. The statement is a live scientific disagreement among grounded positions. The board's comparative analysis describes the nature of the disagreement and what would resolve it.
Scientifically unsupported at this event. No contribution addressing the statement met the grounding standard, or no contribution addressed the statement at all. The statement did not receive grounded scientific treatment at this event.
Outside scientific reach. The grounded contributions presented and the arguments offered indicate that the statement cannot be evaluated by scientific means — either because the relevant science does not exist, cannot exist, or because the statement is not framed in a form that scientific evaluation can address.
Statement verdicts are not verdicts on truth. Statement-level verdicts describe the scientific state of the statement as addressed at this event. They do not declare the statement true or false in any ultimate sense. A statement judged opposed at this event may still be true; the opposition simply prevailed among grounded contributions. A statement judged supported may still be wrong; later events may surface opposition that was not presented at this one.
Procedural rules
No adjudication between grounded approaches. The board does not declare which path to AGI is correct or rank grounded approaches. The grounding verdict filters for scientific seriousness; the statement verdict reports the scientific state of the statement as addressed at this event. Neither verdict settles the scientific question of AGI itself.
Evaluations are final. Subject to the right of response under Rule 8 and Rule 11.13, the board's evaluations are not subject to revision or override by GEA or by presenters.
Rule 7
Editorial Role of GEA
GEA does not evaluate presentations. GEA does not participate in the Independent Review Board and has no input into the board's evaluations. See Rule 5.4.
GEA serves as editor of the Reports. GEA is responsible for compiling and publishing the City AGI Report and the Annual AGI Report. The editorial role is procedural, not substantive.
Procedural scope. GEA's editorial authority consists of enforcing the rules stated in this document — including scope limits on presenter responses under Rule 8, word limits, formatting, and structural organization of the Reports. GEA does not evaluate whether a response is correct, whether its arguments are sound, or whether it successfully rebuts the board.
Limit on editorial authority. GEA's editorial authority applies to presenter responses and to Report structure, not to the board's evaluations. Board evaluations appear in the Report as the board produced them. GEA may not rewrite, soften, or strengthen board verdicts, including verdicts applied to GEA's own presentations.
GEA's own framework. The Reports include a clearly labeled section for GEA's own theoretical framework. GEA's presentations are evaluated by the board on the same criteria as external presentations. The distinction between GEA's framework and the neutral synthesis sections of the Report is made explicit so readers can trust the framing.
Event-time editorial functions. GEA's editorial role extends to event-time procedural functions, including the audience moderator and live curator roles defined in Rule 10 and the submissions editor role defined in Rule 11. These functions apply the rules stated in this document; they do not evaluate scientific substance.
Analogy. The separation between GEA's editorial role and the board's substantive role is analogous to the distinction between editors and peer reviewers at academic journals. Peer reviewers judge substance; editors enforce submission rules. The same distinction applies here.
Rule 8
Right of Response
The right. Each presenter may submit a written response to the board's evaluation of their presentation. The response is published in the City AGI Report alongside the board's evaluation.
Final word. The presenter's response is the final word on the presentation in the Report. The review board does not respond to the presenter's response. This rule creates a clean endpoint to the evaluation.
Scope of the response. Responses are limited to the following:
addressing specific criticisms raised in the board's review;
correcting factual or interpretive errors in the board's review;
providing clarification on points the board found unclear;
stating agreement, disagreement, or nuanced positions on the board's findings.
What a response may not contain. Responses must not introduce new content that was not part of the original presentation. A response is not a second presentation, a revised version of the presentation, or a vehicle for commentary unrelated to the board's review. Responses are also subject to the Scientific Code of Conduct defined in Rule 12.
Length limit. Responses are limited to 1,000 words.
Editorial review for scope. GEA, acting as editor under Rule 7, reviews each submitted response against the scope defined in Rule 8.3 and Rule 8.4. If a response falls outside that scope, GEA notifies the presenter and provides an opportunity to revise. Presenters have multiple opportunities to bring the response within scope, up to the lockdown date for the City AGI Report.
Presenter's alternative publication. A presenter whose response is not published in the Report under Rule 8.7 is free to publish the response through their own channels, including social media, personal websites, or other venues. This Rule does not restrict the presenter's speech; it defines only the conditions for inclusion in the Report.
No response required. Submitting a response is optional. A presenter may decline to respond, in which case the presentation section in the Report contains only the presentation summary and the board's evaluation.
Rule 9
The City AGI Report and Annual AGI Report
Purpose. The Reports provide a grounded scientific assessment of a set of claims about AGI, for journalists, policymakers, researchers, investors, and the public. Each statement addressed at an event receives a structured verdict on its scientific state given the grounded work presented. The Reports surface where grounded scientific support lies, where grounded opposition lies, where live scientific disagreement exists, and where grounded work is absent.
No adjudication on truth. The Reports do not declare which statements about AGI are ultimately true or false. The grounding standard defined in Rule 4 filters for scientific seriousness; the scientific state verdicts defined in Rule 6.2 report the state of grounded work on each statement. Neither adjudicates truth. See Rule 6.2.d and Rule 6.3.a.
City AGI Report — statement-organized structure. A City AGI Report is published for each event in the series. The City Report is organized around the statements addressed at the event. For each statement, the Report contains:
The statement itself, with context on why it matters;
The scientific state verdict under Rule 6.2.c (supported, opposed, contested, unsupported at this event, or outside scientific reach);
Summaries and grounding verdicts of grounded presentations supporting the statement;
Summaries and grounding verdicts of grounded presentations opposing the statement;
Summaries and grounding verdicts of grounded presentations addressing the statement orthogonally;
Summaries of presentations addressing the statement that did not meet the grounding standard, with the board's reasoning;
Pro-ticket commentary addressing the statement, with the board's evaluation of each submission;
Presenter and submitter responses to board evaluations under Rule 8 and Rule 11.13;
The board's comparative analysis of the grounded contributions and, where the statement is contested, a description of what would resolve the disagreement;
Open questions that remain on the statement.
Meta-analyses and visions. Presentations in the Meta-Analysis track (Rule 4.5) and the Visions track (Rule 4.11) may bear on multiple statements or on the state of the field rather than on a single statement. Where such presentations have a primary statement, they appear in that statement's section. Where they address the field or a question broader than any single statement, they appear in dedicated sections of the Report outside the statement-organized body.
GEA's framework section. Each Report contains a clearly labeled section presenting GEA's own theoretical framework as a whole, distinct from the statement-organized body of the Report. Individual GEA presentations are evaluated and appear within their relevant statement sections on the same terms as external presentations. The framework section is editorial presentation by GEA; readers can see the distinction between GEA's positions and the Report's neutral statement-level analysis.
Annual AGI Report. The Annual AGI Report synthesizes findings across all City Reports published during the year. It is also organized around the statements — tracking how the scientific state of each statement evolved across events, which disagreements recurred, which were resolved by further grounded work, and which statements were addressed for the first time. The Annual Report is the cumulative scientific landscape map; the City Reports are the individual events that build it.
Lockdown and publication. Each Report has a lockdown date after which presenter responses, pro-ticket submissions, and submitter responses are no longer accepted (see Rule 8.6, Rule 8.7, Rule 11.4, and Rule 11.13). After lockdown, the Report is finalized and published. Publication dates and distribution channels are specified in each city's coordinator materials and on the event series website.
Rule 10
Audience Participation
Submission channel. Audience members submit questions and challenges through a web-based application accessible on mobile devices during the event. This channel is the sole means by which audience questions reach the stage and the review board.
Attribution required. All audience submissions must be attributed. Anonymous submissions are not accepted. Transparency is a condition of participation: a participant who does not want their name associated with a question should not submit it.
Initial AI evaluation. Every submission is evaluated by an AI system against the Scientific Code of Conduct defined in Rule 12. The AI returns one of three classifications: clean (no code-of-conduct concern), red-flag (borderline, potential concern), or violation (clear breach of the code).
Moderator review. A designated audience moderator from the GEA event team reviews the full submission queue with AI classifications visible. The moderator may accept the AI's classification, override it in either direction, or downgrade or upgrade the severity. The moderator's determination is final during the event. No appeal is available during the event itself.
Three outcomes. Based on moderator determination, each submission receives one of three outcomes:
Clean. The submission is posted to the queue visible to the presenter, the review board, and other audience members, and is eligible for live curation and inclusion in the on-stage Q&A.
Red-flag warning. The submission is posted to the queue with a visible warning indicating the specific code-of-conduct concern. The warning is visible to all participants — presenter, review board, and audience. Red-flagged submissions are eligible for live curation, subject to the moderator's discretion.
Not posted. Submissions in clear breach of the code of conduct are not posted to the queue and do not appear in the presenter-facing or audience-facing views. The submitter is not notified of the rejection.
Live curation. A designated live curator selects questions from the queue for on-stage Q&A during each presentation's challenge period. Selection criteria include relevance to the presentation, scientific seriousness of the challenge, breadth across audience viewpoints, and time available.
Review board access. All posted submissions — clean and red-flagged — are made available to the review board as input to its evaluation of the presentation. Submissions that were not posted are not provided to the board.
Not published in the Report. Audience submissions are not published in the City AGI Report or the Annual AGI Report. They are working material for the presenter, the board, and the event's internal record.
Presenter response is optional. Presenters may respond in writing to posted submissions after the event through a response channel maintained by GEA. Written responses are shared with the submitter and with the review board. Responding is optional; non-response is not a violation of the code of conduct and is not held against the presenter by the board.
Logging. GEA maintains an internal log of AI classifications and moderator decisions for every submission. The log is not public. A submitter who wishes to know the disposition of their own submission may request the record from GEA.
Rule 11
Pro-Ticket Submissions
Eligibility. Holders of a Pro Ticket for a given event may submit written analyses of one or more presentations from that event for possible inclusion in the City AGI Report.
Scope of a pro-ticket submission. A submission is a written analysis of the scientific content of one or more presentations from the event. Submissions may argue in support of a presentation, in opposition, or from an orthogonal angle. Submissions must engage with the presentation as given — its claims, its grounding, its reasoning, or its falsification path.
Length limit. Pro-ticket submissions are limited to 2,000 words.
Submission deadline. The deadline for pro-ticket submissions is set per event and communicated at ticket purchase and in the event coordinator materials. Submissions received after the deadline are not accepted.
Ownership and publication license. Upon submission, pro-ticket submissions become the property of GEA. Submitters grant GEA an irrevocable, perpetual, worldwide license to publish, reprint, excerpt, translate, and create derivative works from the submission in the City AGI Report, the Annual AGI Report, and related GEA materials. Submitters may not copyright, restrict, withdraw, or otherwise limit GEA's use of submissions after the submission deadline.
Attribution. Submissions are published under the submitter's name unless the submitter explicitly requests otherwise at the time of submission.
Editorial handling. Submissions that meet the Scientific Code of Conduct and the board's scientific review are published substantively unedited. GEA performs only copy-editing for clarity, typographical correction, and formatting to match Report standards. GEA does not alter the substance of submissions, including submissions critical of GEA's own research.
Stage one — Code of Conduct review. Submissions are evaluated by an AI system and a designated submissions editor from the GEA editorial team against the Scientific Code of Conduct. If a submission falls outside the code, the submitter receives specific feedback identifying the rule cited and the language that triggered the concern. The submitter may revise and resubmit at any time up to the submission deadline.
Consequence of failure at stage one. Submissions still outside the Scientific Code of Conduct at the submission deadline are not forwarded to the review board and are not included in the Report. The ticket purchase is not refunded. The submitter is free to publish the rejected submission through their own channels, including social media, personal websites, or other venues.
Stage two — Review Board evaluation. Submissions that clear the Code of Conduct review are forwarded to the Independent Review Board. The board evaluates each submission against the five sub-criteria of Rule 4, adapted for the task of commentary rather than primary research. Commentary must identify its foundations, engage with them accurately, acknowledge its extensions, reason coherently, and state its claims in a form that could in principle be shown wrong.
Verdict categories for pro-ticket submissions. Each submission receives one of three verdicts: meets the grounding standard, partially meets the standard, or does not meet the standard. All three verdicts result in publication of the submission in the City AGI Report. The board's verdict and reasoning are published alongside the submission.
Submitter's right of response. A pro-ticket submitter whose submission is evaluated by the review board has the right to respond to the board's evaluation under terms parallel to Rule 8. Submitter responses are limited to 1,000 words, must fall within the scope defined in Rule 8.3 and Rule 8.4, are subject to the Scientific Code of Conduct, and appear in the Report as the submitter's final word on the matter. The board does not respond to submitter responses.
Separate Pro-Ticket Submission Agreement. The legal terms governing ownership, license grant, warranties, and indemnification are set out in a separate Pro-Ticket Submission Agreement to which pro-ticket holders explicitly agree at the time of purchase. The terms in this Rule 11 summarize those provisions; in any conflict between this summary and the Pro-Ticket Submission Agreement, the Agreement governs.
No refunds. Pro-ticket purchases are non-refundable. This applies regardless of whether a submission is accepted, rejected at stage one, receives a particular verdict at stage two, or whether a submitter chooses not to submit at all.
Rule 12
Scientific Code of Conduct
Purpose and scope. The Scientific Code of Conduct is the floor standard for non-presenter participation in the event and the Reports. The Code applies to audience submissions under Rule 10, pro-ticket submissions under Rule 11, and presenter responses to the review board under Rule 8 and to audience challenge under Rule 1.5.
No commercial framing. Claims and challenges must be grounded in the scientific question at hand, not in commercial positioning, market comparisons, or commercial benchmarks. Comparing a presented claim to commercial technology for purposes of commercial benchmarking falls outside the code.
No profit-based arguments. Whether something can be profitable or whether there is a market for it is not a scientific objection to a claim about intelligence, cognition, or AGI. Profit motive and scientific soundness are separate questions.
No personal attacks. Attack ideas, concepts, reasoning, and claims. Do not attack persons, their character, their credentials, or their affiliations. Challenges to what was said are welcome; challenges to the person who said it are not.
In scope. Questions and submissions must be in scope, meaning they must engage with the presentation as given, its scientific claims, or the event's grounding criteria under Rule 4. Questions that go outside the presentation's subject matter or the event's scientific framing fall outside the code.
Scientific purpose. Questions and submissions must serve a scientific purpose: identifying gaps in reasoning, seeking clarification, proposing alternative interpretations of the same evidence, testing the falsification path, or extending the work's logic to new cases. Questions that do not serve one of these purposes fall outside the code.
Honest representation. Do not misrepresent what a presenter or cited researcher claimed. Where a question or submission depends on an interpretation of what was said, state it as an interpretation and invite correction. Straw-man questions, paraphrased in a form the presenter did not use, fall outside the code.
Engage with what was said. Respond to the presentation as given, not to a position the presenter did not take, to the presenter's broader body of work, or to the presenter's institutional affiliations. The unit of engagement is the presentation, not the presenter's career.
No authority-based arguments. Appeals to authority or consensus are not substitutes for argument. Statements such as "mainstream researchers disagree" or "no major lab has pursued this" do not establish a scientific claim. Who disagrees with a claim does not settle whether the claim is sound.
Evidence runs both ways. Lack of evidence to prove something is not evidence to disprove it. Claims need evidence in both directions. "Industry, academia, or government is secretly coordinating to suppress X" needs evidence. "Industry, academia, or government would never do X" also needs evidence. Specific documented actions by specific actors are in scope when supported by evidence — for example, "this agency did this documented thing on this date." Claims without that kind of evidence fall outside the code, whether they assert something happened or assert it did not. When evidence does not exist for either side, the honest answer is that the question is open.
Engage the work as it is. Early-stage work is not evaluated as if it were mature; deferred-falsification claims are not evaluated as if they required present-day testing. The relevant standard is the one the event applies under Rule 4 for the track in which the work is classified. Questions or submissions that attack work for failing to meet a standard that does not apply to it fall outside the code.