{"id":"0349e1cd-2715-43f6-996d-d0c41908799f","arxiv_id":"2508.19749","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"AGI hype is reframed as 'deep hype': a long-term, unverifiable overpromissory dynamic sustained by sociotechnical fictions and venture capital speculation.","lead":"This paper introduces the concept of 'deep hype' to describe how AGI stays perpetually promising through a mix of uncertainty, speculative fiction, and venture capital logic. It argues this dynamic shapes investment and policy while advancing a cyberlibertarian agenda, and it matters for anyone asking who governs AI's future.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central diagnosis hinges on AGI's infeasibility, which the paper asserts but does not establish; its own cited sources (Bubeck et al. 2023; Agüera y Arcas & Norvig 2023) assert feasibility or sparks.","rationale":"I focused on the infeasibility premise because it is the most logically load-bearing. The reader's weakest_assumption also identified this issue and the corpus representativeness issue; I focus on feasibility because it is internal to the paper's own framework and can be resolved by argumentative revision, whereas corpus representativeness is an empirical transparency issue that is secondary to the central conceptual claim. The paper is a theoretical contribution, so the central claim lives or dies on whether 'sociotechnical fiction' is a stable category for AGI. The paper asserts infeasibility as a premise, but its own evidence includes feasibility claims. If those claims are ignored, the analysis becomes cherry-picking; if they are taken seriously, the 'precisely because' clause fails. A concrete test—replacing the clause with 'contested feasibility' and checking the argument—would settle whether the diagnosis is robust. Because the paper's contribution could survive a softened premise, I would not reject it; but the current formulation is internally unstable, so conditional acceptance is appropriate. I agree partially with the reader's weakest_assumption because the reader also mentioned corpus representativeness, which I do not address as the primary concern. No ad hominem is intended; this is purely an argument-level critique.","tokens_in":29883,"tokens_out":3996,"duration_ms":44346,"concrete_test":"Rewrite the Discussion's central clause, replacing 'not technically feasible' with 'contested feasibility'. Trace whether the rest of the argument (deep hype, recursive uncertainty, VC speculation) still follows. If it does, the infeasibility premise is not load-bearing and the paper should be amended accordingly. If it does not, the conclusion is conditional on resolving an empirical debate that the paper itself leaves open. Additionally, inspect the definition of sociotechnical fiction (Belsunces, 2025): if it only requires 'not-yet-existing' and 'technoscientific legitimacy', then the added infeasibility condition is unsupported and should be either evidenced or removed.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The load-bearing premise is in the Discussion: 'AGI functions as a sociotechnical fiction precisely because it is not technically feasible.' This infeasibility claim is what separates the 'sociotechnical fiction' concept from ordinary promissory research. Yet the paper itself undercuts it: it cites Kaplan et al. (2020) scaling laws as a 'proof of verifiable truth' that strengthens AGI visions, and acknowledges that 'others claim that LLMs already embody AGI' (Agüera y Arcas and Norvig, 2023) and Bubeck et al. (2023) describe 'sparks' of AGI. If any of these are correct, AGI is not simply fiction; it is a contested emerging technology with plausible timelines. The recursive deep-hype mechanism could still operate, but the 'precisely because' clause requires a technical justification. Without it, the central distinction between deep hype and ordinary hype about a feasible technology collapses, and the claim that AGI is 'the most prominent contemporary sociotechnical fiction' becomes interpretive assertion rather than a supported conclusion. The paper must either defend infeasibility against its own cited counter-evidence or weaken the claim to 'contested feasibility' and show that the diagnosis survives.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper argues that Artificial General Intelligence (AGI) is sustained by a distinct form of hype—'deep hype'—defined in §4 as a long-term, overpromissory dynamic that constructs civilisational-transformation promises through a network of uncertainties extending into an undefined future, making verification nearly impossible in the present. Drawing on an unspecified 'series of statements and documents' from AGI leaders, the paper proposes nine uncertainty arenas (conceptual, application, temporal, value; governance, economic, geopolitical, existential, ethical) and claims that these uncertainties are articulated through 'sociotechnical fictions' that render not-yet-existing technologies intelligible, desirable, and urgent. The central conclusion is that AGI deep hype results from venture-capital logic converging technological and financial speculation, forming a recursive relation between uncertainty, fiction, and hype; this dynamic advances a cyberlibertarian and longtermist programme, sidelines democratic oversight, and functions as a form of governance of AI futures. The paper is framed as a theoretical contribution to hype studies and critical governance studies.","tokens_in":30081,"tokens_out":4738,"duration_ms":55678,"significance":"If the argument holds, the paper makes a useful contribution by moving hype analysis from short product cycles to long-term promissory dynamics and by explicitly linking hype to venture-capital speculation, sociotechnical imaginaries, and governance. It usefully synthesises hype studies, STS, and critical political economy, and it assembles a broad set of contradictory public statements by AGI leaders. The concept of 'sociotechnical fiction' is a potentially valuable extension of Beckert's fictional expectations. The paper's main weaknesses are that the empirical corpus is not specified, and that the load-bearing premise that AGI is 'not technically feasible' is asserted rather than established, with the paper's own cited sources (Kaplan et al. 2020; Bubeck et al. 2023; Agüera y Arcas & Norvig 2023) cutting the other way. The recursive 'deep hype' mechanism is theoretically coherent but risks circularity. With targeted revisions to the infeasibility claim and a transparent account of the corpus, the paper could be a valuable critical intervention.","major_comments":[{"comment":"The argument that AGI is a sociotechnical fiction—and the paper's headline claim—rests on the sentence: 'AGI functions as a sociotechnical fiction precisely because it is not technically feasible.' This premise is load-bearing and is not defended. The paper itself cites Kaplan et al. (2020) scaling laws as a 'proof of verifiable truth' (Application uncertainty section), Bubeck et al. (2023) on 'sparks' of AGI (Temporal uncertainty section), and Agüera y Arcas & Norvig (2023) claiming LLMs already embody AGI. These sources undercut the asserted infeasibility. If AGI is a contested but potentially feasible technology, the distinction between sociotechnical fiction and ordinary promissory research collapses, and the conclusion that AGI is 'the most prominent contemporary sociotechnical fiction' becomes unsupported. Please either engage substantively with the cited feasibility claims (with t","section":"Discussion – The role of Sociotechnical Fictions in AGI deep hype"},{"comment":"The paper repeatedly states that its analysis is 'based on the analysis of a series of statements and documents from key AGI actors,' but it never specifies the corpus: no selection criteria, source list, time period, number of documents, sampling strategy, or search procedure. This makes the nine-uncertainty-arena taxonomy and the fit between evidence and framework uncheckable. The reader cannot tell whether the quotes were selected to illustrate pre-existing categories or whether the categories emerged from a systematic corpus. If the paper is intended as a theoretical essay, the empirical material should be explicitly labelled as illustrative rather than evidential; if it is intended as an empirical discourse analysis, a methodological appendix is required. As written, the empirical grounding is not auditable, which is a significant limitation for a paper whose central claims are abou","section":"The Uncertainties of Artificial General Intelligence (Section 4)"},{"comment":"The concept 'deep hype' is defined with the very features that the paper then discovers in AGI: long-term civilisational promises, a network of uncertainties, unverifiability, and a recursive relationship between hype and fiction. The nine arenas are generated from the same selected corpus, and the recursive relation is asserted in the Discussion rather than demonstrated. This is a circularity risk: the empirical sections can only confirm what was already built into the definition. The paper should explicitly acknowledge that 'deep hype' is an interpretive concept constructed through this case, not a hypothesis tested against it, or it should provide an independent operationalisation (e.g., distinguishing deep hype from other long-term promissory technologies such as nuclear fusion or quantum computing before applying it to AGI). Without this, the central contribution risks being an elab","section":"Definition of deep hype (Section 4) and recursive claim (Discussion)"}],"minor_comments":[{"comment":"Chronology error: the paper says 'Six days after the hearing, on March 22 of the same year, the Future of Life Institute published an open letter...' The Senate hearing was on May 16, 2023; the FLI letter was March 22, 2023, before the hearing. Please correct the chronology.","section":"Governance uncertainty"},{"comment":"Typographical errors: a stray 'Te.' in the paragraph beginning 'Like nuclear fusion...'; 'Andreesseen' for Andreessen; 'Wachawami' for Wadhwani; 'an the Open University' in footnote 1; and the placeholder citation 'AUTORb' in the section on sociotechnical fictions. These should be cleaned up.","section":"Discussion – AGI Deep Hype"},{"comment":"The statement 'AGI is not rooted in empirical scientific progress but in speculative thought, financial interests and technopolitical agendas' is stronger than the evidence presented, especially given the paper's own references to scaling laws and the claimed 'sparks' of AGI. Consider softening to 'not primarily rooted' or providing supporting argumentation.","section":"Discussion – AGI Deep Hype"},{"comment":"The list of nine uncertainty arenas is one of the paper's main analytic outputs, but it is presented as a simple enumeration. A table summarising each arena, its key actors, and representative quotes would improve readability and make the analytical structure easier to follow.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good fit for a critical STS or governance-oriented journal, though it sits at the boundary of hype studies and political economy. The main risk is not the political stance but the circularity and the unwarrented infeasibility premise. I would encourage the editor to send the revised version to a reviewer with expertise in AGI feasibility debates, since the author's own cited sources span both sides of that debate. There is also a citation-practice concern: the paper cites the author's own prior work via 'AUTORb', which is fine as a placeholder but should be resolved before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper deserves your attention if you work on hype, expectations, and AI governance. Its main contribution is a clear conceptualization of 'deep hype' as distinct from ordinary hype: a long-term, overpromissory dynamic built on unresolved uncertainties and civilizational promises that are nearly impossible to verify in the present. The recursive relation between venture capital speculation, sociotechnical fictions, and uncertainty is well illustrated with concrete quotes and funding data. This gives hype studies a sharper vocabulary for technologies like AGI that sustain excitement over decades rather than business cycles.\n\nThe strongest part is the empirical texture: testimonies, open letters, blog posts, and funding rounds are woven together to show how AGI leaders manage ambiguity strategically. The governance argument — that deep hype narrows the political debate and advances a cyberlibertarian and longtermist agenda — is plausible and timely.\n\nNow the soft spots, in proportion. The load-bearing premise is that AGI is a sociotechnical fiction 'precisely because it is not technically feasible.' That is asserted, not defended. The paper's own citations (Kaplan et al. scaling laws, Bubeck et al.'s sparks, Agüera y Arcas and Norvig) point to contested feasibility, not settled infeasibility. If AGI is on a plausible technical path, the fiction label becomes questionable. The author should either defend infeasibility against those sources or soften the claim to 'contested feasibility' and show that the recursive mechanism still operates. I think it would survive that revision, but it needs a different footing.\n\nThe remaining issues are real but minor. The corpus is described only as 'a series of statements and documents,' with no selection rules, so the evidence-to-framework fit is not checkable. There is an internal date contradiction in the governance section (the FLI letter is dated after the May 2023 hearing while also given a March 22 date). The 'AUTORb' placeholder and several missing bibliography entries are editorial slips, not conceptual problems.\n\nOverall, this is a solid, competent theoretical paper. The concept is novel enough to justify publication, and the political stakes are clear. The feasibility premise needs fixing, but the central interpretation does not collapse if you weaken it. I would send it to peer review and, if I were working on AI hype, I would cite it (with a critical note on that premise).\n\nBest,\n[Your name]","headline":"Deep hype is a genuinely useful concept for long-cycle technological promises, but the paper's 'not technically feasible' premise is asserted against its own cited sources and the corpus is too vague to fully check.","tokens_in":30682,"tokens_out":2996,"would_cite":true,"duration_ms":33369,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"AGI's lasting hype is not an accident but a recursive loop: venture capital speculation, uncertainty, and sociotechnical fictions feed one another, keeping AGI permanently out of reach while steering AI governance.","keywords":["Artificial General Intelligence","Deep Hype","Sociotechnical Fictions","Uncertainty","Venture Capital","Cyberlibertarianism","Longtermism","Governance of AI Futures"],"falsifier":"A verifiable demonstration of general intelligence — a system passing independently audited, pre-registered benchmarks for cross-domain learning and adaptation — would falsify the claim that AGI is a sociotechnical fiction rather than an engineering target. Short of that, a replicable corpus study with explicit selection rules showing that AGI discourse does not actually revolve around the nine uncertainty arenas, or evidence that funding follows demonstrated capability rather than promissory narrative, would undercut the recursive-loop mechanism.","tokens_in":29658,"feed_emoji":"🔮","tokens_out":10591,"duration_ms":100116,"temperature":0.7,"pith_summary":"This paper sets out to explain why Artificial General Intelligence keeps attracting record-breaking investment and political attention despite having no agreed definition and no demonstrated feasibility. The answer it proposes is 'deep hype': a long-term, overpromissory dynamic in which a network of uncertainties — what AGI is, when it will arrive, what it will do, what it is worth, what it risks — makes the promise nearly impossible to verify in the present, and thus keeps the window of investment open. The uncertainties are articulated through 'sociotechnical fictions': speculative visions presented with technoscientific legitimacy as if they were facts. The paper argues that venture capital logic converges technological and financial speculation, establishing a recursive relation: fictions tame and excite uncertainty, uncertainty feeds hype, and hype feeds the fictions. A reader should care because this reframes AGI's inflated promise as an engineered dynamic that narrows political debate and advances a cyberlibertarian, longtermist programme in which private actors, not democratic institutions, steer the technological future.","feed_headline":"AGI hype is a self-sustaining loop built on uncertainty","feed_subtitle":"Venture capital, uncertainty, and speculative fiction feed each other to steer AI governance, the paper argues.","key_machinery":"Two concepts carry the argument. 'Deep hype' is a long-term, overpromissory dynamic that constructs visions of civilisational transformation through a network of uncertainties extending into an undefined future, making promises nearly impossible to verify in the present — unlike conventional hype's short cycles. 'Sociotechnical fictions' are mediated forms of imagination produced within science and technology, sheltered by technoscientific legitimacy and so not recognised as fiction; they expand on the idea of fictional expectations and perform epistemic, aesthetic, affective, and normative work — rendering imaginary futures tangible, generating wonder and fear, and scripting collective acti","core_discovery":"On its own terms, the central claim is that AGI deep hype is the result of venture capital logic that converges technological and financial speculation, and that the sociotechnical fiction of AGI establishes a recursive relation to deep hype: hype is sustained by uncertainties that sociotechnical fictions tame and excite, while the fictions are fed by the hype and uncertainty they produce. AGI is presented as the most prominent contemporary sociotechnical fiction — an imagined entity sheltered by technoscientific legitimacy and treated as an inevitable fact. The paper maps nine arenas of uncertainty (conceptual, application, temporal, value, governance, economic, geopolitical, existential, e","pith_inferences":["Testable prediction the paper leaves implicit: each time a concrete AI milestone is reached, the goalposts will move rather than AGI being declared achieved, because a defined and verified AGI would collapse the fiction that sustains current valuations. A timeline study of how milestones are relabelled (e.g., as 'narrow AI' or 'tools') would test this.","The nine uncertainty arenas could serve as an analytic checklist for other speculative technologies; if fusion or quantum discourse shows the same recursive pattern, that would generalise deep hype beyond AI; if not, the pattern may be specific to AGI's particular combination of vagueness and financial scale.","Because the paper draws on an unspecified series of statements and documents, the cleanest way to pressure its diagnosis is a replicable discourse analysis: if a systematically sampled corpus of AGI communications is not dominated by these uncertainty arenas, the recursive-loop claim would need revision.","A governance countermeasure follows from the paper's logic though the author does not state it: anchoring regulation to demonstrated, verifiable capability rather than to promised futures would break the loop at its weakest point, since the loop depends on promises being unverifiable in the present."],"forward_implications":["AGI's vague definition and shifting timelines are structural features, not failures: they keep the promissory framework open and make the hype unfalsifiable in the present, which is exactly why investors and promoters cultivate ambiguity.","Regulatory and public attention that responds to AGI's promises rather than to demonstrated capabilities plays into the loop, because existential-risk narratives are themselves sociotechnical fictions that legitimise the actors who raise the alarms.","AGI hype functions as a form of soft anticipatory governance: it narrows political debate, frames the AGI future as unavoidable, steers research and investment, and consolidates the cyberlibertarian and longtermist agenda while displacing near-term concerns such as bias, environmental impact, and labour displacement.","Because hype and sociotechnical fictions remain reactivatable, AGI hype should not be expected to follow the classic hype-cycle collapse; instead, expectations can be renewed and restaged after any disappointment.","The same deep-hype dynamic should be visible in other perpetually imminent technologies — the paper names nuclear fusion, quantum computing, and brain-computer interfaces — wherever civilisational promises meet unresolved uncertainties and speculative capital."],"supporting_citations":[{"why":"Supplies the concept of fictional expectations that the paper's sociotechnical fictions expand on to explain economic decision-making under future uncertainty.","marker":"Beckert, 2016"},{"why":"Defines sociotechnical fictions, the paper's central concept; the claim that AGI is the largest such fiction depends on this definition.","marker":"Belsunces, 2025"},{"why":"Provides the performative theory of hype and the 'expansive repertoire of expectations' that explains why vague technologies produce more resilient hype.","marker":"van Lente, Spitters and Peine, 2013"},{"why":"Frames hype as an opportunistic manufactured event and a rapidly closing window, the basis for the paper's account of how deep hype sustains urgency.","marker":"Bareis, Roßmann & Bordignon, 2023"},{"why":"Supports the claim that vague applications and ungrounded belief lead investors to spin optimistic tales, linking uncertainty to speculative financing.","marker":"Goldfarb and Kirsch, 2019"},{"why":"The scaling-laws result the paper cites as the 'proof of verifiable truth' that blurs the line between fiction and fact and lends AGI material traction.","marker":"Kaplan et al, 2020"},{"why":"Underpins the argument that ambiguity and contradiction are strategically mobilised to redirect regulations, investment and power.","marker":"Pfaffenberger, 1992"},{"why":"The charter that mainstreamed AGI as a corporate mission; the paper's account of AGI's rise and its political programme depends on this founding claim.","marker":"OpenAI, 2018"},{"why":"Situates AI hype within the recurring logic of speculative tech bubbles, the comparison the paper builds on to distinguish AGI deep hype.","marker":"Floridi, 2024"}],"fun_headline_variants":["AGI hype: self-sustaining uncertainty and speculation","How AGI hype becomes a self-sustaining speculative fiction","AGI hype sidelines democratic oversight with speculative fictions","Deep hype: AGI's promise is a VC-driven sociotechnical fiction"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The diagnosis depends on AGI being technically infeasible — the paper states that AGI functions as a sociotechnical fiction precisely because it is not technically feasible — and if AGI turns out to be on a plausible path to being built, the line between fiction and ordinary promissory research blurs and the central claim weakens.","fun_headline_variants_meta":{"raw":{"variants":["AGI hype: self-sustaining uncertainty and speculation","How AGI hype becomes a self-sustaining speculative fiction","AGI hype sidelines democratic oversight with speculative fictions","Deep hype: AGI's promise is a VC-driven sociotechnical fiction"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000767,"raw_usage":{"total_tokens":3184,"prompt_tokens":636,"completion_tokens":2548,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":380,"completion_tokens_details":{"reasoning_tokens":2478}},"tokens_in":380,"tokens_out":2548,"duration_ms":20932,"temperature":1.0,"reasoning_tokens":2478,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T15:29:27.348771+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A verifiable demonstration of general intelligence — a system passing independently audited, pre-registered benchmarks for cross-domain learning and adaptation — would falsify the claim that AGI is a sociotechnical fiction rather than an engineering target. Short of that, a replicable corpus study with explicit selection rules showing that AGI discourse does not actually revolve around the nine uncertainty arenas, or evidence that funding follows demonstrated capability rather than promissory narrative, would undercut the recursive-loop mechanism.","supporting_citations":[],"review_version":1}