{ "guide": "gloucester-companion", "corpus_version": "2026-09-17.1", "generated_at": "2026-09-17T08:14:23Z", "source_page": "https://companion.e5enclave.com", "honesty_policy": "This guide is a 30-day field scan (Aug 17-Sep 16, 2026) of new evidence on AI capability, work, infrastructure, and Black economic agency, written in plain language for a non-technical audience. The evidence_strength label on each card is the author's judgment, not a peer-review verdict: strong = large survey, established dataset, or auditable catalog; moderate = single study, scenario model, vendor estimate, or opinion; emerging = early signals; unverified = treat as a question, not a settled fact. An empty caveats array means 'no explicit caveat was stated on the page' - not 'no caveat exists.' Card order and card-NN ids are stable only within a corpus_version; cite as corpus_version + card id + content_hash prefix so staleness is detectable.", "corpus_hash": "sha256:98fcca64a245285c4fb40408bd69e7f9c1b0b94c0b98fd64abb7f30d19725305", "cards": [ { "id": "card-01", "anchor": "#card-01", "rank": 1, "category": "Economic scenarios", "title": "Scenarios for our Economic Future", "publisher": "Anthropic Economics", "date": "2026-09-09", "date_display": "Sep 9, 2026", "summary": "Three possible US economies in 2030, from modest gains to extreme growth and displacement. The “substantial” path puts GDP about 8.3% above baseline while knowledge-worker wages stay broadly flat.", "takeaway": "AI can create abundance without distributing it. Deliberate ownership, training, and institution-building determine who benefits.", "equity_note": "Direct to the distribution question: growth alone does not determine who gains.", "tags": [ "work", "capability" ], "plain_language": true, "source_urls": [ { "url": "https://www.anthropic.com/institute/econ-scenarios", "label": "Open source" } ], "caveats": [ "2030 scenario modeling - projections, not predictions." ], "evidence_strength": "moderate", "content_hash": "sha256:30fc6669727d0ba144c61b082572a57f59f8a576f4e30148011167ff49a4b379" }, { "id": "card-02", "anchor": "#card-02", "rank": 2, "category": "Black economic agency", "title": "AI Fluency Initiative", "publisher": "National Action Network · reported by Urban Journal News", "date": "2026-09-16", "date_display": "Sep 16, 2026", "summary": "A practical training initiative for students, workers, entrepreneurs, clergy, and community leaders. Its premise is direct: Black communities are not simply labor to be displaced or demographics to reach after disruption.", "takeaway": "This is a current Black institutional answer to AI: repeated practice, broad access, and participation as builders.", "equity_note": "Explicit—the initiative is designed for Black communities as builders, not only people affected by disruption.", "tags": [ "equity", "work" ], "plain_language": true, "source_urls": [ { "url": "https://urbanjournalnews.com/2026/09/16/nan-launches-ai-initiative-to-help-black-communities-prepare-for-changing-workforce/", "label": "Open source" } ], "caveats": [ "Program announcement as reported by a single news outlet; outcomes not yet measured." ], "evidence_strength": "moderate", "content_hash": "sha256:657c6f2e6a4aaa11906432cbeef753213e837305a08f099cc11ab266e56f0ff4" }, { "id": "card-03", "anchor": "#card-03", "rank": 3, "category": "Black workers", "title": "Automation Threatens the Future of Black Workers in America", "publisher": "National Urban League · Marc H. Morial", "date": "2026-09-16", "date_display": "Sep 16, 2026", "summary": "The op-ed cites an established McKinsey warning: AI could disrupt 4.5 million jobs held by African Americans, who face a 10% greater likelihood of automation-based job loss.", "takeaway": "The 2026 acceleration evidence makes an older warning urgent. The data are not new research from 2026; the institutional framing is.", "equity_note": "Explicit—African American workers and occupational exposure are the focus.", "tags": [ "equity", "work" ], "plain_language": true, "source_urls": [ { "url": "https://NUL.org/news/op-ed-automation-threatens-future-black-workers-america", "label": "Open source" } ], "caveats": [ "Op-ed citing an established McKinsey warning; the data are not new research from 2026." ], "evidence_strength": "moderate", "content_hash": "sha256:aefe46e12c010fd6cc3154884b8304a620d6794a72e011a2cdf5beaba1ac7312" }, { "id": "card-04", "anchor": "#card-04", "rank": 4, "category": "Organizational adoption", "title": "The State of AI: Global Survey 2026", "publisher": "McKinsey & Company · QuantumBlack", "date": null, "date_display": "2026 · exact date not listed", "summary": "Nearly 9 in 10 respondents use AI somewhere. Yet 80% report individual productivity gains while only 37% report positive EBIT impact; just 6% qualify as high performers.", "takeaway": "Buying access is not transformation. High performers redesign workflows; most organizations still bolt AI onto old ways of working.", "equity_note": "Not explicit in the survey; the gap matters for smaller Black-led institutions that cannot afford failed adoption.", "tags": [ "work", "capability" ], "plain_language": true, "source_urls": [ { "url": "https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai", "label": "Open source" } ], "caveats": [ "Publication date not stated on the survey page." ], "evidence_strength": "strong", "content_hash": "sha256:8ab2c00d71004a9dc631405d3634c472ca10471bceec506b0f16f002d783d72f" }, { "id": "card-05", "anchor": "#card-05", "rank": 5, "category": "Workforce scenarios", "title": "AI Could Reshape the US Workforce in 4 Very Different Ways", "publisher": "The Conference Board · public release", "date": "2026-09-15", "date_display": "Sep 15, 2026", "summary": "Four futures range from gradual augmentation to uneven disruption. Within three years, 60–70% of cognitive jobs could involve human–AI collaboration, compared with 15–25% remaining human-only.", "takeaway": "Forecasts differ, but the preparation agenda does not: build better indicators, train people, and make public systems responsive.", "equity_note": "Not explicit; the scenarios help leaders prepare institutions before unequal shocks arrive.", "tags": [ "work" ], "plain_language": true, "source_urls": [ { "url": "https://www.prnewswire.com/news-releases/report-ai-could-reshape-the-us-workforce-in-4-very-different-ways-302879196.html", "label": "Open source" } ], "caveats": [ "Scenario forecasts - ranges, not point predictions." ], "evidence_strength": "moderate", "content_hash": "sha256:9e33ed22c49894ec91ebe7e253f54267d2ff0e099c6f74956d6231b73fbf7ce3" }, { "id": "card-06", "anchor": "#card-06", "rank": 6, "category": "Productivity & access", "title": "AI adoption and the productivity promise", "publisher": "European Central Bank", "date": "2026-08-26", "date_display": "Aug 26, 2026", "summary": "The median AI user reports saving about 3 hours a week, or 7.7% of median work time. Across the economy, the implied saving is only about 3.8% because fewer than half both use AI and save time.", "takeaway": "The gap is not just access to software. About half of workers say training and a clearer understanding of usefulness would encourage adoption.", "equity_note": "Not explicit; the age, education, and training gaps identify where equitable access work can begin.", "tags": [ "work" ], "plain_language": true, "source_urls": [ { "url": "https://www.ecb.europa.eu/press/blog/date/2026/html/ecb.blog20260826~e1c1a89999.ga.html?utm_source=chatgpt.com", "label": "Open source" } ], "caveats": [ "Self-reported survey figures; the economy-wide 3.8% is the guide's own implied calculation, not the source's." ], "evidence_strength": "moderate", "content_hash": "sha256:af3fdedbfcae43d7151ac6ffa14f27654e3a093699efd6074b6d0432bde45a3a" }, { "id": "card-07", "anchor": "#card-07", "rank": 7, "category": "Capability benchmark", "title": "MirrorCode: AI can rebuild entire programs from behavior alone", "publisher": "Epoch AI · co-developed with METR", "date": null, "date_display": "Approx. Aug 2026", "summary": "Claude Opus 4.7 reimplemented a roughly 16,000-line bioinformatics toolkit in 14 hours for $251, passing 2,000 of 2,001 tests. The task was estimated at 2–17 human-weeks.", "takeaway": "This turns “longer tasks” into “whole systems.” Epoch also warns that long autonomous runs change evaluation budgets and discloses possible memorization risk.", "equity_note": "Not explicit; it demonstrates the scale of technical capacity institutions can now access.", "tags": [ "capability", "tools" ], "plain_language": false, "source_urls": [ { "url": "https://epoch.ai/MirrorCode", "label": "Explainer" }, { "url": "https://epoch.ai/files/MirrorCode.pdf", "label": "Paper" }, { "url": "https://github.com/epochresearch/MirrorCode/", "label": "GitHub" } ], "caveats": [ "Single benchmark result; Epoch AI itself warns that long autonomous runs change evaluation budgets." ], "evidence_strength": "moderate", "content_hash": "sha256:32c188c60ef1432e5e2bc295e5310e39201840a1e6937f980d99c37aa1142d95" }, { "id": "card-08", "anchor": "#card-08", "rank": 8, "category": "Research tool", "title": "Benchmark Radar", "publisher": "Koutian Wu and collaborators", "date": "2026-09-14", "date_display": "Active daily · paper Sep 2026", "summary": "A living catalog of 14,810+ benchmark, evaluation, and dataset records from 37 public sources, with daily discovery, a dashboard, command-line tools, and a downloadable dataset.", "takeaway": "Today’s frontier is perishable. This project keeps the evidence linked, ranked, and updateable rather than freezing progress into one score.", "equity_note": "Not explicit; transparent evidence infrastructure helps institutions make independent choices.", "tags": [ "tools", "capability" ], "plain_language": false, "source_urls": [ { "url": "https://github.com/ktwu01/benchmark-radar", "label": "GitHub" }, { "url": "https://arxiv.org/abs/2609.11115", "label": "Paper" } ], "caveats": [ "Living dataset - record counts change daily; verify current figures at the source." ], "evidence_strength": "moderate", "content_hash": "sha256:0008694cc085eb3d4a1a532c2e836fcb4d727146199d4d413f8595d7275c9850" }, { "id": "card-09", "anchor": "#card-09", "rank": 9, "category": "Research paper", "title": "Benchmark Radar: a living database for AI evaluation", "publisher": "Koutian Wu, Junjie Zhou, Ergan Shang, and collaborators", "date": null, "date_display": "Sep 2026 · exact day not verified", "summary": "The paper explains how the living catalog is collected, searched, and audited, including how benchmarks become less useful when frontier models saturate them.", "takeaway": "It shows why institutions need a continuing evidence practice rather than a single frozen score or annual handout.", "equity_note": "Not explicit in the paper; useful as infrastructure for transparent, continuously updated decision-making.", "tags": [ "tools", "capability" ], "plain_language": false, "source_urls": [ { "url": "https://arxiv.org/abs/2609.11115", "label": "Open paper" } ], "caveats": [ "Exact publication day not verified (Sep 2026)." ], "evidence_strength": "strong", "content_hash": "sha256:a33a96c4cd121a4e9a7c5af9ef9739311eda75bed6d10057b3bcf1e7dea54995" }, { "id": "card-10", "anchor": "#card-10", "rank": 10, "category": "Capability & cost", "title": "From BERT to Frontier Agents", "publisher": "Pranav Kumar Kaliaperumal", "date": null, "date_display": "Aug 2026", "summary": "The paper reports roughly 5.8× annual growth in the odds of resolving real GitHub issues since October 2024, alongside a roughly 60× fall in input-token price from GPT-3 to a 2026 budget-tier model.", "takeaway": "Capability is becoming affordable, and task-targeted model routing can beat dependence on a single frontier system.", "equity_note": "Not explicit; falling costs lower the entry barrier for Black-led organizations and firms.", "tags": [ "capability" ], "plain_language": false, "source_urls": [ { "url": "https://arxiv.org/html/2608.13675v1", "label": "Open paper" } ], "caveats": [ "arXiv working paper; some figures mix vendor and independent evaluations." ], "evidence_strength": "moderate", "content_hash": "sha256:1903cb7c83edd3e1397ef6b44e2b201516807f94cf055218f77e70508f730741" }, { "id": "card-11", "anchor": "#card-11", "rank": 11, "category": "Infrastructure buildout", "title": "Data-center capital spending grew 92%", "publisher": "Dell’Oro Group · public summary via PR Newswire", "date": "2026-09-16", "date_display": "Sep 16, 2026", "summary": "Worldwide data-center capital spending rose 92% year over year in the second quarter of 2026. US and China hyperscalers doubled spending.", "takeaway": "The AI economy is being built in physical infrastructure now. Who supplies, owns, powers, secures, and maintains it is an open economic question.", "equity_note": "Not explicit; it surfaces ownership, contracting, workforce, and supplier opportunities.", "tags": [ "infrastructure" ], "plain_language": true, "source_urls": [ { "url": "https://www.prnewswire.com/news-releases/data-center-capex-grew-92-percent-in-2q-2026-driven-by-surging-ai-demand-and-memory-costs-according-to-delloro-group-302879910.html", "label": "Open summary" } ], "caveats": [ "Vendor capex estimate via PR summary; the full report is paid." ], "evidence_strength": "moderate", "content_hash": "sha256:f5c408f66a1b5be8a36ddc80ba8113805eeee23e1fa7b4dbcb1392b66a8b8adc" }, { "id": "card-12", "anchor": "#card-12", "rank": 12, "category": "Counterpoint", "title": "Fast-embraced AI will be slow to lift productivity", "publisher": "Reuters Breakingviews", "date": "2026-09-16", "date_display": "Sep 16, 2026", "summary": "About 44% of US workplaces used AI by May 2026 — roughly three times faster than the historical PC rollout — while broad productivity gains remain difficult to see.", "takeaway": "Transformation has costs: retraining, consulting, workflow change, and measurement. This commentary is a useful skeptic’s check, not a primary report.", "equity_note": "Not explicit; it warns against promising productivity gains without funding the transition work.", "tags": [ "work" ], "plain_language": true, "source_urls": [ { "url": "https://www.reuters.com/commentary/breakingviews/fast-embraced-ai-will-be-slow-lift-productivity-2026-09-16/", "label": "Open commentary" } ], "caveats": [ "Opinion commentary, not original research." ], "evidence_strength": "moderate", "content_hash": "sha256:5a4bb1c811010d1160404138d2e21a9fe0f1b9216ccacd9b1e66ef82176fa44b" }, { "id": "card-13", "anchor": "#card-13", "rank": 13, "category": "Equity question", "title": "Are Black women absorbing the labor shock first?", "publisher": "The Digital Economist · PR-distributed article", "date": "2026-08-25", "date_display": "Aug 25, 2026", "summary": "The article claims 326,000 Black women exited the workforce amid AI-driven shifts, with about $37 billion in lost GDP. The underlying analysis was not independently verified in the recon.", "takeaway": "Treat this as a question demanding better evidence, not as a settled statistic. The source is promotional and not peer-reviewed.", "equity_note": "Explicit—Black women’s labor outcomes are the subject, but the underlying estimate remains unverified.", "tags": [ "equity", "work" ], "plain_language": true, "source_urls": [ { "url": "https://markets.financialcontent.com/ibtimes/news/article/digitaljournal-2026-8-25-the-digital-economist-explores-ais-impact-on-black-womens-jobs-326000-exits-37b-gdp-loss", "label": "Open source" } ], "caveats": [ "Treat as a question demanding better evidence, not a settled statistic. The source is promotional and not peer-reviewed; the 326,000 estimate is unverified." ], "evidence_strength": "unverified", "content_hash": "sha256:75abdce28ae2e298cdf8bee64d54f1c07ac1e247629bde8b0e92105ee988129b" }, { "id": "card-14", "anchor": "#card-14", "rank": 14, "category": "Wealth research", "title": "Federal mortgages built the wealth gap — 2% of FHA loans reached Black borrowers", "publisher": "UC Santa Barbara · Nature study", "date": "2026-07-29", "date_display": "Jul 29, 2026", "summary": "The first national-level analysis of FHA and VA lending in 1935–1947, published in Nature, found only 2% of FHA loans and 5% of VA loans went to Black borrowers, who were 10% of the U.S. population and 10% of veterans. The team digitized more than 100,000 paper mortgage records and matched them to Census data; foreign-born borrowers, by contrast, received loans at rates comparable to their share of the population.", "takeaway": "Documented mechanism, not theory: federal policy choices in one generation set the terms of wealth accumulation for the next. The researchers warn that any new housing policy should be judged on what it sets in motion 50 years later.", "equity_note": "Explicit — Black borrowers' exclusion is the subject; it quantifies the mechanism behind the racial wealth gap.", "tags": [ "equity" ], "plain_language": true, "source_urls": [ { "url": "https://news.ucsb.edu/2026/022720/exclusionary-federal-mortgage-lending-widened-racial-wealth-gap-us", "label": "Open source" } ], "caveats": [ "Historical window is 1935-1947 only; the study quantifies the lending mechanism, not a dollar figure for today's wealth gap.", "Method is archival record linkage (100,000+ digitized mortgages matched to Census data), not a controlled experiment." ], "evidence_strength": "strong", "content_hash": "sha256:66fadb29ef2e9f21642a0f8a9509a079e038e76ac000213eb7eb5c7aec3cd56f" }, { "id": "card-15", "anchor": "#card-15", "rank": 15, "category": "Economic outlook", "title": "Capabilities double every four months; institutions cannot keep up", "publisher": "McKinsey Global Institute", "date": "2026-09-16", "date_display": "Sep 16, 2026", "summary": "The complexity of tasks that AI can reliably perform has been doubling roughly every four months since 2023, while data centers, power capacity, and chip manufacturing expand far more slowly. The report finds nearly nine in ten organizations use AI in at least one business function — but outside a small group of high performers, only about one-quarter redesigned their workflows around AI.", "takeaway": "The bottleneck is no longer the model — it is power, chips, and organizational change. Whoever owns the infrastructure and does the workflow redesign captures the value.", "equity_note": "Not explicit; the finding that gains come from workflow redesign and infrastructure — not access alone — warns institutions that cannot afford failed adoption.", "tags": [ "work", "capability", "infrastructure" ], "plain_language": true, "source_urls": [ { "url": "https://www.mckinsey.com/mgi/our-research/The-AI-economy-Interconnected-forces-feedback-loops-and-speeds-of-change", "label": "Open source" } ], "caveats": [ "Primary document read in full 2026-09-17; survey figures are McKinsey's own (n=1,719 across 97 countries, May-Jun 2026; governance survey n~500, Dec 2025-Jan 2026).", "Infrastructure figures (power-plant interconnection timelines, Data Center Watch project counts) are McKinsey-reported, not independently verified." ], "evidence_strength": "moderate", "content_hash": "sha256:fdaff5178bba3433709b311f5f6bc7095f5a146c32873a48fcdc3fd9ea483da1" }, { "id": "card-16", "anchor": "#card-16", "rank": 16, "category": "Policy research", "title": "AI could narrow the racial wealth gap — only if it is guarded", "publisher": "The Leadership Conference on Civil and Human Rights · Center for Civil Rights and Technology", "date": "2026-06-24", "date_display": "Jun 24, 2026", "summary": "Built on qualitative interviews with people of color who say AI decision-making systems affected their pursuit of jobs, financial services, and housing, the report argues AI could narrow the racial wealth gap with legal safeguards and corporate responsibility in place — but will keep widening disparities if left unchecked. It closes with recommendations for legislators, regulators, developers, employers, and educators.", "takeaway": "A current civil-rights agenda for AI: safeguards in hiring, lending, and housing are the conditions under which the technology serves Black economic agency rather than undermining it.", "equity_note": "Explicit — Black communities' financial futures are the stated subject; the case is qualitative, built on interviews rather than statistics.", "tags": [ "equity", "work" ], "plain_language": true, "source_urls": [ { "url": "https://civilrights.org/2026/06/24/new-report-ai-safeguards-may-narrow-racial-wealth-gap/", "label": "Open source" } ], "caveats": [ "Qualitative interview study by an advocacy coalition, not a statistical analysis; interviewees self-selected as impacted by AI decision systems.", "Published June 24, 2026 - predates the living-scan window but is new research for the guide." ], "evidence_strength": "moderate", "content_hash": "sha256:f33fed97db4419799727f57ddd2a2dda1fef48eb49bbae5b8e6f114045aa33e7" } ], "charts": [ { "id": "chart-01", "title": "Adoption is not transformation.", "source": "McKinsey & Company - QuantumBlack, The State of AI: Global Survey 2026", "figures": [ { "label": "Use AI somewhere", "value": "88%" }, { "label": "Report individual gains", "value": "80%" }, { "label": "Report EBIT impact", "value": "37%" }, { "label": "Qualify as high performers", "value": "6%" } ], "caveats": [ "Publication date not stated." ] }, { "id": "chart-02", "title": "Capability compounds.", "source": "Pranav Kumar Kaliaperumal, From BERT to Frontier Agents, Aug 2026", "figures": [ { "label": "Annual growth in the odds of resolving real GitHub issues", "value": "5.8x/yr" }, { "label": "Fall in input-token price, GPT-3 era to a 2026 budget-tier model", "value": "60x" }, { "label": "Measurement window start", "value": "Oct 2024" } ], "caveats": [ "Some figures mix vendor and independent evaluations." ] }, { "id": "chart-03", "title": "Cost collapses.", "source": "Pranav Kumar Kaliaperumal, From BERT to Frontier Agents, Aug 2026", "figures": [ { "label": "Indexed input-token price, GPT-3 era", "value": "100" }, { "label": "Indexed input-token price, 2026 budget tier", "value": "1.7" } ], "caveats": [ "Indexed input-token price; endpoints only - no interpolation." ] }, { "id": "chart-04", "title": "The buildout is physical.", "source": "Dell'Oro Group public summary via PR Newswire, Sep 16, 2026", "figures": [ { "label": "Data-center capex, 2Q 2025 (indexed)", "value": "100" }, { "label": "Data-center capex, 2Q 2026 (indexed)", "value": "192 (+92% YoY)" } ], "caveats": [ "Full report is paid." ] }, { "id": "chart-05", "title": "Time saved, unevenly.", "source": "European Central Bank blog, Aug 26, 2026", "figures": [ { "label": "Median AI user: share of median work time saved", "value": "7.7% (~3 hrs/week)" }, { "label": "Economy-wide: share of median work time saved", "value": "3.8%" } ], "caveats": [ "About half of workers cite training gaps." ] } ] }