Evidence / 63 reviewed records

The source library.

Not a pile of footnotes. A two-way map from claims to original records, with source relationship, geography, study design, evidence type, and caveats attached.

Last reviewed August 23, 2026
Showing 63 of 63 reviewed sources
  1. 01
    U.S. Food and Drug Administration

    Artificial Intelligence-Enabled Medical Devices

    A periodically updated list of AI-enabled devices authorized for U.S. marketing, with links to regulatory records.

    Published 2026-08-23Reviewed 2026-08-23United Statesgovernment regulatory databasePrimaryIndependent
    Connected to 1 claim

    The FDA maintains a growing list of AI-enabled medical devices authorized for U.S. marketing after applicable premarket review.

    View original source ↗
  2. S015
    Commonwealth of Pennsylvania

    Governor Shapiro signs Executive Order on data center development

    Governor Shapiro signs Executive Order on data center development

    Published 2026-08-18Reviewed 2026-08-23Pennsylvaniaexecutive order announcementPrimaryIndependent
    Connected to 1 claim

    Pennsylvania Executive Order 2026-05 applies GRID permitting requirements to certain covered data-center projects above 25 MW.

    View original source ↗
  3. S021
    Pennsylvania Department of Community and Economic Development

    GRID implementation materials

    GRID implementation materials

    Published 2026-08-18Reviewed 2026-08-23Pennsylvaniagovernment implementation indexPrimaryIndependent
    Connected to 0 claims

    State implementation context for workforce, local benefit, and disclosure requirements.

    View original source ↗
  4. S022
    Commonwealth of Pennsylvania

    Governor's remarks on Executive Order 2026-05

    Governor's remarks on Executive Order 2026-05

    Published 2026-08-18Reviewed 2026-08-23Pennsylvaniaofficial remarksPrimaryIndependent
    Connected to 1 claim

    Pennsylvania Executive Order 2026-05 applies GRID permitting requirements to certain covered data-center projects above 25 MW.

    View original source ↗
  5. S023
    OpenAI

    OpenAI joins PORTS-Pike project

    OpenAI joins PORTS-Pike project

    Published 2026-08-17Reviewed 2026-08-23Pike County, Ohiocompany announcementPrimarySponsor-affiliated

    Read with care: Forward-looking sponsor disclosure; engineering intent is not measured performance.

    Connected to 4 claims

    OpenAI announced an agreement for approximately 8 GW of planned IT capacity at PORTS-Pike.

    The developer projects 35,000 construction jobs during a six-year buildout through 2032 and 2,500 long-term operating jobs.

    OpenAI and SB Energy describe $80 million in combined community-fund commitments, plus up to $84 million in student technology credits.

    View original source ↗
  6. S025
    U.S. Department of Energy Office of Environmental Management

    Portsmouth Site

    Portsmouth Site

    Published 2026-08-01Reviewed 2026-08-23Pike County, Ohiogovernment site profilePrimaryIndependent
    Connected to 0 claims

    Documents the former gaseous diffusion plant and continuing federal responsibilities.

    View original source ↗
  7. S037
    Louisiana Public Service Commission

    Louisiana Public Service Commission docket portal

    Louisiana Public Service Commission docket portal

    Published 2026-08-01Reviewed 2026-08-23Louisianaregulatory indexPrimaryIndependent
    Connected to 0 claims

    Authoritative location for orders and filings when direct document URLs change.

    View original source ↗
  8. S056
    Stanford Digital Economy Lab

    Canaries in the coal mine: August 2026 summary

    Canaries in the coal mine: August 2026 summary

    Published 2026-08-01Reviewed 2026-08-23United Statesresearch summaryIndependent

    Read with care: Observational payroll-data pattern; not a causal estimate.

    Connected to 0 claims

    Accessible explanation of age, exposure, hiring, and separation patterns.

    View original source ↗
  9. S057
    Stanford Digital Economy Lab

    Canaries in the Coal Mine: six facts about recent employment effects of AI

    Canaries in the Coal Mine: six facts about recent employment effects of AI

    Published 2026-08-01Reviewed 2026-08-23United Statesresearch publicationPrimaryIndependent

    Read with care: Not an economy-wide causal estimate and not a 19% job-loss claim.

    Connected to 1 claim

    Workers ages 22–25 in highly AI-exposed occupations had employment about 19% below a comparison trajectory based on similarly aged workers in less-exposed occupations.

    View original source ↗
  10. S058
    Stanford Digital Economy Lab

    Canaries in the Coal Mine: August 2026 working paper

    Canaries in the Coal Mine: August 2026 working paper

    Published 2026-08-01Reviewed 2026-08-23United Statesworking paperPrimaryIndependent

    Study design: observational payroll-data analysis · Study group stanford-canaries-2026

    Read with care: Association does not establish that AI caused the full difference.

    Connected to 1 claim

    Workers ages 22–25 in highly AI-exposed occupations had employment about 19% below a comparison trajectory based on similarly aged workers in less-exposed occupations.

    View original source ↗
  11. 11
    U.S. Geological Survey

    Colocating artificial intelligence data centers with energy infrastructure on Federal public lands

    A synthesis of siting, water, land, environmental, and energy considerations for AI data centers.

    Published 2026-07-22Reviewed 2026-08-23United Statesgovernment science synthesisPrimaryIndependent
    Connected to 2 claims

    A data center's water impact depends heavily on cooling design, electricity supply, climate, season, and the condition of the local watershed or aquifer.

    Estimated direct U.S. data-center water consumption rose from about 21.2 billion liters in 2014 to 66 billion liters in 2023 and is projected at roughly 140–280 billion liters annually in 2028.

    View original source ↗
  12. 12
    U.S. Environmental Protection Agency

    Clean Air Act Resources for Data Centers

    Clean Air Act requirements and permitting resources for stationary engines and turbines used at data centers.

    Published 2026-07-17Reviewed 2026-08-23United Statesgovernment regulatory guidancePrimaryIndependent
    Connected to 1 claim

    Data-center backup engines and onsite combustion turbines can create regulated local air emissions and may require Clean Air Act permitting and analysis.

    View original source ↗
  13. S032
    Meta

    Teachers and local businesses win as Meta expands Louisiana data center

    Teachers and local businesses win as Meta expands Louisiana data center

    Published 2026-07-15Reviewed 2026-08-23Richland Parish, Louisianacompany announcementPrimarySponsor-affiliated

    Read with care: Contracts, infrastructure, and job figures are company-reported.

    Connected to 1 claim

    Meta's expanded sponsor-described Richland Parish vision is about 5 GW, with more than $1.6 billion in Louisiana business contracts, more than $1 billion in described infrastructure improvements, and more than 1,000 projected operating roles.

    View original source ↗
  14. S030
    Microsoft

    Microsoft completes first Mount Pleasant datacenter facility

    Microsoft completes first Mount Pleasant datacenter facility

    Published 2026-06-23Reviewed 2026-08-23Mount Pleasant, Wisconsincompany announcementPrimarySponsor-affiliated
    Connected to 1 claim

    Microsoft reported that nearly 10,000 construction workers contributed across about two years and about 550 full-time employees supported the first operating Mount Pleasant facility as of June 2026.

    View original source ↗
  15. S001
    Lawrence Berkeley National Laboratory

    United States Data Center Energy Usage Report: 2025 Update

    United States Data Center Energy Usage Report: 2025 Update

    Published 2026-06-01Reviewed 2026-08-23United Statesgovernment research reportPrimaryIndependent
    Connected to 1 claim

    U.S. data-center electricity use is forecast at 649 TWh in the 2030 reference case, with compounded uncertainty of 521–843 TWh (9.5%–15.3% of U.S. electricity).

    View original source ↗
  16. S017
    Pennsylvania Public Utility Commission

    Load Forecast Accountability Report

    Load Forecast Accountability Report

    Published 2026-06-01Reviewed 2026-08-23Pennsylvaniaregulatory report indexPrimaryIndependent

    Read with care: The specific report is reached through the PUC document system; the library retains the agency index as a stable fallback.

    Connected to 0 claims

    Establishes expectations for utilities to test whether speculative large-load requests should enter forecasts.

    View original source ↗
  17. S059
    Stanford Digital Economy Lab

    AI Economic Indicators: June 2026 update

    AI Economic Indicators: June 2026 update

    Published 2026-06-01Reviewed 2026-08-23United Statesresearch indicatorsPrimaryIndependent
    Connected to 0 claims

    Broader labor-market context that prevents one subgroup pattern from becoming an economy-wide claim.

    View original source ↗
  18. S016
    Pennsylvania Public Utility Commission

    PUC releases final order establishing large-load model tariff framework

    PUC releases final order establishing large-load model tariff framework

    Published 2026-05-13Reviewed 2026-08-23Pennsylvaniaregulatory order summaryPrimaryIndependent
    Connected to 1 claim

    Pennsylvania's PUC model tariff generally addresses loads above 50 MW individually or 100 MW in aggregate.

    View original source ↗
  19. 19
    International Energy Agency

    Key Questions on Energy and AI

    An update tracking 2025 demand growth, efficiency gains, and physical bottlenecks.

    Published 2026-04-16Reviewed 2026-08-23Globalintergovernmental reportPrimaryIndependent
    Connected to 1 claim

    Global data-center electricity demand grew by an estimated 17% in 2025.

    View original source ↗
  20. S014
    Pew Research Center

    Most new data centers in the U.S. are coming to rural areas

    Most new data centers in the U.S. are coming to rural areas

    Published 2026-04-13Reviewed 2026-08-23United Statesindependent survey analysisIndependent
    Connected to 1 claim

    Pew estimated that about 67% of planned U.S. data centers were in rural areas, compared with about 13% of operating data centers.

    View original source ↗
  21. 21
    Virginia State Corporation Commission

    SCC Data Center Initiatives — Dominion Energy Virginia

    A summary of Virginia's GS-5 rate class and contract, minimum-charge, collateral, and reliability safeguards.

    Published 2026-03-26Reviewed 2026-08-23Virginiastate regulatory summaryPrimaryIndependent
    Connected to 2 claims

    A data center does not automatically raise every resident's bill, but weak cost-allocation rules can leave other customers carrying infrastructure risk or cost.

    Virginia created a separate GS-5 class for qualifying large loads, with a 14-year service obligation and minimum monthly transmission and distribution charges set at 85% of contracted levels.

    View original source ↗
  22. S046
    Virginia State Corporation Commission

    SCC issues order on Dominion Energy Virginia biennial review

    SCC issues order on Dominion Energy Virginia biennial review

    Published 2026-03-26Reviewed 2026-08-23Dominion Energy Virginia territoryregulatory order summaryPrimaryIndependent
    Connected to 1 claim

    Dominion's GS-5 framework applies to qualifying customers at or above 25 MW and includes minimum requirements around 85% of contracted transmission/distribution demand and 60% of generation demand.

    View original source ↗
  23. S050
    Nature Medicine

    AI-based chest X-ray prioritization in the lung cancer diagnostic pathway

    AI-based chest X-ray prioritization in the lung cancer diagnostic pathway

    Published 2026-03-24Reviewed 2026-08-23Englandpeer-reviewed randomized trialPrimaryIndependent

    Study design: multicenter randomized controlled trial · Sample 97,731 · Study group lungimpact

    Read with care: Tested prioritization within a pathway, not every form of AI image interpretation.

    Connected to 1 claim

    In LungIMPACT, median time to CT was about 53 days in both arms and time to lung-cancer diagnosis was about 44 versus 46 days, with no statistically meaningful acceleration of the targeted pathway.

    View original source ↗
  24. S049
    Nature Medicine

    AI-supported breast screening in Córdoba (AITIC)

    AI-supported breast screening in Córdoba (AITIC)

    Published 2026-03-19Reviewed 2026-08-23Córdoba, Spainpeer-reviewed clinical studyPrimaryIndependent

    Study design: prospective paired noninferiority study · Sample 31,301 · Study group aitic-mammography

    Read with care: One screening program and one AI-supported workflow; mortality was not the endpoint.

    Connected to 2 claims

    In AITIC, cancer detection increased from about 6.3 to 7.3 per 1,000 screenings while radiologist reading workload fell about 63.6%.

    In AITIC, recall rates increased from about 4.8% to 5.5%—a tradeoff alongside higher cancer detection.

    View original source ↗
  25. S013
    Pew Research Center

    How Americans view data centers' impact

    How Americans view data centers' impact

    Published 2026-03-12Reviewed 2026-08-23United Statesindependent survey analysisIndependent

    Read with care: Opinion measures perception and trust, not physical impact.

    Connected to 1 claim

    In Pew's 2026 survey, 39% said data centers are mostly bad for the environment and 38% said mostly bad for home energy costs, while 4% and 6% respectively said mostly good.

    View original source ↗
  26. 26
    Georgia Public Service Commission

    Data Center Fact Sheet

    A regulator's account of Georgia's large-load rules, hearings, generation planning, and ratepayer safeguards.

    Published 2026-03-01Reviewed 2026-08-23Georgiastate regulatory fact sheetPrimaryIndependent
    Connected to 2 claims

    A data center does not automatically raise every resident's bill, but weak cost-allocation rules can leave other customers carrying infrastructure risk or cost.

    Georgia's large-load rule permits longer contracts and minimum billing for new customers above 100 MW and requires regulator review of covered contracts.

    View original source ↗
  27. S028
    Ohio Consumers' Counsel

    Quick facts: data centers in Ohio

    Quick facts: data centers in Ohio

    Published 2026-01-01Reviewed 2026-08-23Ohioconsumer advocate fact sheetIndependent
    Connected to 0 claims

    Consumer-side explanation of large-load cost and risk questions.

    View original source ↗
  28. S040
    Georgia Power

    Data centers and Georgia Power's customer pledge

    Data centers and Georgia Power's customer pledge

    Published 2026-01-01Reviewed 2026-08-23Georgia Power territoryutility positionPrimarySponsor-affiliated
    Connected to 0 claims

    Utility commitment that must remain distinct from independently verified rate outcomes.

    View original source ↗
  29. S041
    Georgia Power

    Georgia Power highlights new customer contracts and updated forecasts

    Georgia Power highlights new customer contracts and updated forecasts

    Published 2026-01-01Reviewed 2026-08-23Georgia Power territoryutility filing summaryPrimarySponsor-affiliated
    Connected to 0 claims

    Current utility characterization of contracted load and planning assumptions.

    View original source ↗
  30. S042
    Georgia Public Service Commission

    Georgia PSC Docket 55378

    Georgia PSC Docket 55378

    Published 2026-01-01Reviewed 2026-08-23Georgiaregulatory docketPrimaryIndependent
    Connected to 0 claims

    Public docket for testing company statements against filings and orders.

    View original source ↗
  31. S043
    Georgia Public Service Commission

    Georgia PSC large-load order

    Georgia PSC large-load order

    Published 2026-01-01Reviewed 2026-08-23Georgia Power territoryregulatory orderPrimaryIndependent
    Connected to 1 claim

    Georgia Power rules for certain very large loads, generally above 100 MW, can require longer agreements, minimum bills, financial security, and infrastructure charges under regulatory oversight.

    View original source ↗
  32. S047
    Virginia General Assembly

    Virginia data center incentive amendment

    Virginia data center incentive amendment

    Published 2026-01-01Reviewed 2026-08-23Virginialegislative budget recordPrimaryIndependent
    Connected to 0 claims

    Shows that a wage threshold can be an incentive eligibility condition rather than evidence of actual wages paid.

    View original source ↗
  33. S048
    Office of the Governor of Virginia

    Virginia executive actions on data centers

    Virginia executive actions on data centers

    Published 2026-01-01Reviewed 2026-08-23Virginiagovernment announcement indexPrimaryIndependent

    Read with care: Executive announcements do not substitute for regulatory orders or project-level evidence.

    Connected to 0 claims

    Tracks current statewide policy and sponsor announcements without merging them into JLARC's independent model.

    View original source ↗
  34. S051
    OECD

    OECD Digital Education Outlook 2026

    OECD Digital Education Outlook 2026

    Published 2026-01-01Reviewed 2026-08-23OECD countriesintergovernmental synthesisIndependent
    Connected to 1 claim

    Students can perform better while AI is available yet learn less for later unaided work when assistance gives away answers; guardrails and tutor-like design can change the outcome.

    View original source ↗
  35. S055
    Stanford SCALE

    AI tutoring is not a monolith

    AI tutoring is not a monolith

    Published 2026-01-01Reviewed 2026-08-23United Statesresearch synthesisIndependent
    Connected to 0 claims

    Explains why answer-giving, hinting, and human-tutor support should not be treated as one intervention.

    View original source ↗
  36. S060
    Stanford Digital Economy Lab

    AI and labor markets: what we know and don't know

    AI and labor markets: what we know and don't know

    Published 2026-01-01Reviewed 2026-08-23United Statesresearch synthesisIndependent
    Connected to 0 claims

    Frames the limits of current labor evidence and the difference between exposure, adoption, and outcomes.

    View original source ↗
  37. 37
    Federal Energy Regulatory Commission

    FERC Directs PJM to Create New Rules for Co-Located Large Loads

    A federal action requiring transparent rules for large loads located beside generation in PJM.

    Published 2025-12-18Reviewed 2026-08-23PJM regionregulatory order summaryPrimaryIndependent
    Connected to 1 claim

    A data center does not automatically raise every resident's bill, but weak cost-allocation rules can leave other customers carrying infrastructure risk or cost.

    View original source ↗
  38. S034
    Meta

    Meta Louisiana project update

    Meta Louisiana project update

    Published 2025-12-01Reviewed 2026-08-23Richland Parish, Louisianacompany update indexPrimarySponsor-affiliated

    Read with care: Earlier scope and schedule should not be merged with the 2026 expansion.

    Connected to 0 claims

    Intermediate evidence vintage for a project whose scale changed.

    View original source ↗
  39. S020
    Blackstone

    Blackstone announces Pennsylvania digital and energy infrastructure investment

    Blackstone announces Pennsylvania digital and energy infrastructure investment

    Published 2025-07-15Reviewed 2026-08-23Pennsylvaniacompany announcement indexPrimarySponsor-affiliated

    Read with care: Jobs created or supported are economic-impact figures, not direct payroll headcount.

    Connected to 1 claim

    Blackstone/QTS announced more than $25 billion in investment, more than 6,000 jobs created or supported annually during construction, and more than 3,000 created or supported during operations.

    View original source ↗
  40. S026
    Public Utilities Commission of Ohio

    PUCO approves AEP Ohio data center tariff

    PUCO approves AEP Ohio data center tariff

    Published 2025-07-09Reviewed 2026-08-23AEP Ohio territoryregulatory announcementPrimaryIndependent
    Connected to 1 claim

    In AEP Ohio territory, qualifying data-center loads above about 25 MW can face minimum-demand provisions up to roughly 85% of contracted capacity and terms up to 12 years including ramp.

    View original source ↗
  41. S027
    AEP Ohio

    Data center tariff

    Data center tariff

    Published 2025-07-01Reviewed 2026-08-23AEP Ohio territoryutility tariff summaryPrimarySponsor-affiliated
    Connected to 1 claim

    In AEP Ohio territory, qualifying data-center loads above about 25 MW can face minimum-demand provisions up to roughly 85% of contracted capacity and terms up to 12 years including ramp.

    View original source ↗
  42. S029
    American Electric Power

    AEP Ohio large-load tariff background

    AEP Ohio large-load tariff background

    Published 2025-07-01Reviewed 2026-08-23AEP Ohio territoryutility announcementPrimarySponsor-affiliated
    Connected to 0 claims

    Sponsor explanation of the tariff rationale and anticipated large-load growth.

    View original source ↗
  43. S018
    Commonwealth of Pennsylvania

    Amazon announces at least $20 billion Pennsylvania investment

    Amazon announces at least $20 billion Pennsylvania investment

    Published 2025-06-09Reviewed 2026-08-23Pennsylvaniastate announcement indexPrimaryIndependent

    Read with care: Investment and jobs are announced, not independently verified outcomes.

    Connected to 1 claim

    Amazon announced at least $20 billion in planned Pennsylvania investment and at least 1,250 technical jobs associated with supported Luzerne and Bucks County locations.

    View original source ↗
  44. S019
    Amazon

    AWS plans major Pennsylvania data center investment

    AWS plans major Pennsylvania data center investment

    Published 2025-06-09Reviewed 2026-08-23Pennsylvaniacompany announcement indexPrimarySponsor-affiliated

    Read with care: Company commitment; future scope can change.

    Connected to 1 claim

    Amazon announced at least $20 billion in planned Pennsylvania investment and at least 1,250 technical jobs associated with supported Luzerne and Bucks County locations.

    View original source ↗
  45. 45
    International Labour Organization

    Generative AI and Jobs: A Refined Global Index of Occupational Exposure

    A task-level global index estimating occupational exposure to generative AI.

    Published 2025-05-20Reviewed 2026-08-23Globalintergovernmental working paperPrimaryIndependent
    Connected to 1 claim

    The ILO estimates that one in four workers worldwide is in an occupation with some generative-AI exposure, while 3.3% of global employment is in its highest exposure category.

    View original source ↗
  46. 46
    International Energy Agency

    Energy and AI

    A global model of data-center electricity demand, energy supply, emissions, and uncertainty.

    Published 2025-04-10Reviewed 2026-08-23Globalintergovernmental reportPrimaryIndependent
    Connected to 1 claim

    Data centers consumed about 415 TWh globally in 2024, around 1.5% of global electricity use.

    View original source ↗
  47. 47
    Harvard Electricity Law Initiative

    Extracting Profits from the Public: How Utility Ratepayers Are Paying for Big Tech's Power

    A critical analysis arguing that utility regulation can shift data-center infrastructure risk and cost to other customers.

    Published 2025-03-05Reviewed 2026-08-23United Statesacademic legal analysisIndependent
    Connected to 1 claim

    A data center does not automatically raise every resident's bill, but weak cost-allocation rules can leave other customers carrying infrastructure risk or cost.

    View original source ↗
  48. S038
    Amazon

    AWS plans approximately $11 billion Georgia investment

    AWS plans approximately $11 billion Georgia investment

    Published 2025-01-07Reviewed 2026-08-23Georgiacompany announcement indexPrimarySponsor-affiliated

    Read with care: No exact project-level power, water, or employment figure is inferred where undisclosed.

    Connected to 1 claim

    AWS announced approximately $11 billion in planned Georgia investment.

    View original source ↗
  49. S012
    Nature

    Lifecycle assessment of advanced data-center cooling systems

    Lifecycle assessment of advanced data-center cooling systems

    Published 2025-01-01Reviewed 2026-08-23Modeled U.S. scenariospeer-reviewed lifecycle studyPrimaryIndependent

    Study design: comparative lifecycle assessment

    Read with care: Microsoft-affiliated primary data informed the study; modeled results are not measurements from an arbitrary project.

    Connected to 1 claim

    In modeled scenarios, advanced cooling reduced greenhouse-gas emissions about 15%–21%, energy demand 15%–20%, and blue-water consumption 31%–52%.

    View original source ↗
  50. S024
    U.S. Department of Energy Office of Environmental Management

    PORTS redevelopment documentation

    PORTS redevelopment documentation

    Published 2025-01-01Reviewed 2026-08-23Pike County, Ohiofederal project recordPrimaryIndependent
    Connected to 0 claims

    Provides the federal cleanup and redevelopment context that predates the current project scope.

    View original source ↗
  51. S035
    Louisiana Public Service Commission

    Non-binding guidelines for potential large-load customers

    Non-binding guidelines for potential large-load customers

    Published 2025-01-01Reviewed 2026-08-23Louisianaregulatory guidancePrimaryIndependent
    Connected to 0 claims

    Makes the difference between regulatory guidance and binding tariff terms explicit.

    View original source ↗
  52. S036
    Louisiana Public Service Commission

    Entergy Louisiana large-load tariff filing

    Entergy Louisiana large-load tariff filing

    Published 2025-01-01Reviewed 2026-08-23Entergy Louisiana territoryregulatory filingPrimaryIndependent

    Read with care: Some commercial terms may be redacted or project-specific.

    Connected to 1 claim

    Entergy Louisiana's relevant structure covers eligible customers with at least 70 MW of firm load and specified contract terms.

    View original source ↗
  53. S044
    Georgia Power

    Georgia PSC approves plan to freeze base rates through 2028

    Georgia PSC approves plan to freeze base rates through 2028

    Published 2025-01-01Reviewed 2026-08-23Georgia Power territoryutility announcementPrimarySponsor-affiliated

    Read with care: A base-rate freeze does not prove every bill component is unaffected.

    Connected to 0 claims

    Useful but sponsor-framed description of a regulatory outcome.

    View original source ↗
  54. S054
    Proceedings of the National Academy of Sciences

    Generative AI can harm learning when it gives away answers

    Generative AI can harm learning when it gives away answers

    Published 2025-01-01Reviewed 2026-08-23United Statespeer-reviewed studyPrimaryIndependent

    Study design: randomized educational experiment · Study group pnas-ai-guardrails

    Read with care: Performance during assistance and later unaided learning are different endpoints.

    Connected to 1 claim

    Students can perform better while AI is available yet learn less for later unaided work when assistance gives away answers; guardrails and tutor-like design can change the outcome.

    View original source ↗
  55. 55
    U.S. Department of Energy / Lawrence Berkeley National Laboratory

    2024 United States Data Center Energy Usage Report

    A national estimate of historical and projected U.S. data-center electricity use, including a wide scenario range through 2028.

    Published 2024-12-20Reviewed 2026-08-23United Statesgovernment reportPrimaryIndependent
    Connected to 3 claims

    U.S. data centers used an estimated 176 TWh of electricity in 2023 — about 4.4% of total U.S. electricity consumption.

    U.S. data-center electricity use could reach 325–580 TWh in 2028, or roughly 6.7%–12% of U.S. electricity consumption.

    Estimated direct U.S. data-center water consumption rose from about 21.2 billion liters in 2014 to 66 billion liters in 2023 and is projected at roughly 140–280 billion liters annually in 2028.

    View original source ↗
  56. S045
    Virginia Joint Legislative Audit and Review Commission

    Data Centers in Virginia

    Data Centers in Virginia

    Published 2024-12-09Reviewed 2026-08-23Virginiaindependent legislative studyPrimaryIndependent
    Connected to 2 claims

    JLARC modeled about 74,000 data-center-related jobs annually in Virginia: about 59,000 construction-phase and 15,000 operations-phase jobs, including roughly 35,000 direct construction and 4,400 direct operations jobs.

    A representative 250,000-square-foot facility may involve about 1,500 workers at construction peak and about 50 full-time workers in ongoing operations.

    View original source ↗
  57. S033
    Louisiana Economic Development

    Richland Parish data center project announcement

    Richland Parish data center project announcement

    Published 2024-12-04Reviewed 2026-08-23Richland Parish, Louisianastate announcement indexPrimaryIndependent

    Read with care: Public economic-development announcement is not an audit of project outcomes.

    Connected to 0 claims

    Initial state record for project scope, incentives, and projected employment.

    View original source ↗
  58. S031
    Microsoft

    Microsoft Wisconsin investment announcement

    Microsoft Wisconsin investment announcement

    Published 2024-05-08Reviewed 2026-08-23Wisconsincompany announcementPrimarySponsor-affiliated

    Read with care: Earlier project scope; later operational updates supersede some forward-looking details.

    Connected to 0 claims

    Preserves the original project vintage rather than overwriting it with later disclosures.

    View original source ↗
  59. S039
    Georgia Public Service Commission

    Georgia Power large-load contract rules

    Georgia Power large-load contract rules

    Published 2024-01-01Reviewed 2026-08-23Georgia Power territoryregulatory orderPrimaryIndependent
    Connected to 1 claim

    Georgia Power rules for certain very large loads, generally above 100 MW, can require longer agreements, minimum bills, financial security, and infrastructure charges under regulatory oversight.

    View original source ↗
  60. S052
    Stanford National Student Support Accelerator

    Tutor CoPilot: a human-AI approach for scaling real-time expertise

    Tutor CoPilot: a human-AI approach for scaling real-time expertise

    Published 2024-01-01Reviewed 2026-08-23United Statesstudy pageIndependent

    Study design: field experiment · Sample 1,000 · Study group tutor-copilot

    Read with care: This page and S053 describe the same underlying study and are not independent replications.

    Connected to 1 claim

    Students whose human tutors had AI assistance were about four percentage points more likely to master a topic; gains reached about nine percentage points with lower-rated tutors.

    View original source ↗
  61. S053
    Stanford SCALE

    Tutor CoPilot publication

    Tutor CoPilot publication

    Published 2024-01-01Reviewed 2026-08-23United Statesresearch publication pagePrimaryIndependent

    Study design: field experiment · Sample 1,000 · Study group tutor-copilot

    Read with care: Same underlying study as S052.

    Connected to 1 claim

    Students whose human tutors had AI assistance were about four percentage points more likely to master a topic; gains reached about nine percentage points with lower-rated tutors.

    View original source ↗
  62. 62
    Nature

    Highly accurate protein structure prediction with AlphaFold

    The primary paper describing AlphaFold's performance in the blind CASP14 protein-structure assessment.

    Published 2021-07-15Reviewed 2026-08-23Globalpeer-reviewed researchPrimaryIndependent
    Connected to 1 claim

    In the blind CASP14 assessment, AlphaFold predicted protein structures with accuracy competitive with experimental structures in a majority of evaluated cases.

    View original source ↗
  63. 63
    U.S. Geological Survey

    Water-Use Terminology

    Federal definitions for withdrawal, consumption, recirculation, and cooling systems.

    Published 2019-02-27Reviewed 2026-08-23United Statesgovernment referencePrimaryIndependent
    Connected to 1 claim

    Water withdrawal and water consumption are different measures.

    View original source ↗