{"paper":{"title":"Quantifying the Climate Risk of Generative AI: Region-Aware Carbon Accounting with G-TRACE and the AI Sustainability Pyramid","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Generative AI image trends can consume 4,309 MWh and emit 2,068 tCO2 according to the new G-TRACE framework, highlighting climate risks from decentralized inference.","cross_cats":["cs.CL"],"primary_cat":"cs.CY","authors_text":"Mehwish Fatima, Raja Khurram Shahzad, Seemab Latif, Zahida Kausar","submitted_at":"2025-11-06T19:52:02Z","abstract_excerpt":"Generative Artificial Intelligence (GenAI) represents a rapidly expanding digital infrastructure whose energy demand and associated CO2 emissions are emerging as a new category of climate risk. This study introduces G-TRACE (GenAI Transformative Carbon Estimator), a cross-modal, region-aware framework that quantifies training- and inference-related emissions across modalities and deployment geographies. Using real-world analytics and microscopic simulation, G-TRACE measures energy use and carbon intensity per output type (text, image, video) and reveals how decentralized inference amplifies sm"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Through G-TRACE and microscopic simulation, the study estimates 4,309 MWh of energy consumption and 2,068 tCO2 emissions from the Ghibli-style image generation trend, illustrating how decentralized inference amplifies small per-query energy costs into system-level impacts.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The emission estimates rest on the accuracy of real-world analytics and microscopic simulation inputs for per-query energy costs and regional carbon intensities, which are invoked to scale individual actions to tonne-scale consequences but are not detailed or validated in the provided abstract.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"G-TRACE quantifies region-aware GenAI emissions and estimates 4,309 MWh energy use plus 2,068 tCO2 from the Ghibli-style image generation trend, paired with the AI Sustainability Pyramid for translating metrics into policy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Generative AI image trends can consume 4,309 MWh and emit 2,068 tCO2 according to the new G-TRACE framework, highlighting climate risks from decentralized inference.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8ca345f93d23f0ec1dd8340f111d4d847d45e1c6f46ce2e596a95f1a0474e225"},"source":{"id":"2511.04776","kind":"arxiv","version":3},"verdict":{"id":"43364f26-8e96-43a8-aedc-cc8a434cc607","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T00:29:27.747380Z","strongest_claim":"Through G-TRACE and microscopic simulation, the study estimates 4,309 MWh of energy consumption and 2,068 tCO2 emissions from the Ghibli-style image generation trend, illustrating how decentralized inference amplifies small per-query energy costs into system-level impacts.","one_line_summary":"G-TRACE quantifies region-aware GenAI emissions and estimates 4,309 MWh energy use plus 2,068 tCO2 from the Ghibli-style image generation trend, paired with the AI Sustainability Pyramid for translating metrics into policy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The emission estimates rest on the accuracy of real-world analytics and microscopic simulation inputs for per-query energy costs and regional carbon intensities, which are invoked to scale individual actions to tonne-scale consequences but are not detailed or validated in the provided abstract.","pith_extraction_headline":"Generative AI image trends can consume 4,309 MWh and emit 2,068 tCO2 according to the new G-TRACE framework, highlighting climate risks from decentralized inference."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2511.04776/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b81e2090878622a64f7252bedf9c3f477b5e1cfccfdf3bae475435b7bf1c7547"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}