The climate case for AI has not been made

AI leaders have repeatedly promised that artificial intelligence will help solve the climate crisis, yet real‑world evidence shows the technology is falling far short of those claims. At the United Nations climate conference in Antalya, Turkey, host nations urged participants to ensure AI is designed, powered, and deployed to support climate action, but major industries have yet to deliver on the optimistic projections put forward by companies such as OpenAI and Google. In a 2023 Google sustainability op‑ed, AI was touted as capable of cutting global emissions by 5‑10% by 2030, while a peer‑reviewed study by the London School of Economics’ Grantham Research Institute and Systemiq estimated AI could eliminate 3.2‑5.4 billion tonnes of CO₂ annually by 2035, primarily through gains in power generation, meat and dairy production, and vehicle efficiency. An AI‑assisted review of public disclosures from leading firms in these sectors, conducted by the article’s second author, found that few have pursued the specific emissions‑reducing pathways identified in the study, and even fewer can demonstrate meaningful climate outcomes.

The most striking gap between projected and actual AI impact appears in the power sector, where AI‑driven forecasting of supply and demand could boost renewable electricity output from existing solar and wind plants by up to 20%, potentially averting 1.8 billion tonnes of emissions each year. Yet among grid operators in China, the United States, and Europe, only the United Kingdom’s National Grid ESO, in partnership with Open Climate Fix, has deployed a functional AI system that halved solar‑forecast error and avoided hundreds of thousands of tonnes of CO₂ annually. Elsewhere, public records reveal little evidence of comparable AI integration, suggesting that the sector’s anticipated emissions reductions remain largely aspirational. In the meat and dairy arena, AI would need to narrow the taste and cost gap between animal‑based products and alternatives such as plant‑based or cultivated proteins. NotCo’s joint venture with Kraft Heinz, using the Giuseppe platform to mimic animal product characteristics, stands out as a concrete example, but most alternative‑protein firms lack deep AI integration, and the sector faces broader challenges, including sluggish consumer adoption and a sharp decline in venture capital—from a $989 million peak in 2021 to roughly $83 million last year.

Transportation shows a mixed picture, with some firms achieving the AI‑enabled efficiencies envisioned by the LSE study. Chinese battery giant CATL, which commands about 40 % of the global EV‑battery market, introduced an AI system trained on over 100,000 design cases and 600 terabytes of test data, cutting material‑screening time by two orders of magnitude. Independent analysis also found that ride‑hailing platform Didi Chuxing reduced per‑trip emissions by an average of 16 % through AI‑optimized routing. However, the broader climate case for AI is undermined by oil and gas companies’ parallel interest in the technology for purposes that may not align with emissions reductions, further complicating the narrative of AI as a climate panacea. As the Antalya conference unfolds, the disparity between lofty AI‑driven climate promises and the scant, uneven progress across key sectors underscores the need for concrete policy commitments, transparent reporting, and rigorous testing of AI’s real‑world environmental impact.

⚡ Effects Interpreter

🌍 Effects for your country