尽管科技业界普遍宣扬人工智能在应对气候变化方面的潜力,但目前的实际减排效果远未达到预期水平 1。相关分析显示,在电力、肉类奶制品及交通三大关键行业中,仅有少数企业通过有效整合AI技术实现了显著降排 1。2024年,OpenAI首席执行官萨姆·奥特曼曾宣称AI将带来包括“修复气候”在内的惊人成就,而伦敦政治经济学院等机构的研究预测,到2035年AI每年可协助减少32亿至54亿吨二氧化碳排放 1。然而,这些乐观预期正面临数据与现状的严峻考验 1。
在具体的技术应用层面,部分案例确实展现了AI的减排价值。英国National Grid ESO与Open Climate Fix合作,利用AI技术将太阳能预测误差减半,从而每年避免数十万吨二氧化碳排放 1。宁德时代推出了基于超过10万个电池设计案例训练的AI系统,成功将筛选候选材料的时间缩短了两个数量级 1;滴滴出行则通过AI优化路径规划,使平均每次行程的排放量减少了16% 1。与此同时,其他领域的进展却令人担忧,培养肉的风险投资额已从2021年的9.89亿美元峰值大幅下滑至去年的约8300万美元 1。此外,AI基础设施的快速扩张导致数据中心能耗激增,国际能源署(IEA)预测到2035年全球数据中心相关排放可能超过3亿吨,这一趋势使得主要科技公司偏离了其既定气候目标,表明“AI有利于气候”的说法目前仍缺乏充分的实证支持 1。
Despite high-profile claims by tech leaders that artificial intelligence will revolutionize efforts to combat climate change, current evidence suggests the technology’s actual emissions-reduction impact falls far short of expectations 1. In 2024, OpenAI CEO Sam Altman asserted that AI would deliver remarkable achievements, including "fixing the climate" 1. This optimism is mirrored by research from institutions such as the London School of Economics, which predicted that AI could help reduce annual carbon dioxide emissions by between 3.2 billion and 5.4 billion tons by 2035 1. However, an analysis of major sectors—including power, meat and dairy, and transportation—indicates that only a minority of companies have successfully integrated AI to significantly lower their emissions 1.
Concrete examples of AI's potential exist but remain isolated compared to the broader narrative. In the UK, National Grid ESO partnered with Open Climate Fix to use AI to halve solar energy forecasting errors, thereby avoiding hundreds of thousands of tons of CO2 annually 1. Similarly, Chinese battery manufacturer CATL deployed an AI system trained on over 100,000 battery design cases, reducing the time needed to screen candidate materials by two orders of magnitude 1, while ride-hailing giant DiDi optimized route planning via AI to cut emissions per trip by an average of 16% 1. Yet, financial indicators in related green tech sectors show volatility; venture capital investment in cultivated meat plummeted from a peak of $989 million in 2021 to approximately $83 million last year 1.
Complicating the climate argument is the rapid expansion of AI infrastructure itself, which has led to surging energy consumption in data centers and caused major tech firms to deviate from their stated climate goals 1. The International Energy Agency (IEA) projects that global emissions associated with data centers could exceed 300 million tons by 2035 1. Consequently, the assertion that "AI is good for the climate" lacks sufficient empirical support when weighed against the growing environmental footprint of its own operational requirements 1.
评论
还没有评论,欢迎留下第一条。