Latest news with #JesseDodge
Yahoo
6 days ago
- Science
- Yahoo
Scientists Just Found Something Unbelievably Grim About Pollution Generated by AI
Tech companies are hellbent on pushing out ever more advanced artificial intelligence models — but there appears to be a grim cost to that progress. In a new study in the science journal Frontiers in Communication, German researchers found that large language models (LLM) that provide more accurate answers use exponentially more energy — and hence produce more carbon — than their simpler and lower-performing peers. In other words, the findings are a grim sign of things to come for the environmental impacts of the AI industry: the more accurate a model is, the higher its toll on the climate. "Everyone knows that as you increase model size, typically models become more capable, use more electricity and have more emissions," Allen Institute for AI researcher Jesse Dodge, who didn't work on the German research but has conducted similar analysis of his own, told the New York Times. The team examined 14 open source LLMs — they were unable to access the inner workings of commercial offerings like OpenAI's ChatGPT or Anthropic's Claude — of various sizes and fed them 500 multiple choice questions plus 500 "free-response questions." Crunching the numbers, the researchers found that big, more accurate models such as DeepSeek produce the most carbon compared to chatbots with smaller digital brains. So-called "reasoning" chatbots, which break problems down into steps in their attempts to solve them, also produced markedly more emissions than their simpler brethren. There were occasional LLMs that bucked the trend — Cogito 70B achieved slightly higher accuracy than DeepSeek, but with a modestly smaller carbon footprint, for instance — but the overall pattern was stark: the more reliable an AI's outputs, the greater its environmental harm. "We don't always need the biggest, most heavily trained model, to answer simple questions," Maximilian Dauner, a German doctoral student and lead author of the paper, told the NYT. "Smaller models are also capable of doing specific things well. The goal should be to pick the right model for the right task." That brings up an interesting point: do we really need AI in everything? When you go on Google, those annoying AI summaries pop up, no doubt generating pollution for a result that you never asked for in the first place. Each individual query might not count for much, but when you add them all up, the effects on the climate could be immense. OpenAI CEO Sam Altman, for example, recently enthused that a "significant fraction" of the Earth's total power production should eventually go to AI. More on AI: CEOs Using AI to Terrorize Their Employees
![Opinion: [A]rtificial [I]nequality: The Inequitable Environmental Cost of Artificial Intelligence](/_next/image?url=https%3A%2F%2Fi0.wp.com%2Fhighschool.latimes.com%2Fwp-content%2Fuploads%2F2025%2F05%2FAI-App.png%3Ffit%3D713%252C523%26ssl%3D1&w=3840&q=100)
![Opinion: [A]rtificial [I]nequality: The Inequitable Environmental Cost of Artificial Intelligence](/_next/image?url=https%3A%2F%2Fall-logos-bucket.s3.amazonaws.com%2Flatimes.com.png&w=48&q=75){kind=small}
Los Angeles Times
22-05-2025
- Business
- Los Angeles Times
Opinion: [A]rtificial [I]nequality: The Inequitable Environmental Cost of Artificial Intelligence
AI apps have a hidden cost. Walk into a room and ask people their thoughts about the last few years of artificial intelligence (AI) development, and you will get as many opinions as people. Some might say some variation on how Gen AI and large language models are the next evolutionary step in digital technology , freeing us from mundane tasks and elevating our creativity to new heights, while others will criticize it as a technology that cheapens creativity and threatens livelihoods . However, regardless of the benefits or drawbacks of AI, there is the consequence of its environmental footprint. In the rush to advance and implement AI use in the past few years, the environmental costs of increasing electricity and water usage have been relatively unregulated and often more strongly affect less developed nations, reflecting a continuation of global inequalities in which the benefits are reaped by nations primarily in the global north while the consequences felt in the global south. The environmental costs of AI begin before a single command is run. AI programs like ChatGPT or Claude require large data centers equipped with large numbers of graphics processing units (GPUs), such as those made by Nvidia. As the United Nations Environmental Programme notes, these powerful chips 'rely on critical mineral and rare elements, which are often mined unsustainably.' Previously, a majority of rare element extraction has been from China, who initially did not have strong environmental regulations on their extraction; most of the environmental damage caused by the boom in rare element mining was because the cheapest methods were often the most environmentally destructive , leaving China with tens of billions of dollars in costs for environmental cleanup amidst rural communities suffering from the long-term fallout. However, as global demand continues to grow, Chinese mining grows more expensive, and other developed nations seek to reduce their reliance on a single source, Africa is increasingly being seen as a potential source of high quality, low cost rare earth metals . Unfortunately, the past few years have already seen the devastating environmental consequences of rare metal mining in Africa, producing 'significant socio-ecological impacts including driving loss of rich biodiversity' and the 'displacement of communities.' Consequently, the strong demand for new GPU chips for AI is contributing to environmental harms both in the present and potentially the future in regions of the global south like Sub-Saharan Africa. In addition to the minerals for chips, AI is also 'highly energy intensive,' as it requires large data centers running many GPUs in order to perform the complex calculations required to, for example, predict the text for a formal email or generate a picture of an apple. Although it is difficult to exactly calculate the electricity cost of a single AI request, in an interview with NPR, Jesse Dodge, research scientist at the Allen Institute for AI, said 'One query to ChatGPT uses approximately as much electricity as could light one lightbulb for about 20 minutes' . On a larger scale, according to Harvard Business Review, by 2026 AI is expected to 'exceed the annual electricity consumption of a small country like Belgium .' While strides have been made in switching to more sustainable energy sources like wind and solar, the majority of electricity generation in the world is through fossil fuels like oil, gas, and coal. Accordingly, tech giants like Google report a roughly 50% increase in their greenhouse gas emissions , with the power consumption of their new AI data centers primarily responsible. Additionally, there are stark differences between the types of electricity used in data centers in Europe versus those in Asia, such as how Google's Finish data centers used '97% carbon-free energy,' whereas its Asian centers were '4-18%.' In effect, the negative externalities of electrical production have been shifted to less developed regions. A third aspect of AI's environmental impact is its heavy use of water. Due to the high heat generated in data centers, water cooling is required for their safe operation, with estimates at 'up to 9 liters of water' evaporated for every kilowatt-hour of electricity used. The water to cool these data centers is increasingly coming from the same sources as drinking water. For instance, in nations like Chile and Uruguay, Google is planning to 'tap into the same reservoirs that supply drinking water.' This comes at a time in which there is a global freshwater crisis, with 'between two and three billion people,' primarily in the global south, suffering from a lack of potable water at least part of each year. In consequence, as the cooling demands of AI mount, there is a risk of an ever-greater strain on freshwater resources, the burden of which would fall inequitably upon less developed nations in the global south. While AI continues to evolve and society engages with the question of how to most responsibly employ it, the environmental costs associated with it are increasingly part of that conversation. In particular, as the benefits and costs of AI are currently inequitably divided between the global north and south, policymakers should consider how to responsibly manage AI's footprint as part of the larger discussion on environmental sustainability in the twenty-first century. Related
