In the fast-paced world of technology, the Gartner Hype Cycle provides a roadmap for emerging technologies from the initial “technology trigger” to the final “productivity plateau.” Currently, AI, especially generative AI like ChatGPT, has reached the zenith of inflated expectations, promising revolutionary change and ubiquity. But beneath the surface of the AI hysteria lie important questions that are often overlooked. It's about how this technology impacts the environment.
- The hidden environmental costs of AI
- Carbon-intensive training: a one-time cost?
- Urgent need for environmental awareness
- Generative AI’s potential impact on climate change
- Hardware innovation: charting a greener course
- Corporate Responsibility: Leading the Change
AI’s voracious appetite for computing power is emerging as a major environmental problem. The enormous amount of power required to power AI and the heat generated by powerful graphics processing units (GPUs) generates large amounts of carbon dioxide emissions. Despite industry reluctance to disclose detailed environmental data, the impact is clear. The dream of “AI everywhere” could lead to a world dependent on technologies that have a huge environmental impact.
Training AI models remains a carbon-intensive process, with the CO2 emissions of early large-scale language models estimated to be around 300,000kg, but some argue that this is a one-time environmental cost. Some people. The real challenge comes in the “inference” phase, when the AI model enters service and interacts with millions or even billions of users. The energy and carbon footprint of these interactions pose significant environmental problems that require attention.
