The rise of generative artificial intelligence has become impossible to ignore in the online world. From AI-generated summaries on Google search results to prompts to use AI tools on social media platforms, the presence of AI is ever-growing. This surge in AI integration can be attributed to OpenAI’s release of ChatGPT, which sparked an obsession with generative AI in Silicon Valley. However, this rush to incorporate AI into various online interactions has led to a significant environmental impact that cannot be ignored.
One of the major downsides of the proliferation of generative AI systems is the increased demand for computing resources. According to Sajjad Moazeni, a computer engineering researcher at the University of Washington, the algorithms required to run generative AI models are significantly more resource-intensive than traditional online services like Google Search or email. Moazeni estimates that generative AI applications are 100 to 1,000 times more computationally intensive, leading to a surge in energy and water consumption in the online world.
The energy needs for training and deployment of generative AI models have become a pressing issue. Experts have warned about surges in energy demand at data centers where companies develop AI applications. Companies like Google and Microsoft, in their race to build the most advanced AI tools, have seen a significant increase in energy consumption. Junchen Jiang, a networked systems researcher at the University of Chicago, points out that the carbon footprint and energy consumption of data centers are directly proportional to the amount of computation required for AI models. As companies continue to work on larger and more complex AI models, the energy consumption will only increase.
While companies like Google have made efforts to reduce their environmental impact, the challenges remain significant. Corina Standiford, a spokesperson for Google, acknowledges that reducing emissions from suppliers, who are responsible for 75% of the company’s carbon footprint, is extremely challenging. This includes manufacturers of servers, networking equipment, and other technical infrastructure required for data centers. The energy-intensive process of creating physical parts for frontier AI models adds to the overall environmental impact of generative AI.
As the use of generative artificial intelligence continues to grow, so does its environmental impact. The energy and water consumption required to run AI models are significantly higher than traditional online services, leading to a surge in resource-intensive computing processes. Companies must address the sustainability challenges posed by generative AI to mitigate its environmental impact and work towards a more sustainable future for AI development.