The tech industry is one of the major contributors to the global carbon emissions, as data centers and AI models consume huge amounts of electricity and natural resources. However, some experts believe that AI can also be part of the solution, by making the industry more energy efficient and responsible.
The challenge of training and deploying AI models
One of the main sources of the tech industry’s carbon footprint is the creation and use of generative AI tools, such as GPT-4 or Google’s Palm2. These tools can produce realistic and creative content, such as text, images, or music, based on user inputs. However, they also require a lot of computing power and energy to train and deploy.
According to a study by Google and the University of California, Berkeley, training GPT-3 resulted in 552 metric tons of carbon emissions, equivalent to driving a passenger vehicle 1.24 million miles (2 million kilometers). GPT-4, which is the latest generation model by OpenAI, is trained on around 570 times more parameters than GPT-3, and the scale of these systems will only grow as AI becomes more powerful and ubiquitous.
Nvidia, AI’s chip giant, provides the processors that are indispensable for training, known as GPUs. And while they are more energy efficient than typical chips, they remain formidable consumers of power.
The other side of generative AI is deployment, or inference: when the trained model is applied to identify objects, respond to text prompts or whatever the use case may be. Deployment doesn’t necessarily need the computing heft of a Nvidia chip, but taken cumulatively, the endless interactions in the real world far outweigh training in terms of workload.
“Inference is going to be even more of a problem now with ChatGPT, which can be used by anyone and integrated into daily life through apps and web searches,” said Lynn Kaack, assistant professor of computer science at the Hertie School in Berlin.
The potential of responsible AI
However, not all hope is lost. Some researchers and companies are working on ways to make AI more energy efficient and responsible, by using different techniques and technologies.
One example is Untether AI, a highly specialized chip-making company that strives to make AI more energy efficient. Its CEO, Arun Iyengar, said that “Pandora’s box is open. We can utilize AI in ways that enhance the climate requirements or we can ignore the climate requirements and find ourselves facing the consequences in a decade or so in terms of the impact.”
Untether AI claims that its chips can perform inference up to 100 times faster than GPUs, while consuming 10 times less power. This means that they can run more complex AI models with less energy and cost.
Another example is Hugging Face, a startup that provides open-source tools for natural language processing (NLP), one of the most popular domains of AI. Hugging Face has launched a project called Green NLP, which aims to measure and reduce the environmental impact of NLP models.
The project provides a tool called Carbon Tracker, which allows researchers and developers to estimate the carbon emissions of their models during training and inference. It also provides guidelines and best practices for creating more efficient and sustainable NLP models.
The project’s leader, Victor Sanh, said that “We want to raise awareness about the environmental impact of NLP models and provide tools to help researchers and practitioners make informed decisions about their models.”
The future of AI and the environment
The tech industry’s carbon footprint is not only a problem for the environment, but also for its own competitiveness and innovation. As AI becomes more widespread and demanding, the industry will face challenges such as rising energy costs, limited resources, and regulatory pressures.
Therefore, it is in the industry’s best interest to adopt responsible AI practices that can reduce its environmental impact while maintaining its performance and quality. By doing so, the industry can also contribute to solving some of the most pressing global issues, such as climate change, biodiversity loss, and social justice.
As Kaack said, “AI has enormous potential to help us address some of these challenges. But we need to make sure that we are not creating new problems while solving old ones.”
