A new study suggests that artificial intelligence systems are causing massive carbon emissions, and the situation is getting worse.
The new research paper warns that the extra demand for energy to run and train complex models is having a serious impact on the environment.
As systems get better, they require more computing power and more energy to run. For example, OpenAI’s current GPT-4 uses 12 times more energy than its previous model.
Training systems is a small process. In fact, running AI tools requires 960 times more energy than the training process.
According to the researchers, the effects of these emissions could be widespread. AI-related emissions could potentially cost the industry more than $10 billion a year, and they have called on governments to come up with measures to measure these emissions, along with new steps to limit them.
