Generative AI (GenAI), like the technology behind chatbots or tools that create images from text, is becoming a big part of our daily lives. But one important aspect that goes unseen as we use these AI models daily is the most fundamental question that one may ask—How sustainable are these AI models? As these AI systems get more advanced, they use up a lot of resources like electricity and water, which can harm the environment.To work properly, GenAI models need to be trained, which means they learn by analysing huge amounts of data. This requires powerful computers running for long periods, consuming a lot of electricity. For example, training something like GPT-3, which powers many AI tools today, needs a ton of energy. This leads to more demand for electricity, much of which still comes from fossil fuels, increasing our carbon footprint. While there are efforts to make these systems more energy-efficient, like making better hardware or optimising the way they work, the growing use of AI might still outpace these improvements.To be continued...
Generative AI (GenAI), like the technology behind chatbots or tools that create images from text, is becoming a big part of our daily lives. But one important aspect that goes unseen as we use these AI models daily is the most fundamental question that one may ask—How sustainable are these AI models? As these AI systems get more advanced, they use up a lot of resources like electricity and water, which can harm the environment.To work properly, GenAI models need to be trained, which means they learn by analysing huge amounts of data. This requires powerful computers running for long periods, consuming a lot of electricity. For example, training something like GPT-3, which powers many AI tools today, needs a ton of energy. This leads to more demand for electricity, much of which still comes from fossil fuels, increasing our carbon footprint. While there are efforts to make these systems more energy-efficient, like making better hardware or optimising the way they work, the growing use of AI might still outpace these improvements.To be continued...