The Environmental Dilemma: Unveiling the Impact of Generative AI
Generative AI, the technology powering chatbots and image generators, is gaining popularity. However, concerns arise regarding its environmental impact and raise questions about society’s future carbon footprint.
The term “generative” refers to the capability of an AI algorithm to create complex data, and “discriminative” AI selects from a fixed set of options and provides a single output, like approving a loan application.
Generative AI can produce more intricate outputs such as sentences, paragraphs, realistic photos, and even short videos and has recently advanced to generate human-like language.
The Power Surge of AI:
Creating large AI models requires substantial energy, with size and parameters impacting consumption. For instance, the creation of the GPT-3 model used 1,287 MWh of electricity and emitted 552 tons of CO2eq.
However, according to Google’s research, utilizing more energy-efficient processors, architectures, and data centers can cut carbon emissions by 100 to 1,000 times. As AI usage grows, the carbon footprint of generative AI queries could be four to five times higher than search engine queries.
ChatGPT and other AI bots for search:
The popularity of AI bots like ChatGPT is rising rapidly, with billions of visits recorded. Even Microsoft has integrated ChatGPT into Bing.
Chatbots provide direct answers instead of endless links through a search engine. While accuracy concerns need to be addressed, the efficiency of accessing information may counterbalance the higher energy consumption compared to search engines.
The Future of Generative AI:
As generative AI models become more popular, people will likely rely on them. Therefore, numerous companies develop different AI bots, and energy consumption could become a concern. Research is needed to improve AI efficiency, and Utilizing renewable energy sources can assist cut emissions.