Opinion: Does Generative AI Pose Threats to the Environment? | Eco-Business

The Rise of Generative AI: Is it Worth the Carbon Footprint?

Generative AI is the new buzzword in the tech industry, powering chatbots and image generators. But what is the cost of creating these increasingly powerful AI models? As an AI researcher myself, I worry about the impact of such models on the environment, given their high energy consumption. The more advanced the model, the more energy it requires. So, what does the increase in generative AI models mean for the future of our planet’s carbon footprint?

At its core, generative AI refers to an AI algorithm’s ability to produce complex data. Discriminative AI, on the other hand, chooses between a set number of options and produces a single output. For example, a discriminative AI system could decide whether or not to approve a loan application. Generative AI, however, can create much more complex outputs, such as a sentence, paragraph, image, or even a short video. While generative AI has long been used to generate audio responses for smart speakers, it has only recently become capable of producing human-like language and realistic photos.

However, the creation of generative AI models comes with a high cost. It’s difficult to estimate the exact energy cost of a single AI model, which includes the energy used to manufacture the computing equipment, create the model, and use it in production. For instance, creating a generative AI model called BERT with 110 million parameters consumed as much energy as a round-trip transcontinental flight for one person. In comparison, the much larger GPT-3 with 175 billion parameters consumed 1,287 megawatt hours of electricity and generated 552 tons of carbon dioxide equivalent.

Furthermore, there is limited data on the carbon footprint of a single generative AI query. However, some industry estimates suggest that it could be four to five times higher than that of a search engine query. Chatbots and image generators that continue to grow in popularity could significantly increase the number of queries received each day.

Despite the potential environmental impact, there are ways to make generative AI models more efficient. For instance, using a more efficient model architecture, processor, and data center can significantly reduce carbon emissions. Additionally, AI can run on renewable energy to reduce emissions by a factor of 30 to 40 compared to using a grid dominated by fossil fuels.

In conclusion, generative AI models are here to stay, and their usage will continue to increase. While they might not ruin the environment on their own, the energy use could become a problem if thousands of companies develop slightly different AI bots for different purposes, each used by millions of customers. More research is needed to make generative AI more efficient by bringing the computation to where green energy is more abundant or scheduling computation for times of day when renewable energy is more available.

Editor’s Note: As AI continues to shape our world, it’s crucial to stay updated on the latest AI trends and news. To learn more, check out the latest news and analysis on the AI industry at GPT News Room.

Source link

from GPT News Room https://ift.tt/DEPTac8