A collaborative study carried out by Carnegie Mellon University and Hugging Face, an AI startup, revealed that creating a single AI-generated image consumes as much power as charging a smartphone.
This concerning revelation about the environmental impact of AI looks bothering, given that image generators like Midjourney and Dall-E are being extensively used across different industries.
In a detailed paper, the researchers explained the significant environmental cost due to energy consumption and carbon emissions while generating images using AI.
Using 30 distinct datasets, the researchers carried out a series of experiments involving as many as 88 models. For each of the 13 tasks, they performed 1,000 inferences. This ranges from generating images to summarizing texts.
People think that AI doesn’t have any environmental impacts, that it’s this abstract technological entity that lives on a ‘cloud’.Dr. Sasha Luccioni, lead researcher
Surprisingly, this study revealed that even when you generate a single AI image, it consumes power almost equivalent to what’s needed to charge your smartphone. On the contrary, text-based content requires less power, proving more energy-efficient. Generative texts consume around 16% of the energy required to charge a smartphone.
The Study Explores Co2 Emissions By Different AI Tasks
Interestingly, this isn’t the first time researchers have raised concerns about carbon emissions in AI use. Researchers at the University of Massachusetts revealed that training a single model of AI can require over 626,000 pounds of carbon dioxide.
“We should be conscious of where and how we use generative AI, comparing its cost and its benefits”, a researcher said.
This is equivalent to the power required by 62.6 cars powered by gasoline when they run for a year.
The study explored the emission of carbon dioxide beyond energy consumption, exploring different workloads of AI. Surpassing other tasks, creating images topped the list, which raised concerns about the extensive use of image generation using AI.
The recent boom brought about by the integration of generative AI involves several power models being integrated into different projects. The study focuses on the overall impact on the environment caused by AI models.
For instance, each week, ChatGPT boasts more than 100 million active users. This significantly scales up the impact on the environment due to the use of AI.
Sustainable Practices And Transparency Are The Key
The researchers, in the context of their findings, focused on the need for transparency in the machine-learning community. The tech community and scientists need to actively engage in communications to mitigate the impact of AI models. Considering the energy-intensive nature of generative AI, there’s a concerning need for sustainable practices.
The study can be considered a wake-up call for the AI industry, revealing the pressing environmental impact following the inception of generative AI. It points to the need to embrace sustainable practices in developing AI and encourages a collaborative responsibility in the AI community to mitigate the carbon footprint.
Considering that AI is likely to shape the tech industry in the future, a sustainable stance defines the key to the eco-friendly development of the technology.
>>> Read full article>>>
Copyright for syndicated content belongs to the linked Source : TechReport – https://techreport.com/news/one-ai-generated-image-needs-as-much-power-as-charging-your-smartphone/
